Priority Communication

os

Mns

culcha

lesubjected to a microbiome depletion/transplantation paradigm using microbiota isolated from donors on either ahigh-fat diet (HFD) or control diet. Following re-colonization, mice were subjected to comprehensive behavioral and

within individuals have been suggested to underlie variable host

BiologicalPsychiatryribosomal RNA sequencing and informatics have revealed that

modern diets high in fat and sugar trigger robust alterations in absence of obesity. Nonobese, adult male C57BL/6 mice were& 2015 Society of Biological Psychiatry 607

SEE COMMENTARY ON PAGE 600in learning, memory, and executive function in obese com-pared with nonobese patients (3,4) and likewise link obesity toenhanced depression and anxiety disorders (5,6). However,there are contradictory reports that dispute these ndings(7,8), suggesting that the cause of obesity-associated mentalillness is not obesity per se but rather one or more of thevariable manifestations of obesity.

One potential site whereby diet-induced obesity could affectphysiology is the gut microbiome, as recent advances in 16S

susceptibility to illness (13,14), including neuropsychiatric impair-ment (15,16). For example, specic alterations in colon bacteriaare associated with cognitive impairment in patients with hepaticencephalopathy (17), and clinical studies show that oral pro-biotics decrease anxiety and improve mental outlook (18,19).Furthermore, animal studies have shown that behavior andsynaptic plasticity are altered in germ-free mice and that thisphenotype is reversed by microbiome colonization (20). The aimof the present study was to test the hypothesis that the obesity-concomitant microbiome undermines behavior even in theincreased brain pathology and decreased brain function[reviewed in (2)]. For example, functional studies report decitsRESULTS: The mice given HFD microbiota had signicant and selective disruptions in exploratory, cognitive, andstereotypical behavior compared with mice with control diet microbiota in the absence of signicant differences inbody weight. Sequencing-based phylogenetic analysis conrmed the presence of distinct core microbiota betweengroups, with alterations in - and -diversity, modulation in taxonomic distribution, and statistically signicantalterations to metabolically active taxa. HFD microbiota also disrupted markers of intestinal barrier function,increased circulating endotoxin, and increased lymphocyte expression of ionized calcium-binding adapter molecule1, toll-like receptor 2, and toll-like receptor 4. Finally, evaluation of brain homogenates revealed that HFD-shapedmicrobiota increased neuroinammation and disrupted cerebrovascular homeostasis.CONCLUSIONS: Collectively, these data reinforce the link between gut dysbiosis and neurologic dysfunction andsuggest that dietary and/or pharmacologic manipulation of gut microbiota could attenuate the neurologiccomplications of obesity.

Keywords: Gut dysbiosis, Intestinal permeability, Mental health, Neurobehavior, Neuroinammation, Obesity,Psychiatric disease

http://dx.doi.org/10.1016/j.biopsych.2014.07.012

The etiology of most neuropsychiatric disorders is likelymultifactorial and based on genetic and environmental riskfactors (1). One potentially important environmental driver ofmental illness is obesity, which dramatically increases risk ofdepression, dementia, and stroke, and is associated with

the core gut microbiome (9). The human gastrointestinal tractharbors as many as 100 trillion bacteria from up to 1000 distinctspecies, and this dynamic population of microbes participates innumerous physiologic functions including nutrition/digestion,growth, inammation, immunity, and protection against patho-gens (1012). Accordingly, the varying combinations of bacteriaISSbiochemical analyses.Obese-type Gut MicrobiNeurobehavioral Changeof ObesityAnnadora J. Bruce-Keller, J. Michael Salbaum,Christopher M. Taylor, David A. Welsh, and Ha

ABSTRACTBACKGROUND: The prevalence of mental illness, partiHere, we test the hypothesis that obesity-associatedneurocognitive behavior in mice.METHODS: Conventionally housed, nonobese, adult maN: 0006-3223 Bta Inducein the Absence

eng Luo, Eugene Blanchard IV,-Rudolf Berthoud

arly depression and dementia, is increased by obesity.nges in gut microbiota are intrinsically able to impair

C57BL/6 mice maintained on a normal chow diet wereiological Psychiatry April 1, 2015; 77:607615 www.sobp.org/journal

BiologicalPsychiatryconventionally housed and maintained on chow diet but sub-jected to a microbiome depletion paradigm followed by adoptivetransfer of cecal plus colonic contents collected from donor micefed either a high-fat diet (HFD) or control diet (CD). Recipientmice were subjected to a battery of neuropsychological tests,followed by sequencing of gut microbiota and thorough bio-chemical evaluation of intestine, blood, and brain samples.

METHODS AND MATERIALS

Animals and Treatments

The Pennington Biomedical Research Center InstitutionalAnimal Care and Use Committee approved all experimentalprotocols, which were compliant with National Institutes ofHealth guidelines. To generate microbiota donor material,8-week-old male C57BL/6 mice (Jackson Laboratories, BarHarbor, Maine) were given regular chow diet (13% fat calories,Purina LabDiet 5001; LabDiet, St. Louis, Missouri) or high-fatdiet (60% fat calories, Research Diets D12492; ResearchDiets, Inc., New Brunswick, New Jersey) for 10 weeks (seeTable S1 in Supplement 1 for diet compositions). At the time ofmicrobiota harvest, the high-fat fed mice weighed 37.0 6 1.7 gand the chow-fed mice weighed 24.5 6 1.2 g. Mice wereeuthanatized and cecal plus colonic contents were harvested,pooled, and diluted fortyfold (weight:volume) in sterile water.After centrifugation at 800 rpm, the supernatant was aliquotedunder sterile conditions for storage at 2801C. Recipient3-month-old male C57BL/6 mice (Jackson Laboratories) weregroup-housed under standard laboratory conditions with freeaccess to water and chow diet (Purina LabDiet 5001). Micewere given a cocktail of ampicillin, gentamicin, metronidazole,and neomycin (all at .25 mg/day) and vancomycin (.125 mg/day) once daily for 14 consecutive days by oral gavage (21).Mice were re-colonized 72 hours later via daily oral gavage ofdonor microbiota (100 mL) for 3 days (22,23). To offsetpotential founder and/or cage effects (24) and to reinforcethe donor microbiota genotype, booster inoculations weregiven biweekly throughout the study. Body weight and com-position were measured regularly, and all mice were euthan-ized following behavioral testing. Plasma, lymphocytes,intestines, intestinal contents, and brains were collected, withdata compiled from 10 animals per group.

