Brain, Behavior, and Immunity 60 (2017) 51–62

Contents lists available at ScienceDirect

Brain, Behavior, and Immunity

journal homepage: www.elsevier .com/locate /ybrbi

Sex-specific modulation of the gut microbiome and behavior in Siberianhamsters

http://dx.doi.org/10.1016/j.bbi.2016.10.0230889-1591/� 2016 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: Department of Biology, Indiana University, 1001 E.3rd Street, Bloomington, IN 47405, USA.

E-mail address: ksylvia@indiana.edu (K.E. Sylvia).

Kristyn E. Sylvia ⇑, Cathleen P. Jewell, Nikki M. Rendon, Emma A. St. John, Gregory E. DemasDepartment of Biology and Center for the Integrative Study of Animal Behavior, Indiana University, Bloomington, IN 47405, USA

a r t i c l e i n f o

Article history:Received 29 June 2016Received in revised form 28 September2016Accepted 16 October 2016Available online 2 November 2016

Keywords:AggressionBlood-brain barrierEnrofloxacinGut-brain axisImmune systemInvestigationScent-marking

a b s t r a c t

The gut microbiome is a diverse, host-specific, and symbiotic bacterial environment that is critical formammalian survival and exerts a surprising yet powerful influence on brain and behavior. Gut dysbiosishas been linked to a wide range of physical and psychological disorders, including autism spectrum dis-orders and anxiety, as well as autoimmune and inflammatory disorders. A wealth of information on theeffects of dysbiosis on anxiety and depression has been reported in laboratory model systems (e.g., germ-free mice); however, the effects of microbiome disruption on social behaviors (e.g., aggression) of non-model species that may be particularly important in understanding many aspects of physiology andbehavior have yet to be fully explored. Here we assessed the sex-specific effects of a broad-spectrumantibiotic on the gut microbiome and its effects on social behaviors in male and female Siberian hamsters(Phodopus sungorus). In Experiment 1, we administered a broad-spectrum antibiotic on a short-term basisand found that antibiotic treatment altered the microbial communities in the gut in male and femalehamsters. In Experiment 2, we tested the effects of single versus repeated antibiotic treatment (includinga recovery phase) on behavior, and found that two, but not one, treatments caused marked decreases inaggressive behavior, but not other social behaviors, in males; aggression returned to normal levels fol-lowing recovery. Antibiotic-treated females, in contrast, showed decreased aggression after a single treat-ment, with all other social behaviors unaffected. Unlike males, female aggression did not return to normalduring either recovery period. The present findings demonstrate that modest antibiotic treatment resultsin marked disruption of the gut microbiome in hamsters, akin to research done in other rodent speciesand humans. Further, we show that treatment with a broad-spectrum antibiotic, which has dysbioticeffects, also has robust, sex-specific effects on aggression, a critical behavior in the survival and reproduc-tive success of many rodent species.

� 2016 Elsevier Inc. All rights reserved.

1. Introduction

The large intestine of the gastrointestinal (GI) tract containsabout 100 trillion microorganisms, an amount ten times greaterthan the total number of cells in the body (Wallace et al., 2011).Though much of our attention was not focused on these microor-ganisms until recently. The gut microbiome is not only a diverse,host unique, and symbiotic bacterial environment, but it is criticalfor mammalian survival (Clarke et al., 2014), and further, exerts asurprisingly powerful influence on brain and behavior.

Information from the gut can communicate with the centralnervous system (CNS) and thus microbes living in the gut can influ-ence memory, emotions, and affective behaviors in many species

(Dinan and Cryan, 2012). For example, dysbiosis has been linkedto various physical and psychological disorders, including autismspectrum disorders (ASD) and anxiety, as well as autoimmuneand inflammatory disorders, such as allergies and asthma (Clarkeet al., 2014; Mueller et al., 2015). Individuals of all species musthave the ability to express the appropriate types of behaviors incontext. Therefore studying how disruption of the microbiomemight affect these behaviors is important to understand the fitnessof an individual across a range of species, including humans, andwith such insights we can begin to connect ecological and transla-tional research.

