Modulation of TNF Release by Choline Requires α7 Subunit ...diac puncture. Blood was centrifuged at 1,500g for 15 min to isolate serum. Sera were used for TNF analysis by ELISA (R&D

INTRODUCTIONThe excessive production of the pro-

inflammatory cytokines TNF, high mo-bility group box 1 (HMGB1), and otherinflammatory molecules by immune cellsand their subsequent release into the cir-culation are associated with unrestrainedinflammation: a hallmark of septic shock,sepsis, and other disorders (1,2). Exacer-bated release of TNF and other pro-inflammatory cytokines, and lethalityduring endotoxemia and sepsis, can be

controlled by the efferent vagus nerve-based cholinergic anti-inflammatorypathway (3–7). Recent research hasdemonstrated that the α7 subunit-containing nicotinic acetylcholine recep-tor (α7nAChR) is an important compo-nent of the mechanism underlying theanti-inflammatory efficacy of the cholin-ergic anti-inflammatory pathway (4,8).Activation of this pathway by stimula-tion of the vagus nerve suppresses serumTNF levels in endotoxemic animals (3,4),

but fails to cause statistically significanteffects in mice lacking the α7nAChR (4).Accordingly, α7nAChR agonists, includ-ing GTS-21, reduce systemic pro-inflammatory cytokine levels duringmurine endotoxemia, sepsis (9), andother inflammatory conditions (10,11),and improve survival (9).

Choline is a selective and endogenousα7nAChR agonist (12–14). Choline alsohas other important physiological func-tions; this essential nutrient is a majordonor of methyl groups, a cell membraneconstituent, and a precursor in thebiosynthesis of the neurotransmitteracetylcholine (15–17). Although previousstudies have shown protective effects ofcholine against endotoxin-induced shockand organ damage (18–20), the mecha-nism of the anti-inflammatory action of

M O L M E D 1 4 ( 9 - 1 0 ) 5 6 7 - 5 7 4 , S E P T E M B E R - O C T O B E R 2 0 0 8 | P A R R I S H E T A L . | 5 6 7

Modulation of TNF Release by Choline Requires α7 SubunitNicotinic Acetylcholine Receptor-Mediated Signaling

William R Parrish,1 Mauricio Rosas-Ballina,1 Margot Gallowitsch-Puerta,1 Mahendar Ochani,1,5

Kanta Ochani,1 Li-Hong Yang,1 LaQueta Hudson,1 Xinchun Lin,2 Nirav Patel,1 Sarah M Johnson,1

Sangeeta Chavan,1 Richard S Goldstein,3 Christopher J Czura,1 Edmund J Miller,2,5 Yousef Al-Abed,4,5

Kevin J Tracey,1,5 and Valentin A Pavlov1,5

Address correspondence and reprint requests to Valentin A Pavlov, Laboratory of Biomed-

ical Science, The Feinstein Institute for Medical Research, 350 Community Drive, Manhas-

set, NY 11030. Phone: 516-562-2316; Fax: 516-562-2356; Email: [email protected].

Submitted June 19, 2008. Accepted for publication June 19, 2008; Epub (www.molmed.

org) ahead of print June 20, 2008.

1Laboratory of Biomedical Science, 2Department of Surgery, 3General Clinical Research Center, 4Laboratory of Medicinal Chemistry,and 5Center for Immunology and Inflammation, The Feinstein Institute for Medical Research, North Shore-LIJ Health System,Manhasset, New York, United States of America

The α7 subunit-containing nicotinic acetylcholine receptor (α7nAChR) is an essential component in the vagus nerve-basedcholinergic anti-inflammatory pathway that regulates the levels of TNF, high mobility group box 1 (HMGB1), and other cytokinesduring inflammation. Choline is an essential nutrient, a cell membrane constituent, a precursor in the biosynthesis of acetylcholine,and a selective natural α7nAChR agonist. Here, we studied the anti-inflammatory potential of choline in murine endotoxemia andsepsis, and the role of the α7nAChR in mediating the suppressive effect of choline on TNF release. Choline (0.1–50 mM) dose-dependently suppressed TNF release from endotoxin-activated RAW macrophage-like cells, and this effect was associated withsignificant inhibition of NF-κB activation. Choline (50 mg/kg, intraperitoneally [i.p.]) treatment prior to endotoxin administration inmice significantly reduced systemic TNF levels. In contrast to its TNF suppressive effect in wild type mice, choline (50 mg/kg, i.p.)failed to inhibit systemic TNF levels in α7nAChR knockout mice during endotoxemia. Choline also failed to suppress TNF releasefrom endotoxin-activated peritoneal macrophages isolated from α7nAChR knockout mice. Choline treatment prior to endotoxinresulted in a significantly improved survival rate as compared with saline-treated endotoxemic controls. Choline also suppressedHMGB1 release in vitro and in vivo, and choline treatment initiated 24 h after cecal ligation and puncture (CLP)-induced polymi-crobial sepsis significantly improved survival in mice. In addition, choline suppressed TNF release from endotoxin-activated humanwhole blood and macrophages. Collectively, these data characterize the anti-inflammatory efficacy of choline and demonstratethat the modulation of TNF release by choline requires α7nAChR-mediated signaling.Online address: http://www.molmed.orgdoi: 10.2119/2008-00079.Parrish

this compound is not well understood. Aparticularly important question iswhether the α7nAChR, which is an es-sential component of the cholinergic anti-inflammatory pathway, mediates theanti-inflammatory action of choline dur-ing endotoxemia. Another relevant ques-tion is whether choline suppresses thepro-inflammatory cytokine response andaffects the survival rate during polymi-crobial sepsis.