Behavioral Testing

All behavioral testing was conducted between 7:00 AM and1:00 PM and was recorded/analyzed using Any-Maze software(Stoelting Co., Wood Dale, Illinois) for unbiased quanticationof body location, orientation, distance, speed, and mobility/immobility. Detailed methods on behavioral assays areprovided in Supplement 1. Overall anxiety and exploratorybehavior were assessed using elevated plus (25) and openeld assays (26). Stereotypical behavior was assessed byquantifying marble burying during a 30-minute trial in a novelcage preloaded with 4 cm of clean bedding and 16 evenlyspaced marbles (27) (Figure 1D). Memory was measured using

a video-based fear conditioning system (Med-Associates, St.Albans, Vermont) that pairs a unique context (scent and cage)and unconditioned stimulus (auditory tone) with a repeatedfoot shock (day 1) and then quanties freezing behavior to the

608 Biological Psychiatry April 1, 2015; 77:607615 www.sobp.org/jocontext (day 2) and to the tone (day 3) as measures of memory(28). Behavioral tests were administered in the order listedabove over 2 weeks, beginning 3 weeks after the end ofantibiotic treatment (Figure 1A). To curtail carryover effects,the elevated plus, open eld, and marble-burying assays wereconducted during the rst week of testing with 48 hoursrecovery between each task, while fear conditioning wastested the following week (29,30).

16S Metagenomic Sequencing

Fecal samples were collected under aseptic conditions from allmice during the nal week of behavioral testing, while cecalsamples were collected aseptically at euthanasia. Sequencingand bioinformatics were performed by the Louisiana StateUniversity Microbial Genomics Resource Center. DNA wasisolated using QIAamp DNA Stool kits (Qiagen, Germantown,Maryland) modied to include a bead-beating step. After DNAisolation, 16S ribosomal DNA hypervariable regions V3 andV4 were polymerase chain reaction amplied using primerswith the V3F CCTACGGGAGGCAGCAG and V4R GGAC-TACHVGGGTWTCTAAT gene-specic sequences, Illuminaadaptors, and molecular barcodes as described in Kozichet al. (31) to produce 430 base pair (bp) amplicons. Sampleswere sequenced on an Illumina MiSeq (Illumina, San Diego,California) using V3 sequencing kit (300 bp paired end reads).The forward read les were processed through the UPARSEpipeline (drive5, Tiburon, California) (32), truncating reads to auniform length of 250 bp, then removing reads with qualityscores less than 16. Additional ltering removed reads thatappeared only once throughout all samples (singletons) andremaining unique reads were clustered into operational taxo-nomic units (OTU) at 97% similarity. Chimeric OTUs wereremoved as identied by UCHIME drive5 run against a goldstandard reference database of nonchimeric sequences.Finally, the original ltered reads (before dereplication) weremapped to the OTUs using USEARCH drive5 at 97% identity.QIIME 1.8 (open source, www.qiime.org) was used to pick andalign a representative set. The Ribosomal Database Projectclassier was used to assign a taxonomic classication to eachread in the representative set and a phylogenetic tree wasconstructed from the representative sequences. Among sam-ples, the minimum read count after ltering was 21,182, with amedian read count of 57,537. Relative abundance of each OTUwas examined at phylum, class, order, family, genus, andspecies levels. Alpha (within a community) and beta (betweencommunities) diversity metrics as well as taxonomic commun-ity assessments were produced using QIIME 1.8 scripts.

Plasma and Tissue Analyses

Whole blood was collected by cardiac puncture of terminallyanesthetized mice into ethylenediaminetetraacetic acid treatedtubes, and plasma and lymphocytes were isolated andanalyzed immediately or stored at 2801C. Endotoxin levelsin plasma were measured using a kinetic limulus amebocytelysate test (Lonza Group, Limited, Basel, Switzerland). Levels

Gut Dysbiosis and Neurologic Dysfunctionof bioactive lipids and hormones/adipokines were measuredas previously reported (33). Lymphocyte, colon, jejunum, andbrain (medial prefrontal cortex) samples were homogenizedand processed for Western blot with chemiluminescence as

urnal

Figure 1. High-fat diet-associated microbiota increases anxiety and stereotypical behaviors but decreases memory in mice. (A) Body weight duringdepletion (ABX), recolonization (microbiome transplant), and behavioral protocols shows no difference between mice transplanted with microbiota from high-fat diet (HFD) fed donors or control diet (CD) fed donors. (B) Time spent exploring the open arms of the elevated plus maze was signicantly reduced in HFDas compared with CD mice. (C) Time spent in the inner zone of the open eld (left panel) but not mean speed (center panel) or total distance traveled (rightpanel) was signicantly decreased in HFD mice as compared with CD mice. (D) Marble-burying behavior was signicantly increased in HFD versus CD mice,as shown by representative images of marble placement before (Init) or after the 30-minute trail with mice transplanted with CD- or HFD-associatedmicrobiota and quantitative analysis (right panel). (E) Following fear conditioning, freezing behavior to context on training day 2 was not different betweengroups (left panel), but conditioned freezing to the tone on day 3 was signicantly reduced in HFD as compared with CD mice (right panel). All data arepresented as mean 6 SEM of 10 mice per group and *p , .05 based on t tests or analysis of variance.

Gut Dysbiosis and Neurologic Dysfunction

Biological Psychiatry April 1, 2015; 77:607615 www.sobp.org/journal 609

BiologicalPsychiatry

(Figure 1E). In addition, the averaged slopes of behavioral

BiologicalPsychiatrypreviously reported (33). For accurate quantication acrossblots, samples from both treatment groups were included ineach individual blot. Data were rst calculated as a ratio ofexpression over tubulin expression, and then expression inmice with HFD microbiota was calculated/presented as per-cent expression in control (CD) mice.

Statistical Analyses

All behavioral and biochemical data were analyzed using Prismsoftware (GraphPad Software, Inc., La Jolla, California) and aredisplayed as mean 6 standard error of measurement. Bodyweight and fear conditioning behavior were analyzed with two-way repeated measures analysis of variance (ANOVA) todetermine main effects of treatment and duration, followedby planned Bonferroni post hoc comparisons to determinedifferences between groups. All other behavioral and bio-chemical data were analyzed by unpaired t tests. Statisticalsignicance for all analyses was accepted at p , .05.

