Sex-related effects of nutritional supplementation of Escherichia coli: Relevance to eating disorders

lable at ScienceDirect

Nutrition 31 (2015) 498–507

Contents lists avai

Nutrition

journal homepage: www.nutr i t ionjrnl .com

Basic nutritional investigation

Sex-related effects of nutritional supplementation of Escherichiacoli: Relevance to eating disorders

Naouel Tennoune Ph.D. a,b, Romain Legrand M.Sc. a,b, Wassila Ouelaa Ph.D. a,b,Jonathan Breton M.Sc. a,b, Nicolas Lucas M.Sc. a,b, Christine Bole-Feysot B.Sc. a,b,Jean-Claude do Rego Ph.D. b,c, Pierre D�echelotte M.D., Ph.D. a,b,d,Sergueï O. Fetissov M.D., Ph.D. a,b,*a Inserm UMR1073, Nutrition, Gut and Brain Laboratory, Rouen, Franceb Institute for Research and Innovation in Biomedicine (IRIB), Rouen University, Normandy University, Rouen, FrancecAnimal behavior platform (SCAC), Rouen, FrancedRouen University Hospital, CHU Charles Nicolle, Rouen, France

a r t i c l e i n f o

Article history:Received 21 October 2014Accepted 26 November 2014

Keywords:Animal modelsFeedingBody weightAnxietyEating disordersAnorexia nervosaBulimiaNeuropeptidesMicrobiota

This study received support from the EU INTERR(7-003-FR_TC2 N) and the Region of Haute Normandieno conflicts of interest to declare.* Corresponding author. Tel.: þ33 235 148 255; fax

E-mail address: [email protected] (S.

http://dx.doi.org/10.1016/j.nut.2014.11.0030899-9007/� 2015 Elsevier Inc. All rights reserved.

a b s t r a c t

Objectives: The biological background of sex-related differences in the development of eatingdisorders (EDs) is unknown. Recent data showed that gut bacteria Escherichia coli induce auto-antibodies against anorexigenic a-melanocyte-stimulating hormone (a-MSH) associated withpsychopathology in ED. The aim of this study was to compare the effects of E. coli on feeding andautoantibodies against a-MSH and adrenocorticotropic hormone (ACTH), between female andmale rats.Methods: Commensal E. coli K12 were given in a culture medium daily to adult Wistar rats byintragastric gavage over a 3-wk period; control rats received culture medium only.Results: Before gavage, E. coli K12 DNA was detected in feces of female but not male rats. E. coliprovision was accompanied by an increase in body weight gain in females, but a decrease in bodyweight gain and food intake in males. Independent of E. coli treatment, plasma levels of anti-a-MSH and ACTH immunoglobulin (Ig)G were higher in female than male rats. Females responded toE. coli by increasing a-MSH IgG levels and affinity, but males by increasing a-MSH IgM levels.Affinity of IgG for ACTH was increased in both E. coli-treated females and males, although withdifferent kinetics. IgG from females stimulated more efficiently a-MSH-induced cyclic adenosinemonophosphate production by melanocortin 4 receptor-expressing cells compared with IgG frommales.Discussion: Sex-related response to how E. coli affects feeding and anti-melanocortin hormoneantibody production may depend on the presence of these bacteria in the gut before E. coli sup-plementation. These data suggest that sex-related presence of certain gut bacteria may represent arisk factor for ED development.

� 2015 Elsevier Inc. All rights reserved.

Introduction metabolism are physiologically influenced by sex hormones [2,

The main types of eating disorders (EDs), anorexia nervosa(AN) and bulimia, occur about 10 times more frequently in fe-males than in males [1]. Although the eating behavior and

EG IVA 2 Seas Program, France. The authors have

: þ33 235 148 226.O. Fetissov).

3], their role in the sex-related prevalence of EDs is unknown.The present study aimed to clarify the biological background of asex-dependent risk factor for ED by exploring sex-related effectsof gut bacteria on feeding and production of autoantibodies(autoAbs)-reactive with the host peptide hormones controllingappetite.

Melanocortin peptides, a-melanocyte-stimulating hormone(a-MSH) and adrenocorticotropic hormone (ACTH) are productsof the same pro-opiomelanocortin (POMC) precursor and arecritically involved in the regulation of energy balance via binding

N. Tennoune et al. / Nutrition 31 (2015) 498–507 499

to melanocortin receptor type 4 (MC4 R) resulting in increasedsatiation [4] and thermogenesis [5]. Indeed, in both humans andmice, gene mutations of POMC [6,7] and MC4 R [8,9] lead toovereating and obesity. Furthermore, MC4 R is involved insignaling mood and emotion [10,11]. ACTH is also a majormediator of the hypothalamic–pituitary–adrenal (HPA) axis,acting via MC2 R in the adrenal cortex to increase cortisolsecretion in response to stress.

Anorexia nervosa and bulimia are characterized by alteredfeeding behavior, anxiety, depression, and altered stressresponse [12], indicating that the melanocortin system may bepotentially involved in the ED pathophysiology. However, nosignificant associations have been found between genetic vari-ants of MC4 R and AN [13], bulimia [14] or binge-eating [15],suggesting alternative mechanisms of altered melanocortinsignaling in ED. One such possibility may involve immunoglob-ulins (Ig), that is, autoAbs that can modulate biological effects ofpeptide hormones (e.g., by increasing their stability) [16].

Indeed, it was previously shown that in both humans and rats,circulating IgG may bind to a-MSH and ACTH [17,18] and thatplasma levels of these autoAbs correlate with some typicalbehavioral traits in female patients with EDs [19]. Furthermore,a-MSH autoAbs were associated with anxiety and depression inhealthy individuals [20] and ACTH autoAbs with antisocialbehavior in adolescents [21], suggesting that autoAbs reactivewith the melanocortin hormones may physiologically regulatebehavioral traits in humans. It was also shown that the repeatedmild stress in rats changed binding properties of a-MSH autoAbs,linking them to the control of food intake and anxiety [22].

Regarding the mechanisms underlying the origin of autoAbsagainst melanocortin hormones, our recent work showed that a-MSH-reactive autoAbs can be stimulated by homologous to a-MSH antigenic proteins produced by commensal gut bacteria. Infact, we identified that the heat-shock ClpB protein of Escherichiacoli is a conformational mimetic of a-MSH and that providing E.coli to male mice increased plasma levels of anti-ClpB IgG cross-reactive a-MSH [23]. Furthermore, correlations between ClpBIgG levels and psychopathologic traits found in ED further sup-ported that ClpB-containing bacteria may be involved in theetiology of ED [23]. It is, hence, possible that there might existsex-related differences in response to E. coli to produce autoAbsagainst a-MSH and ACTH.