In this study, we provide evidence thatcholine functions as an anti-inflammatorymolecule through an α7nAChR-dependent mechanism. In contrast to itsanti-inflammatory effect in wild typemice, choline failed to reduce endotoxin-induced serum TNF levels in α7nAChRKO mice and TNF release from peri-toneal macrophages isolated from thesemice. These findings represent the firstdirect experimental evidence that theanti-inflammatory activity of a choliner-gic agonist is mediated in vivo throughan α7nAChR-dependent mechanism. Wealso show that choline suppresses the re-lease of HMGB1, and choline treatmentinitiated within a clinically relevant timeframe significantly improves survival inmice with severe sepsis.

MATERIALS AND METHODS

AnimalsMale mice (BALB/c at 25–28 g

[Taconic], and wild type [WT] orα7nAChR knockout [KO] C57BL/6 at8–12 wk old) were used for in vivo stud-ies, and WT or α7nAChR KO C57BL/6 female mice at 8–12 wk old were usedin ex vivo peritoneal macrophage stud-ies. All C57BL/6 animals were bred onsite from heterozygous α7nAChR KOanimals obtained from Jackson Labora-tories (Bar Harbor, ME, USA). The geno-type of the α7nAChR locus (CHRNA7)of all progeny was determined by ge-nomic PCR using the Extract and Ampkit (Sigma, St. Louis, MO, USA). Ani-mals were housed in standard condi-tions (room temperature 22° C with a12-h light:dark cycle) with free access toregular chow and water. Animals were

allowed to acclimate for at least 14 d be-fore the corresponding experiment. Allanimal experiments were performed inaccordance with the National Institutesof Health Guidelines under protocolsapproved by the Institutional AnimalCare and Use Committee of the Fein-stein Institute for Medical Research,North Shore-LIJ Health System, Man-hasset, New York, United States ofAmerica.

RAW Cells, Drug Treatment, TNF, andHMGB1 Determination

RAW 264.7 cells were purchased fromthe American Type Culture Collection(ATCC TIB-71, Manassas, VA, USA) andmaintained in DMEM supplementedwith 10% heat-inactivated FBS, 2 mMglutamine (Biowhittaker, Walkersville,MD, USA) and 100 U/mL penicillin, 100 μg/mL streptomycin (both Gibco,Carlsbad, CA, USA). Cell cultures weremaintained at 37° C, 5% CO2. Cells wereseeded in 48-well tissue culture plates at5 × 105 cells per well, and were allowedto adhere for 24 h. Prior to adding com-pounds, media were removed and re-placed with serum-free Optimem media(Gibco). Cells were exposed to lipopoly-saccharide (LPS; endotoxin) (Escherichiacoli, L4130 0111:B4; Sigma) (4 mg/mL)in the presence or absence of choline atthe concentrations indicated. Cell cul-ture media were harvested 4 h after LPSaddition and centrifuged at 800g for 5min to sediment cell debris. SecretedTNF was assayed from the media byEnzyme-Linked ImmunoSorbent Assay(ELISA) (R&D Systems, Minneapolis,MN, USA) according to the manufac-turer’s recommendations. For HMGB1determination, cell culture media werecollected 24 h after endotoxin treatmentand centrifuged at 800g for 5 min to re-move cellular debris. HMGB1 was ana-lyzed from cleared media by Westernblot as described previously (9). Briefly,cleared medium was filtered throughCentricon YM-100 (Millipore, Billerica,MA, USA) diluted 1:2 with 2 × Laemmlisample buffer (BioRad, Hercules, CA,USA), and subjected to electrophoresis

through 10%–20% Tris-HCl acrylamidegels (BioRad). Proteins were immobi-lized onto PVDF membranes (Amer-sham Pharmacia Biotech, Uppsala,Sweden) and probed with polyclonalanti-HMGB1 antibodies. Membraneswere developed using ECL Westernblotting detection reagents (AmershamPharmacia Biotech). Autoradiographfilms were scanned and densitometricanalyses performed using QuantityOne Software (BioRad). Standard curvesof human recombinant HMGB1 wereconstructed and used to interpolateHMGB1 levels in the samples. Cell via-bility was monitored using trypan blueexclusion.

Nuclear Protein Extraction for NF-κBActivity Determination andElectrophoretic Mobility-Shift Assay(EMSA)

RAW 264.7 macrophages were treatedwith the indicated concentrations ofcholine followed by LPS (4 ng/mL).Two h after LPS stimulation, cells wereprocessed for nuclear protein extractionas described previously (9). EMSA wasperformed using the Nushift NF-κB p65kit (Active Motif, Carlsbad, CA, USA) ac-cording to the manufacturer’s instruc-tions, as described previously (9).

Endotoxemia and Drug TreatmentEndotoxemia was induced by injecting

mice intraperitoneally (i.p.) with 6 mg/kgendotoxin, which caused ~80% mortality.Mice were treated i.p. with the indicateddose of choline or vehicle (sterile saline)at 6 h, and at 30 min prior to endotoxinadministration. Animals were euthanizedby CO2 asphyxiation 1.5 h after endotoxininjection, and blood was collected by car-diac puncture. Blood was centrifuged at1,500g for 15 min to isolate serum. Serawere used for TNF analysis by ELISA(R&D Systems) according to the manu-facturer’s recommendations. In survivalexperiments, mice were treated i.p. withcholine (50 mg/kg or 5 mg/kg) or sterilesaline (controls) at 6 h and at 30 min priorto endotoxin injection. Survival was mon-itored for 2 wks.