For sequencing data, alpha diversity rarefaction curveswere produced by plotting several diversity metrics againstthe number of sequences considered from a sample. Subse-quent analysis of diversity was performed at a depthof 20,000 sequences per sample. Statistical signicancewas compared using a nonparametric permutation test fora pairwise comparison of categories. The p values wereBonferroni-corrected for multiple testing. Beta diversity, prin-ciple coordinates analysis plots were produced, using bothweighted (considers abundance of each species) andunweighted (considers presence/absence of species) UniFracmetrics. Plots were visualized using the Emperor 3D Viewer(open source, www.emperor.colorado.com). Statistical signi-cance was assessed with a nonparametric permutation test tocompare a chosen category. ANOVA was used to test fordifferences in relative abundance of specic OTUs for eachgroup. An unweighted g-test was used to evaluate thestatistical signicance of the presence/absence of OTUsacross categories. DESeq2 software (open source, www.bioconductor.org) was also used to test for differential repre-sentation of OTUs and also to identify in an unbiased mannerall individual OTUs in which the group difference reachedstatistical signicance using mean normalized sequencecount level higher than 30 counts and false discovery rate(FDR) , .05 criteria (Wald statistics with Benjamini-Hochbergcorrection) in either the CD or HFD group.

RESULTS

HFD-Derived Gut Microbiota Impair BehavioralPerformance in Mice

All animals tolerated the antibiotic regimen with no overteffects other than a mild, approximate 10% loss of bodyweight (Figure 1A). Quantitative real-time polymerase chainreaction based analyses of 16S RNA levels in fecal samplescollected from mice midway through the antibiotic treatmentrevealed an approximate 90% to 95% reduction in fecal

bacteria burden compared with matched but untreated mice(fecal DNA concentration 82,502.1 6 18,255 mg/g in controlsamples, 3417.4 6 1212 mg/g in samples following antibioticexposure). Mice were subjected to thorough behavioral

610 Biological Psychiatry April 1, 2015; 77:607615 www.sobp.org/jowaveforms depicting freezing of individual mice in response totone were signicantly different between the groups (211.8 61.48 in CD mice; 25.5 6 .6 in HFD mice; t18 5 4.05, p , .001),suggesting attenuated within-session extinction of fear behav-ior in mice with HFD microbiota.

Microbiota Transplantation Results in DistinctPhylogenetic Proles

Fecal and cecal microbiota compositions in recipient micewere analyzed by V3 16S ribosomal DNA phylogenetics asdescribed in Methods and Materials. Analysis of cecal andfecal samples from all mice demonstrated that the adoptivetransfer protocol was successful in producing distinct coremicrobiomes in the two groups of mice (Tables S2 and S3 inSupplement 1). Additionally, HFD microbiota demonstratedsignicantly reduced alpha diversity relative to CD samples, aswell as greater evenness (cecal samples: Chao1 p 5 .0362,observed species p 5 .0302; fecal samples: Chao1 p 5 .0033,observed species p 5 .004; Figure 2A and Figure S1 inSupplement 1). Rarefaction curves of fecal and cecal diversity(Figures S1 and S2 in Supplement 1) also support thisinterpretation as well. Evaluation of beta diversity metricsbased on unweighted UniFrac distances showed that thecommunity structures observed in the HFD samples weresignicantly different (p 5 .0001) from the communitiesdetected in the CD samples (Figure 2B). Visualization byprincipal coordinates analysis demonstrated that CD andHFD samples formed distinct clusters and that for eachcondition, cecal and fecal samples formed a cluster aroundthe initial inoculum (Figure 2B). The taxonomical distributionphenotyping starting 3 weeks after recolonization with eithermicrobiota from high-fat or chow-fed mice (Figure 1A). Explor-atory and anxiety-based behavior assessed using the elevatedplus maze revealed that mice with HFD-associated microbiotaspent signicantly less time (t18 5 2.32, p , .05) in the openarms of the maze (Figure 1B). The open eld assay likewiserevealed that mice with HFD-associated microbiota spentsignicantly less time (t18 5 2.13, p , .05) in the inner zoneof the open eld (Figure 1C). Overall, locomotor activityassessed in the open eld showed no differences in meanspeed or total distance traveled between CD and HFD groups(Figure 1C), suggesting that decreased exploratory behavior inmice with HFD microbiota reects increased anxiety notdecreased motor function. Mice were tested for marble bury-ing, a measure of compulsive, anxiety-like behavior (34). HFD-shaped microbiota were associated with a signicant increase(t18 5 2.64, p , .05) in marble burying (Figure 1D). Finally, thefear conditioning assay was used to measure memory. Sig-nicant differences in freezing behavior were observed on thethird day of the fear conditioning test, when cued learning(freezing in a novel context in response to the tone) wasassessed. Post hoc analyses revealed that freezing inresponse to tone was signicantly decreased in mice withHFD microbiota as compared with mice with CD microbiota

Gut Dysbiosis and Neurologic Dysfunctionwithin groups at phylum, family, and genus levels for the cecal(Figure 2C; Figure S3 in Supplement 1) and fecal samples(Figure S4 in Supplement 1) revealed the divergent composi-tion of communities in CD compared with HFD samples.

urnal

BiologicalsychiatryGut Dysbiosis and Neurologic DysfunctionThe UPARSE pipeline (32) was used to identify OTUs, andANOVA revealed signicant differences in the relative abun-dance of OTUs between HFD and CD microbiota (see Tables S4and S5 in Supplement 1 for Bonferroni-corrected and FDRcontrolled values). To corroborate ANOVA data with statisticalmethods better suited for sequence count data, DESeq2software was used to test for differential representation of OTUsin cecal samples. Individual OTUs in which the group differencereached statistical signicance using mean normalizedsequence count level higher than 30 counts and FDR ,.05criteria (Wald statistics with Benjamini-Hochberg correction) ineither the CD or HFD group were identied (Table S6 inSupplement 1). Of the 104 OTUs passing these DESeq2 lters,53 OTUs were higher in the HFD samples compared with CD,whereas 51 OTUs were higher in CD. Of the 104 signicantlydifferent OTUs, 91 belong to the phylum Firmicutes, with 90 ofthese coming from class Clostridiales. These unbiased analysesshow that the overall distinction between HFD and CD is basedon shifts in the representation of individual OTUs within