Thus, in the present work, we tested our hypothesis that E.coli nutritional supplementation may influence production of a-MSH and ACTH autoAbs differently in male and female ratsassociated with changes in feeding and anxiety.

Materials and methods

Animals

Animal care and experimentation were in accordance with guidelinesestablished by the U.S. National Institutes of Health, and complied with bothFrench and European community regulations (Official Journal of the EuropeanCommunity L 358, 18/12/1986). Male and female Wistar rats (N ¼ 48), bodyweight 220 to 250 g, were purchased from Janvier Labs (Le Genest-St-Isle, France)and were maintained at 24�C with a 12-h light–dark cycle in a specialized animalfacility. The rats were fed with standard pelleted rodent chow (RM1 diet, SDS,UK) when kept in standard holding cages for 1 wk. After acclimation, rats weretransferred to individual metabolism cages (Tecniplast, Lyon, France) where theywere fed with the same RM1 but grounded chow (SDS). Drinking water was al-ways available. Body weight and intake of food and water were measured daily.Male and female rats were studied in two separate but identical experiments.Although the estrous cycle is known to influencemeal pattern in female rats [24],in the present study it was not taken into consideration because total food intakewas measured and the study duration included more than four to five cycles. Theday before starting the gavage, blood samples (200 mL) were taken by the retro-

orbital puncture in rats anesthetized by an intraperitoneal (IP) injection (25 mg/kg body weight) of Brietal (Lilly, Nieuwegein, Netherlands). Plasma was sepa-rated by centrifugation at 3500g for 10 min at 4�C and stored at �80�C beforeassay.

E. coli K12 culture and gavage protocols

The bacterial strain used in this studywas E. coli K12. This strainwas availablefrom the UMR 6270 CNRS Laboratory in Rouen University, France. From thefrozen stocks (–80�C), E. coli K12 were suspended in 250 mL of Luria broth (LB)liquid medium (MP, Illkrich, France) and plated in LB agar to detect the con-taminations. The medium containing bacteria were cultured anaerobically at37�C for 24 h in 50-mL tubes. To estimate the number of E. coli bacteria, opticaldensity (OD) of bacterial culture was measured by a spectrophotometer atl ¼ 600 nm with OD 0.1 corresponding to about 108 cells/mL. Rats were intra-gastrically gavaged with 4 mL of LB medium containing 108 E. coli K12 (n¼ 12) orwith 4 mL of LB medium without bacteria (controls, n ¼ 12). The gavage proce-dure was carried out daily between 0900 and 1000 for 21 d.

E. coli K12 bacterial DNA assay

The DNAwas purified from feces of males and females rats with the QIAampR

DNA Stool Mini Kit (QIAGEN, France), DNA of E. coli K12 was also extracted fromthe cultures of the strain; bacteria were dissolved in water and warmed at 100�Cfor 5 min, after 1 min of centrifugation at 11000g, the supernatant containing theDNA was stored at �20�C.

Based on the published study [25], we used the following nucleotide primers,K12-R: 50-ATCCTGCGCACCAATCAACAA-30 and K12-F: 50-CGCGATGGAAGATGCTCTGTA-3’ (Invitrogen Custom Primers, Cergy Pontoise, France) to amplifya region of the orf264 as a marker of E. coli K12. Polymerase chain reaction wasperformed in a thermocycler with MicroAmp tubes (Eppendorf, Hambourg,Germany). The reactionwas carried out in a 50-mL volume containing 25 mL of GoTaq R Green Master Mix 2 X (Promega, Madison, WI), 1 mL (20 pmol) of eachprimer, 21 mL of bi-distilled water and 1 mL of bacterial DNA. PCR conditions wereas follows: 3 min at 94�C followed by 35 cycles at 94�C for 30 sec, 60�C for 30 sec,and 72�C for 1.5 min. PCR products were analyzed on a 1% agarose gel (Sigma,St Louis, MO, USA), the size of the amplification product corresponding to0.97-kb.

Plasma autoantibody assay

Plasma levels of IgG or IgM autoAbs reacting with a-MSH, ACTH (IgG only)and ClpB were measured using enzyme-linked immunosorbent assay (ELISA)technique according to a published protocol [26]. Briefly, a-MSH or rat ACTHpeptides (BachemAG, Bubendorf, Switzerland) or E. coli ClpB (Delphi Genetics SA,Gosselies, Belgium) were coated on Maxisorp plates (Nunc, Rochester, NY, USA)using 100 mL and a concentration of 2 mg/mL in 100mMNaHCO3 buffer, pH 9.6 for72 h at 4�C. Plates were washed (5 min � 3) in phosphate-buffered saline (PBS)with 0.05% Tween 200, pH 7.4, and then incubated overnight at 4�C with 100 mLof rat plasma diluted 1:200 in PBS to determine free autoAbs levels or diluted1:200 in dissociative 3 M NaCl, 1.5 M glycine buffer, pH 8.9 to determine totalautoAbs levels. The optimal dilutions of plasma were determined as 1:200. Theplates were washed (3 � ) and incubated with 100 mL of alkaline phosphatase-conjugated goat anti-rat IgG (1:2000) or antirat IgM (1:1000) both from Jack-son Immuno Research Laboratories, Inc. (West Grove, PA, USA). Followingwashing (3� ), 100 mL of p-nitrophenyl phosphate solution (Sigma) was added asalkaline phosphatase substrate. After 40 min of incubation at room temperature,the reactionwas stopped by adding 3 NNaOH. The ODwas determined at 405 nmusing a microplate reader Metertech 960 (Metertech Inc., Taipei, Taiwan). BlankOD values resulting from the reading of plates without addition of rat plasmawere subtracted from the sample OD values. Each determination was done induplicate. The variation between duplicate values was <5%. Plasma concentra-tions of total IgG and IgM were measured using Rat IgG and IgM ELISA Quanti-tation Sets (Bethyl Laboratories Inc., Montgomery, TX, USA).

IgG purification from plasma

Purification of plasma IgG for the analysis of their affinity was done accordingto a published protocol [27]. Plasma globulins were separated on C-18 SEP col-umns (Phoenix Pharmaceuticals, Burlingame, CA, USA) after plasma acidificationby mixing 500 mL of rat plasma was 500 mL of buffer A (1% trifluoroacetic acid[TFA] in water). The columnwas activated in 1 mL of buffer B (60% acetonitrile in1% TFA) by 3 min centrifugationwith 700g and rinsed 3 times with 3 mL of bufferA. Diluted plasma (1:1 in buffer A) was added to the column and the effluent(1 mL) was saved and frozen at �20�C. Total IgG were purified from the effluentsof rat plasma samples using the Melon Gel Kit (Thermo Fisher Scientific, Rock-ford, IL, USA). In brief, plasma effluents diluted 1:4 in the kit’s purification bufferwas added on washed melon gel deposited in a column. Column was spinned

Fig. 1. Polymerase chain reaction detection of a 0.97 kbp fragment from E. coli K12DNA. Lanes: (1) Ready-load TM 100 bp DNA ladder; (2) DNA extracted from culturedE. coli K12; (3) DNA from feces of male rats before gavage; (4) DNA from feces offemale rats before gavage.