5 6 8 | P A R R I S H E T A L . | M O L M E D 1 4 ( 9 - 1 0 ) 5 6 7 - 5 7 4 , S E P T E M B E R - O C T O B E R 2 0 0 8

C H O L I N E A N T I - I N F L A M M A T O R Y E F F I C A C Y

Isolation and Treatment of MousePeritoneal Macrophages

WT or α7nAchR KO C57BL/6J femalemice were injected with 2 mL of 9%thioglycollate broth i.p. to elicit peri-toneal macrophages. Animals were eu-thanized by CO2 asphyxiation 42–48 hlater, and cells were collected by lavageof the peritoneal cavity three times with5 mL ice-cold 11.6% sucrose. Cells werewashed three times with phosphate-buffered saline (PBS) and once withRPMI 1640 medium and resuspendedin RPMI 1640 medium supplementedwith 10% heat-inactivated FBS, 2 mMglutamine (Biowhittaker) and 100 U/mLpenicillin, 100 μg/mL streptomycin(Gibco). Cells were seeded at 1.5 × 106

per well into 24-well Falcon Primariatissue culture dishes and were allowedto adhere for 2 h at 37° C under 5% CO2.Then, cells were washed twice with PBSand supplied with fresh medium as de-scribed above, returned to the incuba-tor, and allowed to rest for 18–24 h.Prior to treatment, media were removedand replaced with serum free Optimemmedia. Cells were incubated with theindicated concentration of choline for10 min prior to LPS (100 ng/mL)exposure. Cell culture media wereharvested 4 h after LPS treatment andTNF was assayed by ELISA (R&D Sys-tems) according to the manufacturer’srecommendations.

Cecal Ligation and Puncture Surgeryand Drug Treatment

Severe polymicrobial sepsis was in-duced by cecal ligation and puncture(CLP). Mice were anesthetized using ketamine (100 mg/kg) and xylazine (8 mg/kg) administered intramuscularly.Abdominal access was gained via a mid-line incision. The cecum was isolated andligated with a 6-0 silk ligature below theileocecal valve and then punctured oncewith a 22 G needle. Stool (approximately1 mm) was extruded from the hole, andthe cecum placed back into the abdomi-nal cavity. The abdomen was closed withtwo layers of 6-0 Ethilon sutures. An anti-biotic (Imipenem-Cilastatin, 0.5 mg/kg,

subcutaneously, in a total volume of 0.5 mL/mouse) was administered imme-diately after CLP as part of the resuscita-tion fluid. Sham-operated animals hadthe cecum isolated and then returned tothe peritoneal cavity without being lig-ated or punctured. Sham animals also re-ceived an antibiotic treatment and resus-citative fluid as described above. ForHMGB1 determination, mice were ran-domized 24 h after CLP and were in-jected i.p. with either sterile saline orcholine (25 mg/kg). Mice received addi-tional treatments at 30 h and 44 h afterCLP. Blood was collected by cardiacpuncture at 45 h after CLP, and serumHMGB1 levels were determined byquantitative Western blot analysis as de-scribed above. For survival studies, 24 hafter CLP, mice were randomized and in-jected i.p. with either sterile saline orcholine (25 mg/kg or 5 mg/kg). Thistreatment was repeated 6 h later (30 hafter CLP) and then twice daily for 2 dfor a total of six treatments. Survival wasroutinely monitored for 2 wks.

Isolation and Treatment of HumanMacrophages

Peripheral blood mononuclear cells(PBMCs) were isolated by density gradi-ent fractionation from whole blood thatwas obtained from anonymous donorsthrough the Long Island Blood Services(Westbury, NY, USA), and were differen-tiated to macrophages in culture as de-scribed previously (4). Briefly, PBMCswere harvested from the plasma/Ficol-Hypaque interface after centrifugationfor 30 min at 550g. Cells were washedtwice with PBS, once with RPMI 1640medium, and resuspended in RPMI 1640supplemented with 10% heat-inactivatedhuman serum, 2 mM glutamine(Biowhittaker), and 100U/mL penicillin,100 μg/mL streptomycin (Gibco). Cellswere then seeded at 5 × 107 per 10 cmFalcon Primaria tissue culture plate andincubated for 2 h at 37° C in 5% CO2 toallow attachment. Plates were thenwashed twice with PBS lacking Ca2+ andMg2+, and adherent cells were detachedby gentle scraping in PBS lacking Ca2+

and Mg2+ and containing 1mM EDTA.Cells were washed once and resus-pended in medium supplemented withhuman recombinant macrophage colony-stimulating factor (hrMCSF), and wereseeded into 24-well Falcon Primaria tis-sue culture plates at 1 × 106 cells perwell. Cultures were incubated for 6 d inthe presence of hrMCSF to promote mac-rophage differentiation. Macrophageswere rested for 24 h in the absence ofhrMCSF prior to use. Media were re-moved and replaced with serum free Op-timem media prior to treatment. Cellswere incubated with the indicated con-centration of choline for 10 min prior toexposure to LPS (20 ng/mL). Cell culturemedia were harvested 4 h after LPS treat-ment and secreted TNF was assayed byELISA (R&D Systems) as describedabove.