Figure 2. Effects of transplantation protocol on recipient gut microbiome divdonor and recipient mice were analyzed using 16S ribosomal RNA sequencing. (show that mice with high-fat diet (HFD) microbiota exhibited a statistically signic(CD) microbiota. Red line: median; black lines: range of values. (B) Scaled principcecal and fecal samples from individual recipient mice. Red and orange circles derepresent cecal and fecal samples from HFD-treated mice. The pooled samplesmagenta for the HFD donor pool. -diversity was found to be statistically signiMicrobiota membership is reected in bar diagrams depicting the taxonomic distfamily, and genus levels. Microbiota from the HFD-treated group show higher reimages together with the detailed color codes are shown in Figures S3 and S4 inmice at the level of Lachnospiraceae and Ruminococcaceae, two families withinHFD-treated group versus the CD-treated group were determined in DESeq2. Seither to Lachnospiraceae or to Ruminococcaceae (family level as maximum taxoand plotted relative to the CD group. This analysis demonstrated shifts in reprCoordinate 2.

BiologicalPClostridiales rather than a binary shift from, for example,Bacteroidetes in the CD samples to Firmicutes in the HFDsamples. Looking at specic orders within Clostridiales(Figure 2D; Figure S5 in Supplement 1), some orders are presentat signicantly higher levels in the HFD group (17 members ofLachnospiraceae; 9 members of Ruminococcaceae), whereasothers are lower (21 members of Lachnospiraceae; 9 membersof Ruminococcaceae). Similar ndings were obtained in fecalsamples (data not shown). Finally, the list of differently repre-sented OTUs was queried for presumed benecial bacteria,such as Akkermansia muciniphila. Akkermansia muciniphila was5.4-fold lower in HFD samples compared with CD (FDR 5 .06),indicating that this species of bacteria may be associated with ahealthier microbiome, as suggested previously (35). Likewise,the presumably detrimental Bilophila sp. (belonging to Desulfo-vibrionaceae) was strongly enriched ( 300-fold; FDR 5 2.5 310225) in HFD microbiota, comprising .78% of the microbialcommunity in this group. Conversely, Bilophila sp. was barelydetectable in CD samples at .0024%.

ersity and population. Cecal and fecal microbiome populations from bothA) Box plots that were generated to depict differences in Chao1 -diversityant (*p 5 .0362) reduction in -diversity compared with mice with control dietal coordinate analysis to visualize the unweighted UniFrac distances of bothpict cecal and fecal samples from CD-treated mice; green and yellow circlesused as donor microbiota are shown in blue for the CD donor pool and incantly different between the CD- and HFD-treated groups (p 5 .0001). (C)ribution within cecal samples within the CD and HFD groups at the phylum,presentation of Clostridiales (family: purple; genus: blue). Higher resolutionSupplement 1. (D) Microbiome differences between CD- and HFD-treatedthe order of Clostridiales. Statistically signicant fold changes between theignicant fold changes for individual operational taxonomic units belongingnomical depth for these operational taxonomic units) were log2-transformedesentation within each family. PC1, Principle Coordinate 1; PC2, Principle

Psychiatry April 1, 2015; 77:607615 www.sobp.org/journal 611

Gut Dysbiosis and Neurologic DysfunctionBiologicalPsychiatryHFD-Derived Gut Microbiota Increase IntestinalPermeability, Systemic Inammation, and BrainInammation

Postmortem studies were conducted to identify potentialpathways whereby altered gut microbiota impaired behavior.Analysis of blood glucose and bioactive hormones/lipids inplasma, as well as body weight and body composition,revealed no signicant group differences, demonstrating thatobesity and metabolic syndrome/dysfunction were notinduced by HFD microbiota (Table S7 in Supplement 1).

To determine if transplantation with HFD-shaped micro-biota altered intestinal barrier function, the expression ofmarkers of intestinal inammation and permeability and alsocirculating endotoxin and inammatory markers wereassessed (see Figure S6 in Supplement 1 for representativeWestern blots). Compared with CD, mice with HFD micro-biota had signicantly decreased occludin (t18 5 4.95, p ,.001) expression in the jejunum (Figure 3A). Additionally,expression of inducible nitric oxide synthase (t18 5 3.70, p, .01) and phosphorylation of the p65 subunit of nuclearfactor kappa B (t18 5 4.13, p , .001) were increased, whileoccludin (t18 5 3.32, p , .01) and claudin-3 (t18 5 4.13, p, .001) were decreased, in colons of HFD mice (Figure 3A),indicating increased intestinal inammation and permeabilityin mice with HFD microbiota. In addition, data showedsignicantly increased plasma endotoxin (t18 5 2.64,p , .05) in mice with HFD microbiota (Figure 3B), whileevaluation of isolated lymphocytes revealed increasedexpression of the macrophage marker ionized calcium-binding adapter molecule 1 (t16 5 2.59, p , .05) and toll-like receptor 4 (t16 5 2.73, p , .05) in mice with HFDmicrobiota (Figure 3B).

To determine the effects of HFD-shaped microbiota onbrain, protein markers of brain injury and inammation werequantied (see Figure S6 in Supplement 1 for representativeWestern blots). Analyses were thematically split into evalua-tions of inammation/gliosis, cerebrovascular integrity, andsynaptic density and were conducted in the medial prefrontalcortex, a brain structure involved in both anxiety and cognitivebehaviors in mice (36). Compared with mice with CD-shapedmicrobiota, expression of the microglial marker ionizedcalcium-binding adapter molecule 1 (t18 5 3.48, p , .01),toll-like receptor 2 (t18 5 2.72, p , .05), and toll-like receptor 4(t18 5 2.83, p , .05) were increased in HFD mice (Figure 3B).Additionally, mice with HFD-associated microbiota haddecreased levels of the tight junction proteins zona occludensprotein 1 (ZO-1) (t18 5 2.32, p , .05) and claudin-5 (t18 5 4.11,p , .001) and increased expression of matrix metallo-proteinase 9 (t18 5 2.29, p , .05; Figure 3B). While overallexpression of the synaptic marker proteins synapse-associated protein 97 and synapsin 1 were similar in bothgroups, levels of phosphorylated synapsin 1 were signicantlyreduced (t18 5 2.26, p , .05) in mice with HFD-shapedmicrobiota (Figure 3B). Finally, levels of brain-derived neuro-trophic factor were assessed in the medial prefrontal cortex,

but there were no signicant differences in soluble brain-derived neurotrophic factor in mice with HFD microbiota ascompared with mice with CD microbiota (Table S7 inSupplement 1).