N. Tennoune et al. / Nutrition 31 (2015) 498–507500

1 min at 6000g and the IgG-containing effluent was saved and frozen at �20�Cbefore lyophilization. Lyophilized IgG were reconstituted in the HBS-EP buffer(GE Healthcare, Piscataway, NJ, USA).

Affinity kinetics assay of IgG for a-MSH and ACTH

Affinity kinetics of rat IgG for a-MSH and ACTH was determined by thesurface plasmon resonance (SPR) phenomenon on a BIAcore 1000 instrument(GE Healthcare). a-MSH or rat ACTH peptides (Bachem) were diluted 0.5 mg/mLin 10 mM sodium acetate buffer, pH 5.0 (GE Healthcare) and were covalentlycoupled on the sensor chip CM5 (GE Healthcare), using the amine coupling kit(GE Healthcare). To compare relative binding response between a-MSH and IgGfrommale and female rats, 25 m of purified IgG (0.5 mg/mL or 3360 nmol in HBS-EP buffer) were injected into the flow conduit of the BIAcore instrument withflow speed 5 mL/min at 25

�C. The values of resonance units of the sensorgram

were recorded 5 min after dissociation. All samples were analyzed in a randomorder and alternating injections of IgG from E. coli and control groups. For theaffinity kinetic analysis between a-MSH or rat ACTH and IgG, a multicyclemethod was run with five serial dilutions of each IgG sample: 3360, 1680, 840,420, and 210 (nmol) including a duplicate of 840 nmol and a blank sample (HBS-EP buffer only). Each cycle included 2 min of analyte injection and 5 min ofdissociation with flow speed 30 mL/min at 25

�C. Between injections of each

sample, the binding surface was regenerated with 10 mM NaOH, resulting in thesame baseline level of the sensorgram. The affinity kinetic data were analyzedusing BiaEvaluation 4.1.1 program (GE Healthcare). For fitting kinetic data, theLangmuir’s 1:1 model was used and the sample values were corrected by sub-tracting the blank values.

Food intake after food restriction

After the last day of gavage, the rats underwent a 5-d food restrictionregimen with food available 2 h/d for 5 d. After day 5 of food restriction, all ratswere injected with a-MSH (Bachem, 100 mg/mL IP) 1000 and the food was pro-vided ad libitum. Food intake during refeeding was measured for 24 h.

Locomotor activity and anxiety tests

Five days after food restriction, rats were analyzed for locomotor activityusing a Versamax Animal Activity Monitor (AccuScan Instruments, Inc., Colum-bus, OH, USA). The day after the locomotor activity test, all rats were injectedwith a-MSH (Bachem, 100 mg/mL IP) and 15 min later tested for anxiety in anelevated O-maze. The elevated O-maze is a variation of more commonly usedelevated-plus-maze pharmacologically validated for anxiety testing in rats [28].The advantage of the O-maze is that it lacks the ambiguous central square of thetraditional plus-maze. The O-maze (Med Associate, Inc., St. Albans, VT, USA)consisted of a circular infrared platform (outer diameter 120 cm) elevated 80 cmabove the floor, featuring two open and two closed segments made of grayplastic. The closed segments were enclosed by walls extending 30 cm above thesurface of the maze and covered with a black Plexiglas lid. Each test started byplacing the rat into one of the two closed segments. The test lasted 5min and wasrecorded using a video camera placed above the O-maze and the EthoVisionvideo tracking software (Noldus IT, Wageningen, The Netherlands). Measure-ments of distance and time spent in the open and closed segments wereanalyzed. Between each rat tests, the O-maze was cleaned with a 30% ethanol.Two hours after the O-maze test, rats were sacrificed by decapitation in a guil-lotine and trunk blood was collected into EDTA-containing tubes for plasmaseparation as described previously.

In vitro cAMP assay

Stable cell line of human embryonic kidney (HEK) 293 cells expressing hu-man MC4 R was generated using a lentiviral transduction technology and pur-chased from Amsbio (Oxon, UK). High expression of MC4 R mRNA in transfectedcells was validated by reverse transcription-PCR in Amsbio and in our laboratory.The presence of the transgene in cells before each experiment was verified by thevisualization at a fluorescence microscope of the green-fluorescent protein (GFP)which gene was inserted in the same with MC4 R lentivector but under adifferent promoter. The day before the experiment, four IgG pools correspondingto each experimental group of rats were made from six individual rat IgG sam-ples. Each IgG pool was diluted to 0.5 mg/mL in PBS. The a-MSH peptide(Bachem) was diluted in the induction buffer: PBS, 500 mM IBMX, 100 mM RO 20to 1724 (Sigma-Aldrich), 20 mM MgCl2 to the final concentrations of 2 mM, 1 mM,750 nM, 500 nM, 250 nM, 100 nM, 75 nM, 50 nM, 10 nM corresponding to thea-MSH doses of 0.6, 3, 4.5, 6, 15, 30, 45, 60, and 120 pmol, respectively, and alsoincluded one blank sample. Each dilution of a-MSH was incubated with 3 mL ofone of the four pools of rat IgG or with PBS at 4�C overnight.

After unfreezing, the cells were cultured in 250 mL tissue culture flasks(BD-Falcon, Becton-Dickinson, Bedford, MA, USA) in Dulbecco’s Modified Eagle

Medium 4,5 g/L glucose (Eurobio, Courtaboeuf, France) supplemented with(2 mM L-glutamine; 10% fetal calf serum; 0,1 mM nonessential amino-acids; 1%penicillin-streptavidin) in humidified cell culture incubator at 37�C, 5% CO2 for 8to 10 d. On the day of experiment, cultured MC4 R HEK293 cells were treatedwith 0,25% trypsin-EDTA (Sigma-Aldrich) and cell pellets were suspended in PBSto obtain about 5000 cells/well (10 mL) in a nontreated bioluminescence white96-microwells plate (Nunc, Roskilde, Denmark). The cyclic adenosine mono-phosphate (cAMP) production by MC4 R expressing HEK 293 cells was measuredusing the bioluminescent assay cAMP-GloTM Max Assay kit (Promega) accordingto themanufacturer’s instructions. Briefly, the cells were incubated with differentdilutions of a-MSH peptide alone or a-MSH with rat IgG for 15 min at roomtemperature. Serial dilutions of cAMP standard (provided by the kit) wereassayed on the samemicroplate. cAMP detection solutionwas added to eachwell,then the cells were homogenized by agitation and centrifuged 2 min at 1000gand then incubated for 20 min at 23�C. Kinase-Glo reagent substrate was addedin each well, and after 10 min of incubation at 23�C, the luminescence was readwith a bioluminescence instrument (Safas Spectrometer, Monaco). In eachexperiment, the measurements were performed in triplicate and the sameexperiment was performed three times on separate days resulting in nine foreach point of the cAMP-activation curve. The negative controls included incu-bation of a-MSH peptide with HEK 293 cells that were not transfected with MC4R (Amsbio).