A wealth of information has been reported in model systems(e.g., germ-free mice) (Clarke et al., 2014; Collins et al., 2012;O’Mahony et al., 2009); however, the effects of microbiome disrup-tion on social behaviors particularly important in the success ofnon-model species (e.g., aggression) have yet to be fully explored.In studying these model systems, researchers are only partially

52 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

able to determine the role that the microbiota play in the naturalfunction of the mammalian system. Although they have therapeu-tic benefits, antibiotics can alter the structure, function, and ulti-mately evolution of host microbial communities in the gut(Archie and Theis, 2011). Therefore antibiotics are a useful tool tomanipulate the gut microbiome, and further, doing so in a non-model species provides insight into the potential cause of themicrobiota-dependent changes we see in physiology and behaviorand will help complement the important strides the field has takenin understanding the gut-brain axis thus far.

Recent work in BALB/c mice has shown that disruption of themicrobiota via antibiotics increases exploration, a non-socialbehavior, yet the same change in behavior is not seen in germ-free mice, unless the germ-free mice are colonized with themicrobiota from other strains (Bercik et al., 2011). Because mostsocial behaviors are essential for the maintenance of appropriateinteractions with conspecifics, both in reproductive and non-reproductive contexts, understanding the ways in which they areaffected are important to our understanding of behavior andphysiology. The influence of the gut microbiome on social behaviorhas become of recent interest, yet the effects on aggression, animportant behavior for a number of rodent species and humansalike, has not yet been determined. Aggressive behavior variesgreatly across species, however, a number of indicators can be usedto identify it. Often, an aggressive act involves two or moreindividuals competing for resources (e.g., food, mates, territory),and these acts of aggression may result in severe injury, or evendeath (Gould and Zeigler, 2007; Soma et al., 2015). Aggressivebehaviors include attacking and wrestling, chasing and biting(Jasnow et al., 2002; Nelson et al., 1995), as well as pushing andjump-fighting. Many of these aggressive behaviors are importantin signaling reproductive condition to potential mates (Gould andZeigler, 2007).

To provide information about identity, sex, and reproductivestate, which are important for survival, many rodents, includingmale and female hamsters, have a sexually dimorphic gland onthe midline of their ventral surface that they use for scent-marking, in which they rub their ventral gland on a protruding sur-face (Johnston, 1993; Lai and Johnston, 1994; Reasner andJohnston, 1987; Rendon et al., 2016b). These types of behaviorsare essential for recognition of conspecifics, and are particularlyimportant in a social context. When animals are first introduced,they will investigate one another, via nose-to-nose and nose-to-anogenital sniffing, in order to use the chemical signals to identifysex, age, reproductive status, quality, and other characteristics oftheir conspecifics (Pellis and Pellis, 1988; Rendon et al., 2016a,b).Because investigative behaviors can influence how often othersocial behaviors (e.g., exploration, aggression, reproduction) mayoccur, they are critical to our understanding of the behavioral phe-notype (Wynne-Edwards and Lisk, 1987; Wynne-Edwards, 2003).Investigating how the microbiome affects these important behav-iors may help us to determine a more thorough understanding ofpsychiatric disorders often associated with unusual social interac-tions, such as autism or schizophrenia (Scattoni et al., 2011).

The goal of the present study was to examine if thebroad-spectrum antibiotic, enrofloxacin, can be used as a tool tomanipulate the gut microbiome and whether this tool also hasthe potential to affect social behaviors in male and female Siberianhamsters. We hypothesized that a 7-day antibiotic treatmentwould be sufficient to alter the microbial composition of the gutas well as to initiate changes in both aggressive and investigativebehaviors, therefore altering aspects of the adult behavioralphenotype. The results of these studies will help to further ourunderstanding of the influence of the gut microbiome on physiol-ogy and behavior, and potential sex differences seen in responseto disruption of the gut microbiome.

2. Materials and methods

2.1. Experiment 1: Effect of antibiotics on the gut microbiome

2.1.1. Animal housing conditionsMale and female adult (>60 days of age) hamsters were reared

and maintained under long days (light:dark, 16:8 h), and individu-ally housed in polypropylene cages (28 � 17 � 12 cm). Males andfemales were run together in cohorts in the same animal holdingroom. All procedures were the same for all cohorts. Ambient tem-perature was maintained at 20 ± 2 �C, and relative humidity wasmaintained at 55 ± 5%. Hamsters were given ad libitum access topurified tap water and standard laboratory rodent chow (Lab Diet5001, PMI Nutrition). All procedures were performed in accordancewith the National Institutes of Health Guide for the Care and Use ofLaboratory Animals and were approved by the Bloomington Insti-tutional Animal Care and Use Committee at Indiana University.