612 Biological Psychiatry April 1, 2015; 77:607615 www.sobp.org/joDISCUSSION

The present ndings represent the rst denitive evidence thathigh-fat diet-induced changes to the gut microbiome aresufcient to disrupt brain physiology and function in theabsence of obesity. Specically, data show that transplantationof microbiota shaped by high-fat diet, but not control low-fatdiet, caused signicant and selective disruptions in exploratory,cognitive, and stereotypical behavior in conventionally housed,nonobese, diet-nave mice. Overall, these data are in agreementwith the extensive body of literature describing the sensitivity ofthe brain to diet-induced obesity (28,37) and the growingnumber of studies linking gut microbiota to central nervoussystem health and behavior (20,38,39). For example, there is areported high comorbidity between psychiatric syndromes,including depression and anxiety, with gastrointestinal disor-ders, while conversely, recent studies link probiotics to positivechanges in mood and behavior [reviewed in (40)]. Furthermore,changes in microbiota appear to mediate weight gain com-monly associated with antipsychotic administration (41,42),which has been likewise linked to improvements in coreschizophrenia symptoms, depression, and overall mental func-tioning (43). It should be pointed out, however, that reports haveshown that high-fat diet consumption can allay anxiety anddepressive-like behaviors in mice subjected to chronic socialstress (44). Thus, these data underscore the strong but complexinuence of diet-induced changes to the gut microbiome onstress-induced behaviors and emphasize the clinical utility ofthe gut-brain axis as a target for future therapeutic intervention.

The signicant behavioral phenotype of the mice described inthis report, combined with the established association betweenpsychiatric conditions and gastrointestinal symptoms, supportthe concept of a microbiome-gut-brain axis (45,46), but themechanisms whereby gut microbes affect behavior are notunderstood. Gut microbial metabolism is known to producecatecholamines, histamine, and/or other neuroactive mediatorsthat can directly stimulate the local enteric nervous system and/or primary afferent bers of vagal or dorsal root origin (38,47).Indeed, reports have shown that the probiotic bacteriumLactobacillus rhamnosus can directly increase single- andmulti-unit ring rates of the mesenteric nerve bundle and candecrease stress-induced corticosterone and anxiety/depressionin mice (48,49). Moreover, the positive behavioral effects ofLactobacillus rhamnosus are abolished by vagotomy (48). Inaddition to direct interactions with neural processes, immuneactivation and inammation participate in nearly all neurologic/psychiatric disorders (50,51), and gut dysbiosis might alter brainfunction via this pathway. Indeed, the increases in gram-negative Proteobacteria within the gut, endotoxins in the blood,and inammatory markers in the brain collectively suggest thatintestinal permeability and inammation link HFD microbiota tobehavioral dysfunction. In further support of this scenario,transplantation of microbiota from obese donors to germ-freerecipients has been shown to disrupt intestinal tight junctionprotein and increase translocation of bacteria into the blood-stream (52). In relation to neurologic impairment, alterations to

gut microbiota and disrupted intestinal barrier function are seenin mouse models of autism spectrum disorder (53). Collectively,these data suggest that unhealthy diet-induced alterations togut microbiota could boost the prevalence and/or severity of

urnal

numerous neurologic conditions that involve inammation,including autoimmune disease, autism, and Alzheimers disease.

While previous reports have shown that gut microbiometransplantations into germ-free mice can replicate manyaspects of the obese phenotype (54,55), this is the rstdemonstration that high-fat shaped gut microbiota canintrinsically and adversely affect neurologic function/physiol-ogy in conventionally housed mice, even in the absence ofaltered diet, adiposity, or metabolic syndrome. A variety oftools and techniques have been developed to study the gutmicrobiome; microbiota, including introduction into germ-freerecipients; antibiotic use; administration of prebiotics andprobiotics; and specic gastrointestinal infection. Indeed, theuse of gnotobiological methods on experimental animals hasbeen indispensable in establishing the signicance of gutmicrobiota to mammalian physiology (56). However, thereare several characteristics of germ-free mice, outside of

Gut Dysbiosis and Neurologic Dysfunction

Biological

BiologicalPsychiatrychanges in cecal size and bowel motility, that undermine theirutility and physiologic relevance. For example, germ-free miceare well known to be smaller than conventional mice, withdecreased cardiac output and notably underdevelopedimmune systems (57,58). As any of these confounds couldaffect the development and function of the brain, we optedinstead to use a strategy whereby donor microbiota wereadoptively transferred to conventionally housed mice followingantibiotic-based microbial depletion. While both the depletionand recolonization protocols are based on established meth-ods (2124), there are limitations of the antibiotic-based modelthat could have affected the outcome of our study. It is likelythat the antibiotic regimen did not entirely deplete the recipientmicrobiome, which could differentially affect recolonization byspecic bacteria. It is also possible that repeated gavage and/or sustained systemic antibiotic exposure could contribute tosome of the behavioral alterations obseved. However, as bothgroups were treated equally, the impact of such potentialartifacts is minimized.