Statistical analysis

Data were analyzed and the graphs were plotted using the GraphPad Prism5.02 (GraphPad Software Inc., San Diego, CA, USA). Normality was evaluated bythe Kolmogorov-Smirnov test. Group differenceswere analyzed by the analysis ofvariance (ANOVA) or the nonparametric Kruskal-Wallis test with the Tukey’s orDunn’s post-tests, according to the normality results. Body weight changes wereanalyzed with the two-way repeated measurements (RM) ANOVA and theBonferroni post-tests. Where appropriate, individual groups were comparedusing the Student’s t test or the Mann-Whitney test according to the normalityresults. The cAMP production was analyzed using a nonlinear regression fit (log[a-MSH] versus normalized cAMP response) for which the equationwas Y¼ 100/[1 þ 10

ˇ

({LogEC50-X}*HillSlope)]. Data are shown as means � SEM, and for alltest, P < 0.05 was considered statistically significant.

Results

Presence of E. coli K12 in rat feces

Presence of E. coli K12 DNA in rat feces was verified by PCR. Asa positive control for E. coli K12, PCR product of the expected size(0.97 kbp) was amplified from the DNA extracted directly fromthe cultured E. coli K12 bacteria (Fig. 1). Before gavage with E. coliK12, a PCR product of 0.97 kbp was abundantly present in thefeces DNA of female rats, whereas in males it was at the limit ofdetection (Fig. 1).

Effects of E. coli K12 gavage on body weight and intake of foodand water

Female and male rats responded differently to E. coli K12gavage, a can be seen by changes in their body weight gain aswell as food and water intake. In females receiving E. coli K12,

N. Tennoune et al. / Nutrition 31 (2015) 498–507 501

body weight gain was more pronounced than in controls,resulting in a significant increase of 4.4% at the end of gavage(Fig. 2A). In males, starting the gavage with E. coli K12 resulted inan immediate significant decrease in body weight lasting for 1 dfollowed by the regain of body weight to the levels below but notsignificantly different from the control group (Fig. 2B).

Food intake did not significantly differ between E. coliK12–treated and control females (Fig. 2C), although mean dailywater intakewas slightly increased in the E. coli K12 group versuscontrols (29.9 � 0.38 versus 28.5 � 0.38 mL, respectively, Stu-dent’s t test P ¼ 0.015). In males, a significant decrease in foodintake was observed for 2 d after beginning of gavage with E. coliK12 (Fig. 2D). The mean daily food and water intake during 3 wkof gavage were also lower in E. coli K12–treated males than incontrols (food intake, 25.6 � 0.6 g versus 27.6 � 0.2 g, respec-tively, Student’s t test P ¼ 0.007; water intake: 21.5 � 0.6 mLversus 23.7 � 0.8 mL respectively, Student’s t test P ¼ 0.03).

Plasma anti-ClpB and total Igs

Plasma levels of anti-ClpB IgG were increased after E. colitreatment in both female and male rats without sex differences(Fig. 3A), but E. coli did not have significant effects on plasmaanti-ClpB IgM (Fig. 3B). Plasma concentrations of total IgG andIgM were higher in females than in males but were not affectedby E. coli treatment (Fig. 3C, D).

Plasma a-MSH autoantibodies

Plasma levels of both free and total (dissociated) a-MSH-reactive IgG were higher in female than male rats before and

Fig. 2. Effects of gavage with E. coli K12 on body weight (A, B) and food intake (C, D) in femin E. coli K12–treated females at the end of gavage (Student’s t test *P < 0.05, two-way RMmales after beginning of gavage (two-way RM ANOVA, P ¼ 0.03, Bonferroni posttest *P <

found in E. coli K12–treated male rats after beginning of gavage (two-way RM ANOVAanalysis of variance.

after the gavage in both E. coli K12–treated and control rats(Fig. 4A–D). In both groups of E. coli K12–treated and controlfemale and male rats, the levels of a-MSH IgG increasedcompared with before gavage (Fig. 4A–D), however, after thegavage, levels of total but not free a-MSH-reactive IgG werehigher in E. coli K12–treated female rats (Fig. 4C), whereas nosignificant differences was observed between E. coli K12–treatedand control males rats.

Plasma levels of a-MSH-reactive IgM were not differentbefore gavage between female and male rats (Fig. 4E, F). Afterproviding E. coli K12, increased a-MSH IgM levels versus controlswas found only in males (Fig. 4F), whereas in females, bothgroups showed similar a-MSH IgM levels that were higher thanbefore gavage (Fig. 4E).

Affinity kinetics of a-MSH-reactive Igs

Affinity kinetics between a-MSH and IgG extracted fromplasma of female andmale rats after gavage were analyzed usingSPR. Examples of the SPR responses in resonance units to sixdifferent concentrations of rat IgG are shown for one control(Fig. 5A) and one E. coli K12–treated females rats (Fig. 5B). Fittingeach individual rat SPR responses to different IgG concentrationswas done to obtain affinity kinetics parameters (Fig. 5D–F). MeanSPR response measured 5 min after dissociation at low flowspeed was lower in male than in female rats and it was alsodecreased in both female and male E. coli K12–treated rats(Fig. 5C). In female rats, the dissociation equilibrium constant(KD) was lower in E. coli K12–treated rats versus controls, indi-cating an increased affinity for a-MSH (Fig. 5D). The mean KDvalues of control females did not significantly differ from the KD

ale (A, C) and male (B, D) rats. (A) Higher versus controls body weight was observedANOVA, P ¼ 0.07). (B) Decreased body weight was observed in E. coli K12–treated0.05). (C) Food intake did not differ in female rats. (D) Decreased food intake was

, P ¼ 0.02, Bonferroni post-test ***P < 0.001). RM ANOVA, repeated measurement

Fig. 3. Effects of gavage with E. coli K12 on plasma levels of anti-ClpB IgG (A) and IgM (B) and plasma concentrations of total IgG (C) and IgM (D) in female and male rats. (A)Student’s t test, *P < 0.05. Sex differences, female versus male rats, aversus control males, bversus E. coli K12-treated males, ANOVA, Tukey’s post-tests **P < 0.01, ***P < 0.001.ANOVA, analysis of variance; IG, immunoglobulin.