Statistical AnalysisData are expressed as mean ± SEM.

Significant differences were assessed byusing one way analysis of variance(ANOVA) followed by a Student t test.The statistical significance of differencesbetween groups of animals in survivalexperiments was analyzed by the log-rank test. Differences with P < 0.05 wereconsidered statistically significant.

RESULTS

Choline Inhibits Endotoxin-InducedTNF Release and NF-κB Activation

The anti-inflammatory efficacy ofcholinergic agonists has been tested pre-viously by using RAW-264.7 mousemacrophage-like cells (9), a well-established cell culture system for in vitrostudies of innate immune cell inflamma-tory responses. We studied the efficiencyof choline in inhibiting TNF release fromendotoxin-activated RAW cells. The cellswere pre-incubated for 10 min with in-creasing choline concentrations prior toendotoxin activation. As shown in Fig-ure 1A, choline dose-dependently sup-pressed endotoxin-stimulated TNF re-lease from RAW cells. NF-κB is a keytranscription factor that is activated in

R E S E A R C H A R T I C L E


response to endotoxin for the productionof inflammatory mediators such as TNF.Therefore, we tested whether thecholine-induced suppression of TNF wasassociated with inhibition of NF-κB acti-vation. RAW cells were incubated for 10min in the absence or presence of in-creasing concentrations of choline priorto exposure to endotoxin. Nuclear ex-tracts were prepared and electrophoreticmobility shift assays (EMSA) were con-ducted to measure activated NF-κB.Choline dose-dependently suppressedNF-κB activation in response to endo-toxin (Figure 1B).

Choline Suppresses Systemic TNFLevels During Endotoxemia Throughan α7nAChR-Mediated Mechanism

Based on our in vitro data, we nexttested whether choline reduces systemicTNF during endotoxemia. Choline(5 mg/kg or 50 mg/kg) or vehicle(saline) was injected into BALB/c mice at6 h and at 30 min prior to endotoxin(6 mg/kg i.p.) administration. This doseof endotoxin previously was shown tocause about 80% mortality (9). Thehigher choline dose (50 mg/kg) signifi-cantly suppressed serum TNF levels(Figure 2A). The lower choline dose(5 mg/kg) failed to alter serum TNF lev-els (see Figure 2A). In light of the re-cently discovered role for the α7nAChRin mediating the cholinergic suppressionof systemic TNF during endotoxemia (4),and the fact that choline is a selectiveα7nAChR agonist, we tested the efficacyof choline in suppressing systemic TNFlevels during endotoxemia in WT andα7nAChR KO mice. WT and age-matched α7nAChR KO mice were in-jected i.p. with choline (50 mg/kg) or ve-hicle (saline) at 6 h and at 30 min prior toendotoxin (6 mg/kg, i.p.) administration.Choline significantly suppressed sys-temic TNF levels in endotoxemic WTmice as compared with saline-treated en-dotoxemic controls (Figure 2B, P < 0.05).In contrast, choline administration inα7nAChR KO mice did not alter sys-temic TNF levels significantly, as com-pared with controls (see Figure 2B). Mac-

rophages represent a major source ofTNF during endotoxemia (21,22), and theα7nAChR expressed on macrophagesplays a critical role in mediating cholin-ergic anti-inflammatory signaling (4).Therefore, we reasoned that choline maysuppress macrophage TNF releasethrough an α7nAChR-dependent signal-ing mechanism. Accordingly, we exam-ined the effect of choline on TNF releasefrom endotoxin-activated peritonealmacrophages collected from WT miceand α7nAChR KO mice. Peritoneal mac-rophages were exposed to the indicatedconcentrations of choline 10 min prior tothe addition of endotoxin, and TNF lev-els were measured in media super-natants collected 4 h later. As shown inFigure 2C, choline dose-dependentlysuppressed TNF release by macrophagesfrom WT mice. However, choline treat-ment did not suppress TNF release bymacrophages isolated from α7nAChRKO mice (see Figure 2C).

Choline Improves Survival in LethalEndotoxemia

We have shown previously that theα7nAChR agonist GTS-21 significantlyimproves survival of BALB/c mice dur-ing endotoxemia (9). We next studiedwhether choline improves survival inlethal endotoxemia. Choline (5 mg/kg or50 mg/kg, i.p.) or vehicle (saline, i.p.)was injected into BALB/c mice at 6 hand at 30 min prior to the i.p. adminis-tration of endotoxin (6 mg/kg, i.p.). Ve-hicle-treated mice showed a 27% survivalrate that was not improved by 5 mg/kgcholine (Figure 3). In contrast, treatmentwith 50 mg/kg choline resulted in a sig-nificantly improved survival rate of 63%(see Figure 3).