2.1.2. Antibiotic treatmentTo determine how modest antibiotic treatment affects the

microbial communities in the gut and potential sex differences inthe gut microbiome in Siberian hamsters, 12 males and 12 femaleswere assigned to either a control group (n = 6 males, n = 6 females),in which animals received sterilized water administered via sterilepipette orally once daily, or an experimental group (n = 6 males,n = 6 females), in which animals received a broad-spectrum antibi-otic [Abx: 0.3 ll of enrofloxacin (Baytril, Bayer Animal Health) 10%oral solution per gram of body mass] administered via sterile pip-ette orally once daily (Romick-Rosendale et al., 2009). Enrofloxacinis a fluoroquinolone antimicrobial agent, frequently used for treat-ment in many domesticated animals and does not easily cross theblood brain barrier (BBB) (Alvarez et al., 2010; Ooie et al., 1997a,b;Slate et al., 2014). Enrofloxacin inhibits DNA synthesis and hasbeen given orally to hamsters numerous times at varying dosesand has been documented as safe and effective in our species(Martorell et al., 2010; Romick-Rosendale et al., 2009; Slate et al.,2014; Thomas et al., 2008). On days 1–7 (D1-7) of experimentation(Pre-Treatment), all animals were monitored and weighed regu-larly. During days 8–14 (Treatment), Abx animals received Abxtreatment once daily and control animals received sterilized wateronce daily, during which animals were monitored and weighedregularly. During days 14–21 (Post-Treatment), both the experi-mental group and the control group were monitored and weighedregularly.

2.1.3. Fecal samplingFollowing the Pre-Treatment, Abx/Control Treatment, and the

Post-Treatment Recovery period, effects on the gut microbiomewere assessed by taking fecal samples from each animal. To takefecal samples, animals were removed from their home cage andheld over a sterile container once daily, after which the fecal sam-ples were stored in �80 �C until the samples were processed. Allanimals were returned to their home cage until the followingday. To determine whether restraint stress during fecal samplesaffected the gut microbiome, fecal samples were also taken froma subset of males and females (n = 3 males and n = 3 females) thatreceived neither water (control) treatment nor antibiotic (experi-mental) treatment.

2.1.4. Microbiome analysisFecal samples were sequenced (IDEXX Bioresearch, Columbia,

MO) to determine microbial composition in the gut (n = 9 males;n = 9 females). DNA was extracted from the fecal material and puri-fied over a DNeasy spin column, and the extracted DNA was quan-tified via fluorometry for normalization when plating. Using the

K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62 53

purified DNA, a previously well-validated U515F/806R primer setwas used to amplify a portion of the V4 region of the 16S rRNAgene using the Illumina MiSeq platform yielding a 300 base pairfragment from all bacterial species present in the sample(Ericsson et al., 2015). From each, 200–500 megabases of sequenceinformation were obtained. Sequences were determined usingQIIME (Quantitative Insights Into Microbial Ecology) v1.8 Pipeline,and once sequences were determined, they were BLASTed (BasicLocal Alignment Search Tool) against the Greengenes database(Second Genome, Inc.: University of Colorado, University ofQueensland) to identify operational taxonomic units (OTUs) andtaxonomic classification. We found a mean of 32,335 sequencesper sample. We conducted a principal component analyses (PCA)on the 14 bacterial phyla that were present in fecal samples forboth sexes to examine the effects of antibiotic treatment on micro-bial communities across groups.

2.1.5. Tissue collectionAt the end of the experiment (D21), all animals were euthanized

via a lethal intraperitoneal (i.p.) injection of a ketamine and xyla-zine cocktail in 0.9% saline. Following euthanasia, livers, spleens,and reproductive organs were dissected and weighed to determinethe effects of antibiotic treatment on gross anatomy.

2.2. Experiment 2: Effects of repeated antibiotic treatment on themicrobiome and social behavior

2.2.1. Animal housing conditionsMale and female adult (>60 days of age) hamsters were reared

and maintained under long days (light:dark, 16:8 h), and individu-ally housed in polypropylene cages (28 � 17 � 12 cm). Males andfemales were run together in cohorts in the same animal holdingroom. All procedures were the same for all cohorts. Ambient tem-perature was maintained at 20 ± 2 �C, and relative humidity wasmaintained at 55 ± 5%. Hamsters were given ad libitum access topurified tap water and standard laboratory rodent chow (Lab Diet5001, PMI Nutrition).