N. Tennoune et al. / Nutrition 31 (2015) 498–507502

of male rats, who also showed similar KD values between thecontrol and E. coli K12 groups (Fig. 5D). Themean association rate(small Ka) was increased (Fig. 5E) and the dissociation rate (smallKd) was decreased (Fig. 5F) in E. coli K12–treated versus controlfemale rats, but in male rats, the mean values of small Kd werehigher in E. coli K12–treated rats versus controls (Fig. 5F).

Plasma levels and affinity kinetics of ACTH IgG

Plasma levels of both free and total anti-ACTH IgG werehigher in female than in male rats independently from E. colitreatment (Fig. 6A, B). Male but not female E. coli-gavaged ratsshowed increased levels of free anti-ACTH IgG (Fig. 6A). The ra-tios of free to total anti-ACTH IgG were affected in an oppositeway between female and male rats, with male rats showing arelative increase in the free fraction of ACTH IgG (Fig. 6C). Affinitykinetics analysis showed that both female and male E. coli-treated rats increased affinity of their IgG for ACTH (Fig. 6D).These changes were due to an increase of the association rate inboth females and males but the dissociation rate was signifi-cantly decreased only in female rats (Fig. 6F).

Effects of a-MSH on food intake and anxiety

Relevance of changes in a-MSH autoAbs properties to feedingand anxiety were studied after single peripheral administrationof a-MSH. Following the gavage, and then 5 d of food restriction,rats received an injection of a-MSH (100 mg/kg IP) and weregiven ad libitum access to food. No significant differences in 24-hfood intake during refeeding was observed between femaleE. coli K12 and control rats, but male rats that had been gavagedwith E. coli K12 showed lower food intake than controls (Fig. 7A).

Three d after refeeding, the anxiety in rats was tested in theO-maze after a single injection of a-MSH (100 mg/kg IP). No

significant differences in distance and time spent in open orclosed arms of the O-mazewere observed for female rats (Fig. 7C,D). In males, E. coli K12 group showed a tendency for increasedanxiety bymoving shorter distance and spending less time in theopen arms of the O-maze (Fig. 7C, D). Locomotor activity that wasassayed the day before the O-maze test did not significantlydiffer among all study groups of male and female rats (Fig. 7B).

cAMP production

a-MSH dose dependently stimulated cAMP production byHEK293 cells overexpressing MC4 R (Fig. 8A). Adding a-MSH toHEK293 cells that have not been transfected with MC4 R did notresult in a detectable increase of cAMP (data not shown). Addingto MC4 R overexpressing cells a-MSH together with rat IgG(0.5 mg/mL) also resulted in an a-MSH dose-dependent increasein cAMP production, with a left shift from the a-MSH stimulationcurve (Fig. 8A), resulting in two-to threefold increased sensitivityof MC4 R for a-MSH (Fig. 8B). The left shift was observedwith theIgG from bothmale and female rats but was more pronounced bythe IgG from females, resulting also in lower half maximaleffective concentration (EC50) mean values in female controlsversus both groups of male rats (Fig. 8B). No significant differ-ences for EC50 between the E. coli K12–treated and control maleand female rats were observed.

Discussion

The results of this study confirmed our hypothesis thatchronic provision of E. coli K12 to female and male rats sex-dependently affected feeding and anti-melanocortin hormoneantibody production. The E. coli K12 bacteria are largely presentin the environment and are known as a gut commensal strain inhumans and rodents [29]. However, its relative abundance in

Fig. 4. Effects of gavage with E. coli K12 on plasma levels of a-MSH autoAbs in female (A, C, E) and male (B,D,F) rats. (A, B) Plasma levels of free and (C, D) total a-MSH IgGautoAbs. (E, F) Plasma levels of free a-MSH IgM autoAbs. Differences before and after gavage, paired t tests, *P < 0.05, **P < 0.01, ***P < 0.001. Controls versus E. coli K12,Student’s t tests, *P < 0.05. Sex differences, female versus male rats, aversus control males, bversus E. coli K12–treated males, ANOVA, Tukey’s post-tests **P < 0.01,***P < 0.001. ANOVA, analysis of variance; autoAb, autoantibody; Ig, immunoglobulin; MSH, melanocyte-stimulating hormone.

N. Tennoune et al. / Nutrition 31 (2015) 498–507 503

microbiota with regard to the healthy or disease phenotypes andsex is not known. This study showed that E. coli K12were presentas gut commensals before E. coli K12 provision in female but notmale rats. These sex differences might, however, be specific torats and should not be extrapolated directly to humans. Indeed,increased presence of E. coli in gut microbiota was recentlyassociatedwith lower bodymass index inmainly female patientswith AN [30].

Regarding the general composition of gut microbiota, recentstudies revealed that there are indeed sex differences in rodentsand humans [31] and that a link exists between gut microbes andsex hormones [32]. It is possible that sex-related presence ofcertain gut bacteria may contribute to increased prevalence ofEDs in women. Our recent study revealed that ClpB proteinproduced by E. coli K12 and some other gram-negative bacteria isinvolved in the regulation of host appetite and emotion viastimulation of anti-ClpB antibodies cross-reactive with a-MSH,and that such mechanism may underlie involvement of gut

bacteria in the origin of EDs [23]. In the present study, we foundthat E. coli increased the plasma levels of anti-ClpB IgG in bothfemale and male rats and a tendency of increased anti-ClpB IgMwas seen in males. These results confirm a normal response bythe rat immune system to the bacterial-specific antigensfollowing E. coli gastrointestinal supplementation. This is inagreement with previous data showing that E. coli K12 are foundin the mesenteric lymph nodes after their intragastric load inmice [33]. However, because total IgG and IgM concentrationsdid not increase after E. coli K12 gavage, the immune response tothese bacteria appears to be antigen specific and justify E. coliK12 as a commensal strain of gut bacteria.