Choline Suppresses HMGB1 Releaseand Improves Survival in Mice withSevere Sepsis

HMGB1 is a late pro-inflammatory cy-tokine mediator of inflammation duringexperimental sepsis and an importanttherapeutic target in the treatment of thisdisorder (23–25). We studied whethercholine suppresses HMGB1 release in



Figure 1. Choline inhibits TNF release (A)and NF-κB activation (B) in endotoxin-stimulated RAW-264.7 mouse macrophage-like cells. (A) RAW cells were exposed tothe indicated concentration of choline10 min prior to the addition of endotoxin(4 ng/mL). Culture supernatants wereharvested 4 h later and TNF was deter-mined by ELISA. Data represent themean ± SEM of two representative ex-periments conducted in duplicate (*P < 0.04 as compared, with vehicle [V]treated controls). (B) RAW cells were ex-posed to the indicated concentrationsof choline 10 min prior to the addition ofendotoxin (4 ng/mL) and cells were har-vested 2 h after endotoxin challenge fordetermination of NF-κB activation byEMSA. Autoradiographs were subjectedto densitometry by using Quantity Onesoftware (Biorad). Data represent themean ± SEM of three independent ex-periments (*P < 0.04, **P < 0.006 as com-pared with the lowest choline concen-tration tested).

vitro and in vivo and improves survivalof mice with polymicrobial sepsis. Basedon our data that choline suppresses TNFrelease and NF-κB activation in RAWcells (see Figure 1A,1B), we testedwhether choline also attenuates HMGB1release from RAW cells. Choline dose-dependently reduced HMGB1 releasefrom endotoxin-stimulated RAW cells

(Figure 4A). We next studied whethercholine suppresses serum levels ofHMGB1 in mice with CLP-induced sep-sis. Choline (25 mg/kg) or saline were in-jected i.p. to septic mice 24 h after theCLP surgery. Mice received additionaltreatments 30 h and 44 h after CLP, andserum HMGB1 levels were determined inblood obtained at 45 h after CLP. Serum

HMGB1 levels were reduced by 85% inthe animals receiving choline as com-pared with the saline-administered con-trols (Figure 4B). It is noteworthy that themortality rate at the 45-h time point washigher in saline-treated mice (5/12) com-pared with the choline-treated group(1/12) (data not shown). We then specifi-cally tested whether choline improvessurvival when therapeutically adminis-tered to septic mice. Choline (25 mg/kg,or 5 mg/kg) or vehicle (saline) was ad-ministered i.p. to mice with CLP-inducedsepsis 24 h after surgery. This treatmentwas repeated 6 h later (30 h after surgery)and twice daily for 2 d more. The sur-vival rate for choline-treated (25 mg/kg)mice (64%) was improved significantlywhen compared with control animals(23%) (Figure 4C). The 42% survival rateof septic mice treated with the lowercholine dose (5 mg/kg) was not signifi-cantly different as compared with controlmice (see Figure 4C). These data showthat therapeutically administered cholineattenuates systemic HMGB1 levels andimproves survival in polymicrobial sepsis.

Choline Suppresses TNF Productionfrom Endotoxin-Stimulated HumanWhole Blood and CulturedMacrophages

To examine the anti-inflammatory effi-ciency of choline in human cells, we



Figure 2. Choline suppresses systemic TNF levels during endotoxemia through anα7nAChR-dependent mechanism. (A) Choline or vehicle (V, saline) was injected i.p. inBALB/c mice (n = 10 per group) at 6 h, and at 30 min, prior to endotoxin (6 mg/kg, i.p.)administration. Serum TNF was analyzed by ELISA in blood obtained 90 min after endo-toxin administration. Results show the mean ± SEM for each group (*P < 0.05 as comparedwith vehicle (V) administered controls). (B) Choline (50 mg/kg, i.p.) or saline was injectedi.p. in age-matched WT and α7nAchR KO mice (WT n = 8–9/group, α7nAChR KO n =7–9/group) at 6 h, and at 30 min, prior to endotoxin (6 mg/kg, i.p.) administration. SerumTNF was analyzed by ELISA in blood obtained 90 min after endotoxin administration. Re-sults show the mean ± SEM for each treatment group (*P < 0.001 as compared with salineadministered controls). (C) Peritoneal macrophages from age-matched WT and α7nAChRKO mice were incubated with the indicated concentrations of choline or vehicle (V) for10 min prior to exposure to endotoxin (100 ng/mL). TNF in cell culture media was deter-mined by ELISA 4 h after endotoxin addition. Results represent the mean ± SEM of three in-dependent experiments conducted in duplicate (*P < 0.04, **P < 0.02, ***P < 0.002 ascompared with lowest choline concentration tested).

Figure 3. Choline improves survival in lethalendotoxemia. BALB/c mice (n = 30/ group)were injected i.p. with either vehicle (saline)or choline at 6 h, and at 30 min, prior to en-dotoxin (6 mg/kg, i.p.) administration. Sur-vival was monitored for 14 d (*P < 0.002).

tested whether choline suppressed TNFrelease from endotoxin-activated humanwhole blood and cultured human macro-phages. Blood was collected fromhealthy volunteers and blood sampleswere exposed to increasing concentra-tions of choline or vehicle for 10 minprior to the addition of endotoxin. Asshown in Figure 5A, choline (50 mM)significantly suppressed endotoxin-induced TNF release from human wholeblood. In a parallel set of experiments,

human macrophages that express theα7nAChR (4) were differentiated fromperipheral blood mononuclear cells(PBMCs) and treated with variouscholine concentrations 10 min prior tothe addition of endotoxin. As shown inFigure 5B, choline (1 mM) significantlyreduced TNF release from endotoxin-stimulated human macrophages. Thelevel of TNF suppression did not in-crease with higher concentrations ofcholine (up to 50mM) (data not shown).