Alterations in microbiome composition following manipulationhave been evaluated in the past in an attempt to identify

Figure 3. Transplantation with microbiota shaped by high-fat dietdisrupts intestinal barrier proteins and increases systemic and braininammation. (A) Relative expression of tight junction proteins occludin,claudin-2, and claudin-3 in jejunum (left). Expression of inducible nitric oxidesynthase (iNOS), phosphorylated p65 (phos-p65), and tight junction pro-teins occludin, claudin-2, and claudin-3 in colon (right) in mice with high-fatdiet (HFD) microbiota relative to control-diet (CD) mice. (B) Levels of plasmaendotoxin (Endtxn) and lymphocyte expression of macrophage markers(ionized calcium-binding adapter molecule 1 [Iba1]) and toll-like receptor(TLR) 4 in mice with HFD microbiota as compared with CD mice. (C)Markers of inammation, cerebrovascular integrity, and synaptic density intissue homogenates prepared from the medial prefrontal cortex. Graphsdepict increased microgliosis (Iba1) and TLR2 and TLR4 expression,increased matrix metalloproteinase (MMP) 9 expression, and decreasedexpression of endothelial tight junction proteins (zona occludens protein 1[ZO-1] and claudin-5) and phosphorylated synapsin-1 (P-Synap) in HFDmice. All data depict mean 6 SEM expression in mice with HFD microbiotapresented as % CD mice (100% line on graph). *p , .05, **p , .01, and***p , .001, based on t tests. See Figure S6 in Supplement 1 for

representative images of all Western blot data. GFAP, glial brillary acidicprotein; P-synap, phosphorylated synapsin I; SAP97, synapse-associatedprotein 97; Synap, synapsin I.

Psychiatry April 1, 2015; 77:607615 www.sobp.org/journal 613

Gut Dysbiosis and Neurologic DysfunctionBiologicalPsychiatrybenecial core microbiota. For example, studies on high-fat diet-induced gut microbiota have reported a shift in the relativeabundance of two major phyla, a reduction in Bacteroidetes andan increase in Firmicutes (54,55). Furthermore, abundance ofthese two phyla was shifted in the opposite direction after weightloss or gastric bypass surgery (59,60); thus, it has beenproposed that the balance between these two phyla mightreect the balance of unhealthy and healthy microbiota. How-ever, our data suggest that this binary distinction does notsufciently reect the complexity of diet-induced changes to thegut microbiome, as has been suggested in previous investiga-tions of diet-induced obesity and gut microbiota (61). Speci-cally, our data indicate that shifts within the Firmicute phylumdrive the overall distinction of HFD from CD, rather than phylum-wide shifts from Bacteroidetes to Firmicutes (Figure 2; Figure S5in Supplement 1). While it is currently not possible to identify thespecic alterations or population shifts that drive the observedbehavioral/biochemical alterations, the relative abundance ofpurportedly benecial and harmful species in each group wasprobed. Akkermansia muciniphila, a presumed benecial spe-cies, was 5.4-fold lower in HFD samples compared with CD,indicating that this species of bacteria may be associated with ahealthier microbiome, as suggested previously (35). Likewise, thepresumed detrimental Bilophila sp. (belonging to Desulfovibrio-naceae) was strongly enriched ( 300-fold; FDR 5 2.5 3 10225)in HFD microbiota, comprising .78% of the microbial communityin this group. Conversely, Bilophila sp. was barely detectable inCD samples at .0024%. As member(s) of the phylum Proteobac-teria, Bilophila sp. may be partially responsible for the increase inendotoxin observed in the serum from HFD-treated micecompared with CD-treated mice (Figure 3). Indeed, higherBilophila wadsworthia have been repeatedly found in humanpatients suffering from intestinal diseases (62,63). The collectiveidentication of specic bacterial species/populations drivingadverse physiologic responses to diet will facilitate the futuredesign of personalized microbiomes that optimize physiologicfunction in the context of modern diets/lifestyles. Overall, thesedata strongly suggest that therapeutic manipulation of themicrobiome, which should be highly responsive compared withexisting clinical targets, could dramatically mitigate the preva-lence and/or severity of neuropsychiatric disorders.

ACKNOWLEDGMENTS AND DISCLOSURESThis work was supported by the National Institutes of Health (DK047348 toHRB) and also used Pennington Biomedical Research Center (AnimalPhenotyping) and Louisiana State University (Microbial Genomics ResourceCenter), which are funded, in part, by the National Institutes of Health(P20-RR021945, P30-DK072476, and P60-AA009803).

We thank Dr. Barry Robert for expert veterinary assistance related toantibiotic administration.

The authors declare no biomedical nancial interests or potentialconicts of interest.

ARTICLE INFORMATIONFrom the Pennington Biomedical Research Center (AJB-K, JMS, H-RB),Louisiana State University System, Baton Rouge, Louisiana Department of

Internal Medicine (DAW), Louisiana State University Health Sciences CenterNew Orleans, New Orleans, Louisiana; and Department of Microbiology,Immunology & Parasitology (ML, EB, CMT), Louisiana State University HealthSciences Center New Orleans, New Orleans, Louisiana.

614 Biological Psychiatry April 1, 2015; 77:607615 www.sobp.org/joAddress correspondence to Annadora J. Bruce-Keller, Ph.D., PenningtonBiomedical Research Center/Louisiana State University, Inammation andNeurodegeneration Laboratory, 6400 Perkins Road, Baton Rouge, LA70808; E-mail: annadora.bruce-keller@pbrc.edu.

Received Apr 15, 2014; revised Jun 26, 2014; accepted Jul 6, 2014.

Supplementary material cited in this article is available online at http://dx.doi.org/10.1016/j.biopsych.2014.07.012.

REFERENCES1. Shih RA, Belmonte PL, Zandi PP (2004): A review of the evidence from

family, twin and adoption studies for a genetic contribution to adultpsychiatric disorders. Int Rev Psychiatry 16:260283.

2. Bruce-Keller AJ, Keller JN, Morrison CD (2009): Obesity and vulner-ability of the CNS. Biochim Biophys Acta 1792:395400.

3. Elias MF, Elias PK, Sullivan LM, Wolf PA, DAgostino RB (2003): Lowercognitive function in the presence of obesity and hypertension: TheFramingham Heart Study. Int J Obes Relat Metab Disord 27:260268.

4. Waldstein SR, Katzel LI (2006): Interactive relations of central versustotal obesity and blood pressure to cognitive function. Int J Obes(Lond) 30:201207.

5. Needham BL, Epel ES, Adler NE, Kiefe C (2010): Trajectories ofchange in obesity and symptoms of depression: The CARDIA study.Am J Public Health 100:10401046.

6. Ma J, Xiao L (2010): Obesity and depression in US women: Resultsfrom the 2005-2006 National Health and Nutritional ExaminationSurvey. Obesity (Silver Spring) 18:347353.

7. Atlantis E, Baker M (2008): Obesity effects on depression: Systematicreview of epidemiological studies. Int J Obes (Lond) 6:881891.