An overall opposite effect on energy balance was found afterthe E. coli K12 gavage in female andmale rats. Indeed, female ratsreceiving E. coli K12 showed increased body weight gain,whereas male rats showed a decrease. These data are in agree-ment with a previous study showing an increased body weightgain in female rats receiving another commensal strain of E. coli

Fig. 5. Effects of gavage with E. coli K12 on affinity kinetics between a-MSH and IgG of female and male rats as measured by SPR. Representative SPR sensorgrams and curvefit (thin lines) for the affinity kinetic analysis using the Langmuir’s 1:1 model between a-MSH and IgG from control (A) and E. coli K12–treated (B) female rats, using five serialdilutions of IgG; from top to bottom (nmol): 3360, 1680, 840, 420, and 210 and blank. The corresponding affinity kinetics values: (A) small Ka, 1.51 � 103 M�1 s�1, small Kd,3.02 � 103 s�1, KD, 2.0 � 106 M, c2 66,1; (B) small Ka, 6.88 � 103 M�1 s�1, small Kd, 2.62 � 103 s�1, KD, 3.81 � 107 M, c2 326. (C) Mean SPR response in resonance units (RU)5 min after dissociation. (D) Dissociation equilibrium constants (KD). (E) Association rates (small Ka) and (F) dissociation rates (small Kd). Controls versus E. coli K12, Student’st tests, *P < 0.05, **P < 0.01, ***P < 0.001. Sex differences, female versus male rats, aversus control males, bversus E. coli K12-treated males (C, E) ANOVA, Tukey’s post-tests*P < 0.05, ***P < 0.001. (D, F) Kruskal-Wallis test, Dunn’s post-tests ***P < 0.001. ANOVA, analysis of variance; autoAb, autoantibody; Ig, immunoglobulin; MSH, melanocyte-stimulating hormone; SPR, surface plasmon resonance.

N. Tennoune et al. / Nutrition 31 (2015) 498–507504

for 6 mo [34]. In our study, male rats had a decrease in foodintake after the start of E. coli gavage; a similar effect was seen inmale mice in a previous study [23]. Female rats, in contrast,seemed to completely tolerate E. coli K12 bacteria. Using PCR onfecal DNA, we verified whether the presence of E. coli K12 in therat microbiota before gavage might explain such differences, byinducing tolerance. Indeed, before gavage, E. coli K12 wereabundant in female rats but they were hardly detectable in malerats. Therefore, the sex differences in E. coli K12–induced a-MSHand ACTH autoAb changes might be related to the initial pres-ence of E. coli K12 in females but not in males, and hence, sup-plementation of bacteria to previously exposed or unexposedanimals resulted in different immune responses. Accordingly,female rats displayed higher levels of a-MSH IgG than malesbefore gavage, and the supplementationwith E. coli K12 resultedin an increased affinity of a-MSH IgG; affinity maturation being atypical response following initial exposure to the antigen.Increased levels of total (dissociated) a-MSH IgG in E. coli K12–treated rats also supported existence of higher affinity autoAbsfor a-MSH in female rats, possibly forming immune complexeswith either antigenic microbial proteins [35] or with anti-idiotypic antibodies [36]. In contrast, increased response in a-MSH IgM class of autoAbs was observed only in males, indicating

that E. coli K12 bacteria were rather new to male but not femalerats. Furthermore, increased dissociation rates of a-MSH IgGwere found in males, characterizing an initial antibody response.Hence, E. coli-induced differences in the affinity kinetics of a-MSH IgG in males and females may contribute to the sex-relatedchanges in energy balance.

To validate the relevance of a-MSH reactive autoAbs tofeeding in anxiety, in this study, after 3 wk of gavage, the ratswere challengedwith a small dose of a-MSH, the effects of whichcould be revealed if there was an enhancement of a-MSHsignaling (e.g., by an autoAb). The enhancing role of a-MSHbiological effects of a-MSH–reacting IgG in rats has been sug-gested by the existence of positive correlations between a-MSHIgG levels and a-MSH peptide concentrations [22,37]. We foundthat only male E. coli K12–treated rats displayed lower foodintake during refeeding after food restriction. Furthermore, onlyE. coli K12 males showed a tendency of increased anxietywithout differences in locomotor activity; the increase can belinked to hunger [38]. These data suggest that a-MSH–reactiveautoAbs properties in males were more favorable to enhancea-MSH anorexigenic and anxiogenic effects. This may concernmainly IgM because IgG were more efficient in female rats toactivate the MC4 R. In fact, the results of the in vitro study

Fig. 6. Effects of E. coli K12 on plasma levels and affinity kinetics between ACTH and IgG of female and male rats. Plasma levels of ACTH IgG free (A) and total (B) autoAbs andtheir ratios (C). Dissociation equilibrium constants (D), association (E), and dissociation (F) rates between ACTH and IgG assayed by SPR. A, C–F: Controls versus E. coli K12,Student’s t tests, *P < 0.05, **P < 0.01, ***P < 0.001, D: M-W test, males *P < 0.05. Sex differences, female versus male rats, aversus control males, bversus E. coli K12–treatedmales, Kruskal-Wallis test, Dunn’s post-tests*P < 0.05, **P < 0.01, ***P < 0.001. B: M-W test, #P < 0.05. ACTH, adrenocorticotropic hormone; autoAb, autoantibody; Ig,immunoglobulin; M-W, the Mann-Whitney; SPR, surface plasmon resonance.

N. Tennoune et al. / Nutrition 31 (2015) 498–507 505

revealed that rat IgG did not show any blocking properties ofa-MSH activation of MC4 R, but, instead, enhanced this interac-tion, which was manifested by the significant left shift of thecAMP activation curve after addition of rat IgG. The analysis ofEC50 for a-MSH showed that adding IgG from female ratsdecreased EC50 by three times, and of IgG frommale rats by two

Fig. 7. (A) Effects of a-MSH (100 mg/kg IP) on food intake during refeeding after food restrday before the anxiety test. (C, D) Effects of a-MSH (100 mg/kg IP) on anxiety in female aControls versus E. coli K12, Student’s t tests, *P < 0.05. Sex differences, female versus maleof variance; IP, intraperitoneal; MSH, melanocyte-stimulating hormone.

times. This is in agreement with higher levels of a-MSH-reactiveautoAbs in female versus male rats and a dose-dependent effectof anti-a-MSH IgG to activate MC4 R [39]. Although some ten-dencies of the cAMP curve shift between IgG from E. coliK12–treated and control rats were noted, they were not signifi-cant, which might explain why no significant differences in the

iction in female and male rats 5 d after end of gavage. (B) Locomotor activity test thend male rats as assayed in the elevated O-maze 5 d after the refeeding experiment.rats, aversus control males, ANOVA, Tukey’s post-tests ***P < 0.001. ANOVA, analysis

Fig. 8. Effects of IgG from E. coli K12–treated and control female and male rats on a-MSH–induced cAMP production by MC4 R-expressing HEK293 cells. (A) a-MSHdose dependently increased cAMP with or without rat IgG, but adding IgG shiftedthe activation curves to the left. (B) EC50 for a-MSH–induced activation of cAMP.ANOVA P < 0.0001, Bonferroni post-test ***P < 0.001. Sex differences, female versusmale rats, aversus control males, bversus E. coli K12–treated males, Student’s t test*P < 0.05, **P < 0.01. ANOVA, analysis of variance; cAMP, cyclic adenosine mono-phosphate; Ig, immunoglobulin; MC4 R, melanocortin receptors type 4; MSH,melanocyte-stimulating hormone.