Figure 4. Choline treatment suppresses HMGB1 release and improves survival in severesepsis. (A) Choline suppresses HMGB1 release from endotoxin-stimulated RAW cells. RAWcells were exposed to the indicated concentration of choline or vehicle 10 min prior toLPS (100 ng/mL) addition for 24 h. Culture supernatants were harvested, and secretedHMGB1 was detected by Western blot analysis. HMGB1 was not detected in the super-natant from cells that were not treated with LPS. Data represent the mean ± SEM of fourexperiments conducted in duplicate (*P < 0.05, **P < 0.02, ***P < 0.001 as compared withlowest choline concentration tested). (B) Mice (n = 12) were administered i.p. with eithersaline or choline (25 mg/kg) 24 h after CLP. Mice received additional treatments at 30 hand 44 h after CLP. Serum HMGB1 levels were determined in surviving mice (n = 11 forcholine treatment, n = 7 for control treatment) in blood obtained at 45 h after CLP (*P <0.0008). (C) Mice (n = 26 per group) were subjected to CLP surgery. 24 h after CLP, micewere randomized and injected i.p. with either saline or choline (25 mg/kg). This treatmentwas repeated 6 h later (30 h after CLP) and twice daily for 2 d more for a total of six treat-ments, and survival was monitored for 2 wks (*P < 0.002).

Figure 5. Choline suppresses TNF releasefrom endotoxin-stimulated humanwhole blood (A) and human macro-phages (B). (A) Whole blood wastreated for 10 min with the indicatedconcentrations of choline prior to endo-toxin (10 ng/mL) challenge at 37° C.Plasma TNF was determined by ELISA 4 h later. Data represent the mean ±SEM of duplicate determinations fromfive donors (*P < 0.0001) as comparedwith the lowest choline concentrationtested). (B) Peripheral blood mononu-clear cells (PBMCs) were isolated fromadult donors and were differentiatedinto macrophages. Macrophages weretreated with the indicated concentra-tion of choline 10 min prior to the addi-tion of endotoxin (20 ng/mL). Culture su-pernatants were harvested 4 h laterand the level of TNF secreted into themedia was determined by ELISA. Datarepresent the mean ± SEM of at leastthree experiments conducted in dupli-cate from independent donors (*P <0.02) as compared with the lowestcholine concentration tested.

DISCUSSIONIn this study, we show that choline sup-

presses serum TNF levels in endotoxemicmice and this anti-inflammatory effect ofcholine is dependent on an α7nAChR-mediated signaling. In addition, we dem-onstrate the anti-inflammatory efficacy ofcholine in experimental polymicrobialsepsis and in human cells.

Choline suppressed systemic TNF lev-els in endotoxemic mice, but failed toreduce TNF levels in mice lacking theα7nAchR (α7nAchR KO mice) and thesefindings clearly indicate the α7nAChRdependence of this anti-inflammatory invivo effect of choline. In contrast to itssuppressive effect on TNF release fromendotoxin-stimulated WT peritonealmacrophages, choline did not suppressTNF release from α7nAChR KO cells.These results strengthen the conceptthat α7nAChR expressed on macro-phages and other immune cells plays acritical role in controlling inflammatoryresponses (4,26).

Choline, a byproduct of acetylcholinedegradation, is a stable, natural, and se-lective agonist on α7nAChR. Earlierstudies indicated that acetylcholine sup-presses TNF release from peritonealmouse macrophages and human macro-phages (3,4), but it was unknown previ-ously whether choline also can regulateTNF in these cells. Interestingly, ourdata show that choline concentrationsthat suppress TNF release from theseendotoxin-stimulated immune cells aresignificantly higher than acetylcholineconcentrations (in the presence of anacetylcholinesterase inhibitor) that exertsimilar suppressive effects (3,4). Theseobservations are in line with studiesshowing a lower agonistic efficacy ofcholine on neuronal α7nAChRs as com-pared with acetylcholine (14,27,28) andhave implications for signaling afteracetylcholine release, because choline canpersist after acetylcholine degradation.

The expression of the α7nAChR in sev-eral non-neuronal cells, including macro-phages, monocytes, and dendritic cells,has been documented (8,29,30). How-ever, knowledge about the receptor func-

tion and pharmacological characteristicsrelated to cytokine production is verylimited. Previously, agonistic propertiesof choline have been studied on neuronalα7nAChRs and it has been shown thatcholine is a full agonist on the α7nAChRwith an EC50 of 1.6 mM (12). These datawere suggestive for the drug concentra-tions we used to study the effects ofcholine on TNF release, NF-κB activa-tion, and HMGB1 release in response toendotoxin. Our results show that cholineconcentrations required to cause statisti-cally significant suppression of TNF re-lease and NF-κB activation in vitro aregenerally higher than those that sup-press HMGB1 release. This observationmay indicate that the cellular mecha-nisms governing the release of HMGB1could be more sensitive to choline-stimulated α7nAChR signaling thanthose controlling the release of TNF. Thisin vitro difference also was extrapolatedto the in vivo studies. While a cholinedose of 50 mg/kg was required to sup-press serum TNF significantly and to im-prove survival during endotoxemia, alower choline dose of 25 mg/kg signifi-cantly inhibited serum HMGB1 levelsand improved survival in mice withpolymicrobial sepsis. A possible explana-tion could be related to differences be-tween the underlying inflammatorymechanisms of endotoxemia and CLP-sepsis. The survival-improving effect ofcholine, administered i.p. 6 h and 30 minprior to endotoxin is in line with a previ-ous study, demonstrating that a choline-rich diet improves survival in endotox-emic rats (18). Moreover, our results thatcholine treatment, initiated within a clin-ically relevant time frame, improves sur-vival in polymicrobial sepsis indicate thepotential for clinical development ofcholine.