8. Rivenes AC, Harvey SB, Mykletun A (2009): The relationship betweenabdominal fat, obesity, and common mental disorders: Results fromthe HUNT study. J Psychosom Res 66:269275.

9. Kim KA, Gu W, Lee IA, Joh EH, Kim DH (2012): High fat diet-inducedgut microbiota exacerbates inammation and obesity in mice via theTLR4 signaling pathway. PLoS One 7:e47713.

10. Robles Alonso V, Guarner F (2013): Linking the gut microbiota tohuman health. Br J Nutr 109(suppl 2):S21S26.

11. Collins SM, Kassam Z, Bercik P (2013): The adoptive transfer ofbehavioral phenotype via the intestinal microbiota: Experimentalevidence and clinical implications. Curr Opin Microbiol 16:240245.

12. Jarchum I, Pamer EG (2011): Regulation of innate and adaptive immunityby the commensal microbiota. Curr Opin Immunol 23:353360.

13. Aziz Q, Dor J, Emmanuel A, Guarner F, Quigley EM (2013): Gutmicrobiota and gastrointestinal health: Current concepts and futuredirections. Neurogastroenterol Motil 25:415.

14. Schwabe RF, Jobin C (2013): The microbiome and cancer. Nat RevCancer 13:800812.

15. Douglas-Escobar M, Elliott E, Neu J (2013): Effect of intestinalmicrobial ecology on the developing brain. JAMA Pediatr 167:374379.

16. Dinan TG, Quigley EM (2011): Probiotics in the treatment of depres-sion: Science or science ction? Aust N Z J Psychiatry 45:10231025.

17. Bajaj JS, Ridlon JM, Hylemon PB, Thacker LR, Heuman DM,Smith S, et al. (2012): Linkage of gut microbiome with cognition inhepatic encephalopathy. Am J Physiol Gastrointest Liver Physiol 302:G168G175.

18. Bested AC, Logan AC, Selhub EM (2013): Intestinal microbiota,probiotics and mental health: From Metchnikoff to modern advances.Part III - convergence toward clinical trials. Gut Pathog 5:4.

19. Chen X, DSouza R, Hong ST (2013): The role of gut microbiota in thegut-brain axis: Current challenges and perspectives. Protein Cell 4:403414.

20. Diaz Heijtz R, Wang S, Anuar F, Qian Y, Bjrkholm B, Samuelsson A,et al. (2011): Normal gut microbiota modulates brain development and

behavior. Proc Natl Acad Sci U S A 108:30473052.

21. Reikvam DH, Erofeev A, Sandvik A, Grcic V, Jahnsen FL, Gaustad P,et al. (2011): Depletion of murine intestinal microbiota: Effects on gutmucosa and epithelial gene expression. PLoS One 6:e17996.

urnal

Gut Dysbiosis and Neurologic DysfunctionBiologicalsychiatry22. Heimesaat MM, Plickert R, Fischer A, Gbel UB, Bereswill S (2013):Can microbiota transplantation abrogate murine colonization resist-ance against Campylobacter jejuni? Eur J Microbiol Immunol (Bp) 3:3643.

23. Ubeda C, Bucci V, Caballero S, Djukovic A, Toussaint NC, Equinda M,et al. (2013): Intestinal microbiota containing Barnesiella speciescures vancomycin-resistant Enterococcus faecium colonization. InfectImmun 81:965973.

24. McCafferty J, Mhlbauer M, Gharaibeh RZ, Arthur JC, Perez-ChanonaE, Sha W, et al. (2013): Stochastic changes over time and not foundereffects drive cage effects in microbial community assembly in amouse model. ISME J 7:21162125.

25. Walf AA, Frye CA (2007): The use of the elevated plus maze as anassay of anxiety-related behavior in rodents. Nat Protoc 2:322328.

26. Choleris E, Thomas AW, Kavaliers M, Prato FS (2001): A detailedethological analysis of the mouse open eld test: Effects of diazepam,chlordiazepoxide and an extremely low frequency pulsed magneticeld. Neurosci Biobehav Rev 25:235260.

27. Angoa-Prez M, Kane M, Briggs DI, Francescutti DM, Kuhn DM(2013): Marble burying and nestlet shredding as tests of repetitive,compulsive-like behaviors in mice. J Vis Exp 82:50978.

28. Freeman LR, Zhang L, Nair A, Dasuri K, Francis J, Fernandez-Kim SO,et al. (2013): Obesity increases cerebrocortical reactive oxygenspecies and impairs brain function. Free Radic Biol Med 56:226233.

29. Paylor R, Spencer CM, Yuva-Paylor LA, Pieke-Dahl S (2006): The useof behavioral test batteries, II: Effect of test interval. Physiol Behav 87:95102.

30. Vikar V, Vasar E, Rauvala H (2004): Behavioral alterations induced byrepeated testing in C57BL/6J and 129S2/Sv mice: Implications forphenotyping screens. Genes Brain Behav 3:2738.

31. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD(2013): Development of a dual-index sequencing strategy and curationpipeline for analyzing amplicon sequence data on the MiSeq Illuminasequencing platform. Appl Environ Microbiol 79:51125120.

32. Edgar RC (2013): UPARSE: Highly accurate OTU sequences frommicrobial amplicon reads. Nat Methods 10:996998.

33. Pepping JK, Freeman LR, Gupta S, Keller JN, Bruce-Keller AJ (2013):NOX2 deciency attenuates markers of adiposopathy and brain injuryinduced by high-fat diet. Am J Physiol Endocrinol Metab 304:E392E404.

34. Njunge K, Handley SL (1991): Evaluation of marble-burying behavioras a model of anxiety. Pharmacol Biochem Behav 38:6367.

35. Everard A, Belzer C, Geurts L, Ouwerkerk JP, Druart C, Bindels LB,et al. (2013): Cross-talk between Akkermansia muciniphila andintestinal epithelium controls diet-induced obesity. Proc Natl AcadSci U S A 110:90669071.

36. Han S, Hong S, Lee D, Lee MH, Choi JS, Koh MJ, et al. (2012): Alteredexpression of synaptotagmin 13 mRNA in adult mouse brain aftercontextual fear conditioning. Biochem Biophys Res Commun 425:880885.

37. Stranahan AM, Norman ED, Lee K, Cutler RG, Telljohann RS, EganJM, Mattson MP (2008): Diet-induced insulin resistance impairshippocampal synaptic plasticity and cognition in middle-aged rats.Hippocampus 18:10851088.