N. Tennoune et al. / Nutrition 31 (2015) 498–507506

basal food intake were found between the two groups of rats,regardless of sex, at the end of the study.

Some similarities with a-MSH autoAb responses were seen inE. coli-induced ACTH-reactive IgG between male and female rats,as manifested by opposite change at the relative levels of freeand total autoAbs. Such an increase of free versus total levelsfound in male rats may indicate a recent antigenic stimulationand biologically available IgG to bind ACTH. However, both E.coli-treated male and female rats showed an increase of IgG af-finity for ACTH, suggesting that these autoAbs are induced bydistinct from a-MSH bacterial mimetic proteins in agreementwith previous data [23]. Functional significance of such changesof ACTH IgG properties in the HPA axis and EDs needs to befurther clarified; the previous data showed that plasma levels ofACTH-reactive autoAbs strongly correlatedwithmaturity fears inpatients with EDs [19].

Since the increase in interest in the role played by gutmicrobiota in health and disease [40], several studies revealedthe relevance of microbiota to brain function and behavior[41–43]. Our work showed sex-dependent response to gut bac-teria on host energy metabolism, emotion, and specific immuneresponses, providing further evidence that sex matters in

medico-scientific research [44]. A recent epidemiologic studyconcluded that the immune-mediated mechanisms underlie EDs[45]. Here, we showed that supplementing rats with E. coli K12changed levels and affinity of a-MSH-reactive IgG, which wereincreased in female rats and associated with positive energybalance. In contrast, increased a-MSH IgM levels in male ratswere associated with a-MSH–mediated satiety and anxiety. Thereason behind these sex differences of a-MSH-autoAb produc-tion might be related to the presence of E. coli K12 in the residentmicrobiota before E. coli K12 gastrointestinal supplementation,creating a tolerance against bacterial antigen-mimetic proteinsof melanocortin hormones. Taken together, these findings sug-gest that sex differences in gut microbiota may differently affectfeeding and emotion via production of autoAbs cross-reactingwith a-MSH and ACTH. Further studies are needed to deter-mine if presence of E. coli or other gut bacteria can be involved inthe more frequent occurrence of EDs in women and whatnutritional strategies should be developed for their prevention.

Acknowledgments

The authors acknowledge Dr. Emmanuelle D�e, Rouen Uni-versity, for providing the stock of E. coli K12 and Dr. MartinePestel-Caron, Rouen University, for providing bacterial culturefacilities.

References

[1] Becker AE, Grinspoon SK, Klibanski A, Herzog DB. Eating disorders. N Engl JMed 1999;340:1092–8.

[2] Asarian L, Geary N. Sex differences in the physiology of eating. Am J PhysiolRegul Integr Comp Physiol 2013;305:R1215–67.

[3] Clegg DJ. Minireview: the year in review of estrogen regulation of meta-bolism. Mol Endocrinol 2012;26:1957–60.

[4] Cone RD. Studies on the physiological functions of the melanocortin sys-tem. Endocr Rev 2006;27:736–49.

[5] Berglund ED, Liu T, Kong X, Sohn J-W, Vong L, Deng Z, et al. Melanocortin 4receptors in autonomic neurons regulate thermogenesis and glycemia. NatNeurosci 2014;17:911–3.

[6] Yaswen L, Diehl N, Brennan MB, Hochgeschwender U. Obesity in the mousemodel of pro-opiomelanocortin deficiency responds to peripheral mela-nocortin. Nat Med 1999;5:1066–70.

[7] Krude H, Biebermann H, Luck W, Horn R, Brabant G, Gruters A. Severeearly-onset obesity, adrenal insufficiency and red hair pigmentation causedby POMC mutations in humans. Nat Genet 1998;19:155–7.

[8] Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q,Berkemeier LR, et al. Targeted disruption of the melanocortin-4 receptorresults in obesity in mice. Cell 1997;88:131–41.

[9] Farooqi IS, Keogh JM, Yeo GSH, Lank EJ, Cheetham T, O’Rahilly S. Clinicalspectrum of obesity and mutations in the melanocortin 4 receptor gene. NEngl J Med 2003;348:1085–95.

[10] Kokare DM, Dandekar MP, Singru PS, Gupta GL, Subhedar NK. Involvementof alpha-MSH in the social isolation induced anxiety- and depression-likebehaviors in rat. Neuropharmacology 2010;58:1009–18.

[11] Chaki S, Hirota S, Funakoshi T, Suzuki Y, Suetake S, Okubo T, et al. Anxiolytic-like and antidepressant-like activities of MCL0129 (1-[(S)-2-(4-fluorophenyl)-2-(4-isopropylpiperadin-1-yl)ethyl]-4-[4-(2-met hoxynaphthalen-1-yl)butyl]piperazine), a novel and potent nonpeptide antagonist of the melanocortin-4 receptor. J Pharmacol Exp Ther 2003;304:818–26.

[12] Kaye WH, Klump KL, Frank GK, Strober M. Anorexia and bulimia nervosa.Annu Rev Med 2000;51:299–313.

[13] Brandys MK, van Elburg AA, Loos RJF, Bauer F, Hendriks J, van derSchouw YT, et al. Are recently identified genetic variants regulating BMI inthe general population associated with anorexia nervosa? Am J Med GenetB Neuropsychiatr Genet 2010;153B:695–9.

[14] Hebebrand J, Fichter M, Gerber G, Gorg T, Hermann H, Geller F, et al. Ge-netic predisposition to obesity in bulimia nervosa: a mutation screen of themelanocortin-4 receptor gene. Mol Psychiatry 2002;7:647–51.

[15] Hebebrand J, Geller F, Dempfle A, Heinzel-Gutenbrunner M, Raab M,Gerber G, et al. Binge-eating episodes are not characteristic of carriers ofmelanocortin-4 receptor gene mutations. Mol Psychiatry 2004;9:796–800.

[16] Takagi K, Legrand R, Asakawa A, Amitani H, Francois M, Tennoune N, et al.Anti-ghrelin immunoglobulins modulate ghrelin stability and its orexigeniceffect in obese mice and humans. Nat Commun 2013;4:2685.

N. Tennoune et al. / Nutrition 31 (2015) 498–507 507

[17] Fetissov SO, Hallman J, Oreland L, af Klinteberg B, Grenbäck E, Hulting AL,et al. Autoantibodies against a-MSH, ACTH, and LHRH in anorexia andbulimia nervosa patients. Proc Natl Acad Sci U S A 2002;99:17155–60.

[18] Fetissov SO, Hamze Sinno M, Coëffier M, Bole-Feysot C, Ducrott�e P,Hökfelt T, et al. Autoantibodies against appetite-regulating peptide hor-mones and neuropeptides: putative modulation by gut microflora. Nutri-tion 2008;24:348–59.