Choline deficiency has been shownpreviously to induce liver injury in hu-mans and in rodents, which is exacer-bated upon endotoxin administration(31,32). In contrast, choline (20 mg/kg,intravenous [i.v.]) administration sup-presses TNF release and attenuates in-flammation during endotoxemia in dogs

(19,20). Moreover, choline (60 mg/kg,i.v.) attenuates acid-induced lung injuryin mice (33). The effective doses ofcholine used in the present study (25–50mg/kg, i.p.) are within the dose rangeused in these other studies. It is impor-tant to note that we did not observe anyadverse neurobehavioral effects of thesecholine doses, which are comparablewith the recommended tolerable upperlimit of dietary choline intake in humans(34). Unlike other synthetic α7nAChRagonists, choline is an endogenous mole-cule with important physiological func-tions, including its vital roles in main-taining the structural integrity of cellmembranes and providing methylgroups for the synthesis of betaine,thus participating in methionine, folate,and homocysteine metabolism (34).While some of these metabolic functionshave been linked previously to anti-inflammatory effects of choline (18,35)our data clearly show that α7nAChRsignaling is required for the anti-inflammatory efficacy of this compoundduring endotoxemic shock. Our results(data not shown) also indicate thatα7nAChR plays a role in mediating theanti-inflammatory efficacy of choline inpolymicrobial sepsis. Therefore, our find-ings bring new light to these previousstudies and suggest that endogenouscholine may act on the α7nAChR andplay an important role in regulating in-nate immune responses to maintain ho-meostasis. Choline also is a precursorfor the synthesis of acetylcholine, whichis the principle neurotransmitter of theefferent vagus nerve (36). It is possiblethat a portion of the exogenous choline ismetabolized as a substrate for acetyl-choline biosynthesis, which may con-tribute to anti-inflammatory effects invivo. In the cholinergic synapse, acetyl-choline is degraded rapidly by acetyl-cholinesterases into acetate and choline.Choline generated in this mode may actto prolong α7nAChR activation selec-tively (36). While indicating a criticalanti-inflammatory role for choline as anα7nAChR agonist, we cannot entirely ex-clude the contribution of other effects of



choline to its anti-inflammatory activityin vivo, including the stimulation ofcholinergic signaling in the central nerv-ous system (CNS), which has beenshown recently to play a role in control-ling inflammation during endotoxemia(37). However, it is possible that these al-ternative pathways also culminate inα7nAChR-mediated signaling.

Choline also suppressed TNF releasefrom endotoxin-activated human wholeblood and macrophages effectively, dem-onstrating the anti-inflammatory efficacyof this compound in human cells. Inconclusion, our data provide experimen-tal evidence that the cholinergic agonistcholine suppresses TNF release throughan α7nAChR-dependent mechanism andhas therapeutic potential in the treat-ment of sepsis and other inflammatorydiseases.

ACKNOWLEDGMENTSThis study was supported by a North

Shore-LIJ Health System Research Awardgrant (to VAP), MO1 R018535, andNIGMS R01 GM0557226-08A1 (to KJT).

REFERENCES1. Tracey KJ, et al. (1986) Shock and tissue injury in-

duced by recombinant human cachectin. Science234:470–4.

2. Tracey KJ. (2002) The inflammatory reflex. Nature420:853–9.

3. Borovikova LV, et al. (2000) Vagus nerve stimula-tion attenuates the systemic inflammatory re-sponse to endotoxin. Nature 405:458–62.

4. Wang H, et al. (2003) Nicotinic acetylcholine re-ceptor alpha7 subunit is an essential regulator ofinflammation. Nature 421:384–8.

5. Huston JM, et al. (2007) Transcutaneous vagusnerve stimulation reduces serum high mobilitygroup box 1 levels and improves survival inmurine sepsis. Crit. Care Med. 35:2762–8.

6. Pavlov VA, Wang H, Czura CJ, Friedman SG,and Tracey KJ. (2003) The cholinergic anti-inflammatory pathway: a missing link in neuro-immunomodulation. Mol. Med. 9:125–34.

7. Tracey KJ. (2007) Physiology and immunology ofthe cholinergic antiinflammatory pathway. J. Clin.Invest. 117:289–96.

8. Gallowitsch-Puerta M and Pavlov VA. (2007)Neuro-immune interactions via the cholinergicanti-inflammatory pathway. Life Sci. 80:2325–9.

9. Pavlov VA, et al. (2007) Selective alpha7-nicotinicacetylcholine receptor agonist GTS-21 improvessurvival in murine endotoxemia and severe sep-sis. Crit. Care Med. 35:1139–44.

10. van Westerloo DJ, et al. (2006) The vagus nerveand nicotinic receptors modulate experimentalpancreatitis severity in mice. Gastroenterology130:1822–30.

11. Yeboah MM, et al. (2008) Cholinergic agonists at-tenuate renal ischemia-reperfusion injury in rats.Kidney Int. 74:62–9.

12. Alkondon M, Pereira EF, Cortes WS, Maelicke A,and Albuquerque EX. (1997) Choline is a selec-tive agonist of alpha7 nicotinic acetylcholine re-ceptors in the rat brain neurons. Eur. J. Neurosci.9:2734–42.