38. Neufeld KM, Kang N, Bienenstock J, Foster JA (2011): Reducedanxiety-like behavior and central neurochemical change in germ-freemice. Neurogastroenterol Motil 23:255264.

39. Bercik P, Denou E, Collins J, Jackson W, Lu J, Jury J, et al. (2011):The intestinal microbiota affect central levels of brain-derived neuro-tropic factor and behavior in mice. Gastroenterology 141:599609.

40. Dinan TG, Cryan JF (2013): Melancholic microbes: A link between gutmicrobiota and depression? Neurogastroenterol Motil 25:713719.

41. Davey KJ, OMahony SM, Schellekens H, OSullivan O, BienenstockJ, Cotter PD, et al. (2012): Gender-dependent consequences ofchronic olanzapine in the rat: Effects on body weight, inammatory,

metabolic and microbiota parameters. Psychopharmacology (Berl)221:155169.

Biological42. Davey KJ, Cotter PD, OSullivan O, Crispie F, Dinan TG, Cryan JF,OMahony SM (2013): Antipsychotics and the gut microbiome:Olanzapine-induced metabolic dysfunction is attenuated by antibioticadministration in the rat. Transl Psychiatry 3:e309.

43. Ascher-Svanum H, Stensland M, Zhao Z, Kinon BJ (2005): Acuteweight gain, gender, and therapeutic response to antipsychotics in thetreatment of patients with schizophrenia. BMC Psychiatry 5:3.

44. Finger BC, Dinan TG, Cryan JF (2011): High-fat diet selectivelyprotects against the effects of chronic social stress in the mouse.Neuroscience 192:351360.

45. Montiel-Castro AJ, Gonzlez-Cervantes RM, Bravo-Ruiseco G,Pacheco-Lpez G (2013): The microbiota-gut-brain axis: Neurobeha-vioral correlates, health and sociality. Front Integr Neurosci 7:70.

46. Ochoa-Repraz J, Mielcarz DW, Begum-Haque S, Kasper LH (2011):Gut, bugs, and brain: Role of commensal bacteria in the control ofcentral nervous system disease. Ann Neurol 69:240247.

47. Lyte M (2013): Microbial endocrinology in the microbiome-gut-brainaxis: How bacterial production and utilization of neurochemicalsinuence behavior. PLoS Pathog 9:e1003726.

48. Bravo JA, Forsythe P, Chew MV, Escaravage E, Savignac HM, DinanTG, et al. (2011): Ingestion of Lactobacillus strain regulates emotionalbehavior and central GABA receptor expression in a mouse via thevagus nerve. Proc Natl Acad Sci U S A 108:1605016055.

49. Perez-Burgos A, Wang B, Mao YK, Mistry B, McVey Neufeld KA,Bienenstock J, Kunze W (2013): Psychoactive bacteria Lactobacillusrhamnosus (JB-1) elicits rapid frequency facilitation in vagal afferents.Am J Physiol Gastrointest Liver Physiol 304:G211G220.

50. Raison CL, Miller AH (2011): Is depression an inammatory disorder?Curr Psychiatry Rep 13:467475.

51. Lyman M, Lloyd DG, Ji X, Vizcaychipi MP (2014): Neuroinammation:The role and consequences. Neurosci Res 79C:112.

52. Duca FA, Sakar Y, Lepage P, Devime F, Langelier B, Dor J, CovasaM (2014): Replication of obesity and associated signaling pathwaysthrough transfer of microbiota from obese prone rat. Diabetes 63:16241636.

53. Hsiao EY, McBride SW, Hsien S, Sharon G, Hyde ER, McCue T, et al.(2013): Microbiota modulate behavioral and physiological ab-normalities associated with neurodevelopmental disorders. Cell 155:14511463.

54. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, GordonJI (2006): An obesity-associated gut microbiome with increasedcapacity for energy harvest. Nature 444:10271031.

55. Turnbaugh PJ, Bckhed F, Fulton L, Gordon JI (2008): Diet-inducedobesity is linked to marked but reversible alterations in the mousedistal gut microbiome. Cell Host Microbe 3:213223.

56. Yi P, Li L (2012): The germfree murine animal: An important animalmodel for research on the relationship between gut microbiota and thehost. Vet Microbiol 157:17.

57. Luckey TD (1965): Effects of microbes on germfree animals. Adv ApplMicrobiol 7:169223.

58. Berg RD (1996): The indigenous gastrointestinal microora. TrendsMicrobiol 4:430435.

59. Zhang H, DiBaise JK, Zuccolo A, Kudrna D, Braidotti M, Yu Y, et al.(2009): Human gut microbiota in obesity and after gastric bypass.Proc Natl Acad Sci U S A 106:23652370.

60. Sweeney TE, Morton JM (2013): The human gut microbiome: A reviewof the effect of obesity and surgically induced weight loss. JAMA Surg148:563569.

61. Murphy EF, Cotter PD, Hogan A, OSullivan O, Joyce A, Fouhy F, et al.(2013): Divergent metabolic outcomes arising from targeted manipu-lation of the gut microbiota in diet-induced obesity. Gut 62:220226.

62. Jia W, Whitehead RN, Grifths L, Dawson C, Bai H, Waring RH, et al.(2012): Diversity and distribution of sulphate-reducing bacteria inhuman faeces from healthy subjects and patients with inammatorybowel disease. FEMS Immunol Med Microbiol 65:5568.P63. Baron EJ (1997): Bilophila wadsworthia: A unique Gram-negativeanaerobic rod. Anaerobe 3:8386.

Psychiatry April 1, 2015; 77:607615 www.sobp.org/journal 615

Obese-type Gut Microbiota Induce Neurobehavioral Changes in the Absence of ObesityMethods and MaterialsAnimals and TreatmentsBehavioral Testing16S Metagenomic SequencingPlasma and Tissue AnalysesStatistical Analyses

ResultsHFD-Derived Gut Microbiota Impair Behavioral Performance in MiceMicrobiota Transplantation Results in Distinct Phylogenetic ProfilesHFD-Derived Gut Microbiota Increase Intestinal Permeability, Systemic Inflammation, and Brain Inflammation

DiscussionAcknowledgments and DisclosuresArticle InformationReferences