[19] Fetissov SO, Harro J, Jaanisk M, Järv A, Podar I, Allik J, et al. Autoantibodiesagainst neuropeptides are associated with psychological traits in eatingdisorders. Proc Natl Acad Sci U S A 2005;102:14865–70.

[20] Karaiskos D, Mavragani CP, Sinno MH, D�echelotte P, Zintzaras E,Skopouli FN, et al. Psychopathological and personality features in primarySjogren’s syndromedassociations with autoantibodies to neuropeptides.Rheumatology 2010;49:1762–9.

[21] Schaefer JM, Fetissov SO, Legrand R, Claeyssens S, Hoekstra PJ, Verhulst FC,et al. Corticotropin (ACTH)-reactive immunoglobulins in adolescents inrelation to antisocial behavior and stress-induced cortisol response. TheTRAILS study. Psychoneuroendocrinology 2013;38:3039–47.

[22] Hamze Sinno M, Do Rego JC, Coëffier M, Bole-Feysot C, Ducrotte P,Gilbert D, et al. Regulation of feeding and anxiety by a-MSH reactive au-toantibodies. Psychoneuroendocrinology 2009;34:140–9.

[23] Tennoune N, Chan P, Breton J, Legrand R, Chabane YN, Akkermann K, et al.Bacterial ClpB heat-shock protein, an antigen-mimetic of the anorexigenicpeptide [alpha]-MSH, at the origin of eating disorders. Transl Psychiatry2014;4:e458.

[24] Laviano A, Meguid MM, Gleason JR, Yang ZJ, Renvyle T. Comparison of long-term feeding pattern between male and female Fischer 344 rats: influenceof estrous cycle. Am J Physiol Regul Integr Comp Physiol 1996;270:R413–9.

[25] Kuhnert P, Nicolet J, Frey J. Rapid and accurate identification of Escherichiacoli K-12 strains. Appl Environ Microbiol 1995;61:4135–9.

[26] Fetissov SO. Neuropeptide autoantibodies assay. Methods Mol Biol2011;789:295–302.

[27] Legrand R, Takagi K, Fetissov SO. Immunoglobulin G preparation fromplasma samples and analysis of its affinity kinetic binding to peptide hor-mones. Protocol Exchange; 2014. http://dx.doi.org/10.1038/protex.2014.004.

[28] Crawley JN, Bailey KR. Anxiety-related behaviors in mice. In: Buccafusco JJ,editor. Methods of behavior analysis in neuroscience. 2nd ed. Boca Raton,FL: CRC Press; 2009. p. 77–101.

[29] Mühldorfer I, Blum G, Donohue-Rolfe A, Heier H, Ölschläger T, Tschäpe H,et al. Characterization of Escherichia coli strains isolated from environ-mental water habitats and from stool samples of healthy volunteers. ResMicrobiol 1996;147:625–35.

[30] Million M, Angelakis E, Maraninchi M, Henry M, Giorgi R, Valero R, et al.Correlation between body mass index and gut concentrations of Lactoba-cillus reuteri, Bifidobacterium animalis, Methanobrevibacter smithii andEscherichia coli. Int J Obes 2013;37:1460–6.

[31] Bolnick DI, Snowberg LK, Hirsch PE, Lauber CL, Org E, Parks B, et al. Indi-vidual diet has sex-dependent effects on vertebrate gut microbiota. NatCommun 2014;5:4500.

[32] Markle JGM, Frank DN, Mortin-Toth S, Robertson CE, Feazel LM, Rolle-Kampczyk U, et al. Sex differences in the gut microbiome drive hormone-dependent regulation of autoimmunity. Science 2013;339:1084–8.

[33] Slack E, Hapfelmeier S, Stecher B, Velykoredko Y, Stoel M, Lawson MAE,et al. Innate and adaptive immunity cooperate flexibly to maintain host-microbiota mutualism. Science 2009;325:617–20.

[34] Karlsson CLJ, Molin G, Fak F, Johansson Hagslätt M-L, Jakesevic M,Hakansson A, et al. Effects on weight gain and gut microbiota in rats givenbacterial supplements and a high-energy-dense diet from fetal life throughto 6 mo of age. Br J Nutr 2011;106:887–95.

[35] Oldstone MB, Aoki T, Dixon FJ. The antibody response of mice to murineleukemia virus in spontaneous infection: Absence of classical immunologictolerance (AKR mice-complement-fixing antibodies-lymphocytic chorio-meningitis virus-immunofluorescence-glomerular deposits of antigen-antibody complexes). Proc Natl Acad Sci U S A 1972;69:134–8.

[36] Deloumeau A, Bayard S, Coquerel Q, D�echelotte P, Bole-Feysot C,Carlander B, et al. Increased immune complexes of hypocretin autoanti-bodies in narcolepsy. PLoS One 2010;5:e13320.

[37] Coquerel Q, Hamze Sinno M, Boukhettala N, Coëffier M, Terashi M,Bole-Feysot C, et al. Intestinal inflammation influences a-MSH reactiveautoantibodies: relevance to food intake and body weight. Psychoneuro-endocrinology 2012;37:94–106.

[38] Duclos M, Ouerdani A, Morm�ede P, Konsman JP. Food restriction-inducedhyperactivity: addiction or adaptation to famine? Psychoneur-oendocrinology 2013;38:884–97.

[39] Lucas N, Legrand R, Ouelaa W, Breton J, Tennoune N, Bole-Feysot C, et al.Effects of rabbit anti-a-melanocyte-stimulating hormone (a-MSH) immu-noglobulins on a-MSH signaling related to food intake control. Neuro-peptides 2014;48:21–7.

[40] Clemente JC, Ursell LK, Parfrey LW, Knight R. The impact of the gutmicrobiota on human health: an integrative view. Cell 2012;148:1258–70.

[41] Gonzalez A, Stombaugh J, Lozupone C, Turnbaugh PJ, Gordon JI, Knight R.The mind-body-microbial continuum. Dialogues Clin Neurosci 2011;13:55–62.

[42] Cryan JF, Dinan TG. Mind-altering microorganisms: The impact of the gutmicrobiota on brain and behaviour. Nat Rev Neurosci 2012;13:701–12.

[43] Forsythe P, Kunze W. Voices from within: gut microbes and the CNS. CellMol Life Sci 2013;70:55–69.

[44] Meguid MM. In the matter of sex. sex matters in medical-scientificresearch. Nutrition 2008;24:767–8.

[45] Raevuori A, Haukka J, Vaarala O, Suvisaari JM, Gissler M, Grainger M, et al.The increased risk for autoimmune diseases in patients with eating dis-orders. PLoS One 2014;9:e104845.