13. Mandelzys A, De Koninck P, and Cooper E.(1995) Agonist and toxin sensitivities of ACh-evoked currents on neurons expressing multiplenicotinic ACh receptor subunits. J. Neurophysiol.74:1212–21.

14. Papke RL, Bencherif M, and Lippiello P. (1996)An evaluation of neuronal nicotinic acetylcholinereceptor activation by quaternary nitrogen com-pounds indicates that choline is selective for thealpha 7 subtype. Neurosci. Lett. 213:201–4.

15. Blusztajn JK. (1998) Choline, a vital amine. Sci-ence 281:794–5.

16. Zeisel SH. (2000) Choline: an essential nutrientfor humans. Nutrition 16:669–71.

17. Canty DJ and Zeisel SH. (1994) Lecithin andcholine in human health and disease. Nutr. Rev.52:327–39.

18. Rivera CA, Wheeler MD, Enomoto N, and Thur-man RG. (1998) A choline-rich diet improves sur-vival in a rat model of endotoxin shock. Am. J.Physiol. 275:G862–7.

19. Ilcol YO, Yilmaz Z, and Ulus IH. (2005) Endo-toxin alters serum-free choline and phospho-lipid-bound choline concentrations, and cholineadministration attenuates endotoxin-inducedorgan injury in dogs. Shock 24:288–93.

20. Yilmaz Z, Ilcol YO, Torun S, and Ulus IH. (2006)Intravenous administration of choline or cdp-choline improves platelet count and platelet clo-sure times in endotoxin-treated dogs. Shock25:73–9.

21. Chensue SW, Terebuh PD, Remick DG, ScalesWE, and Kunkel SL. (1991) In vivo biologic andimmunohistochemical analysis of interleukin-1alpha, beta and tumor necrosis factor during ex-perimental endotoxemia. Kinetics, Kupffer cellexpression, and glucocorticoid effects. Am. J.Pathol. 138:395–402.

22. Grivennikov SI, et al. (2005) Distinct and non-redundant in vivo functions of TNF produced byt cells and macrophages/neutrophils: protectiveand deleterious effects. Immunity 22:93–104.

23. Wang H, Yang H, and Tracey KJ. (2004) Extracel-lular role of HMGB1 in inflammation and sepsis.J. Intern. Med. 255:320–31.

24. Lotze MT and Tracey KJ. (2005) High-mobilitygroup box 1 protein (HMGB1): nuclear weaponin the immune arsenal. Nat. Rev. Immunol.5:331–42.

25. Parrish W and Ulloa L. (2007) High-mobilitygroup box-1 isoforms as potential therapeutictargets in sepsis. Methods Mol. Biol. 361:145–62.

26. Saeed RW, et al. (2005) Cholinergic stimulationblocks endothelial cell activation and leukocyterecruitment during inflammation. J. Exp. Med.201:1113–23.

27. Papke RL, et al. (2005) Rhesus monkey alpha7nicotinic acetylcholine receptors: comparisons tohuman alpha7 receptors expressed in Xenopusoocytes. Eur. J. Pharmacol. 524:11–8.

28. Papke RL and Porter Papke JK. (2002) Compara-tive pharmacology of rat and human alpha7nAChR conducted with net charge analysis. Br. J.Pharmacol. 137:49–61.

29. Cloez-Tayarani I and Changeux JP. (2007) Nico-tine and serotonin in immune regulation and in-flammatory processes: a perspective. J. Leukoc.Biol. 81:599–606.

30. Kawashima K, Yoshikawa K, Fujii YX, MoriwakiY, and Misawa H. (2007) Expression and functionof genes encoding cholinergic components inmurine immune cells. Life Sci. 80:2314–9.

31. Nolan JP and Vilayat M. (1968) Endotoxin andthe liver. I. Toxicity in rats with choline deficientfatty livers. Proc. Soc. Exp. Biol. Med. 129:29–31.

32. Eastin CE, McClain CJ, Lee EY, Bagby GJ, andChawla RK. (1997) Choline deficiency augmentsand antibody to tumor necrosis factor-alpha at-tenuates endotoxin-induced hepatic injury. Alco-hol Clin. Exp. Res. 21:1037–41.

33. Su X, et al. (2007) Activation of the alpha7nAChR reduces acid-induced acute lung injuryin mice and rats. Am. J. Respir. Cell Mol. Biol.37:186–92.

34. Zeisel SH. (2006) Choline: critical role duringfetal development and dietary requirements inadults. Annu. Rev. Nutr. 26:229–50.

35. Kim SK and Kim YC. (2002) Attenuation of bac-terial lipopolysaccharide-induced hepatotoxicityby betaine or taurine in rats. Food Chem. Toxicol.40:545–9.

36. Pavlov VA and Tracey KJ. (2005) The cholinergicanti-inflammatory pathway. Brain Behav. Immun.19:493–9.

37. Pavlov VA. et al. (2006) Central muscariniccholinergic regulation of the systemic inflamma-tory response during endotoxemia. Proc. Natl.Acad. Sci. U. S. A. 103:5219–23.



Documents

Modulation of TNF Release by Choline Requires α7 Subunit ...diac puncture. Blood was centrifuged at 1,500g for 15 min to isolate serum. Sera were used for TNF analysis by ELISA (R&D