2.2.2. Antibiotic treatmentWe housed 18 males and 18 females, and assigned them to

either a control group (n = 9 males; n = 9 females), in which ani-mals received sterilized water administered via sterile pipetteorally once daily for two treatment courses, or an experimentalgroup (n = 9 males; n = 9 females), in which animals received abroad-spectrum antibiotic [Abx: 0.3 ll of enrofloxacin per gramof body mass administered via sterile pipette orally once daily]for two treatment courses (Romick-Rosendale et al., 2009). On days1–7 (D1-7) of experimentation (First Treatment Period), Abx ani-mals received Abx treatment and control animals received steril-ized water. On days 8–14 (First Recovery Period), all animalswere monitored and weighed. Again on days 15–21 (Second Treat-ment Period) Abx animals received Abx treatment and control ani-mals received sterilized water. Finally, on days 22–28 (SecondRecovery Period), all animals were monitored and weighed.

2.2.3. Fecal samplingFollowing Treatment Period 2, effects on the gut microbiome

were assessed by taking fecal samples from each animal. To collectfecal samples, we used the same protocol described in Experiment1. All fecal samples were stored in �80 �C until the samples wereprocessed.

2.2.4. Microbiome analysisFecal samples collected at the end of Treatment Period 2 were

sequenced (IDEXX Bioresearch, Columbia, MO) to determinemicrobial composition in the gut (n = 6 males, n = 6 females).

DNA was extracted from the fecal material, purified, and quantifiedaccording to the previously described protocol. The DNA was thenused to amplify a portion of the V4 region of the 16S rRNA geneusing the Illumina MiSeq platform yielding a 300 base pair frag-ment from all bacterial species present in the sample (Ericssonet al., 2015). From each, 200–500 megabases of sequence informa-tion were obtained. Sequences were determined using QIIME v1.8Pipeline and BLASTed against the Greengenes database to identifyoperational taxonomic units (OTUs) and taxonomic classification.We found a mean of 81,313 sequences per sample.

2.2.5. Behavioral interactions and analysesWe recorded and analyzed aggressive behavior, investigative

behavior, scent-marking behavior, and grooming using theresident-intruder paradigm with same-sex social partners per pre-viously outlined methods for this species (Jasnow et al., 2000;Rendon et al., 2016b, 2015). On D7 (Treatment 1), D14 (Recovery 1),D21 (Treatment 2) and D28 (Recovery 2), behavioral testing wasconducted during the first 2 h of the dark phase to control for cir-cadian rhythmicity of behavior. We staged five-minute dyads suchthat they were composed of a resident hamster (n = 18 males andn = 18 females) and a same-sex intruder hamster (n = 9 males andn = 9 females). We paired animals of the same reproductive status,of approximately the same age, with comparable mass (±5%), andfrom different parents. All trials were video recorded (SonyHandyCam Digital Camcorder HDR-SR7) under low-illuminationred lights.

We quantified aggressive, investigative, scent-marking, andgrooming behaviors of the resident hamster using ODLogTM (Macro-pod Software, Eden Prairie, MN). For aggression, we quantifiedlatency to first attack (s), and frequency and duration of attacksand chases (s). For investigation, we quantified the frequency andduration of nose-to-anogenital and nose-to-nose investigation (s).For scent-marking, we quantified the frequency and duration ofscent depositing (s), and for grooming, we quantified the frequencyand duration of the resident animal grooming him or herself (s).We used four separate principal component analyses (PCA) on allaggression variables (69.63–88.28% of the total variance explained,Table 1 in Supplementary material), investigative variables (84.61–95.23% of the total variance explained, Table 2 in Supplementarymaterial), scent-marking variables (96.15–99.98% of the total vari-ance explained, Table 3 in Supplementary material) and groomingvariables (79.70–96.28% of the total variance explained, Table 4 inSupplementary material) for both sexes. All behaviors indepen-dently loaded strongly onto the first component; therefore thecomposite aggression score (PC-AGG), composite investigationscore (PC-INV), composite scent-marking score (PC-SMK) and com-posite grooming score (PC-GRM) were used to examine the effectsof antibiotic treatment on changes in social behavior of residenthamsters.

2.2.6. Tissue collectionAt the end of the experiment (D28), all animals were euthanized

via a lethal i.p. injection of a ketamine and xylazine cocktail in 0.9%saline. Following euthanasia, livers, spleens, reproductive organsand reproductive fat tissue were dissected and weighed to deter-mine the effects of antibiotics on gross anatomy.

3. Statistical analyses

We performed all statistical analyses on microbiome data in R v.3.2.2 (R Core Team 2015), and attributed statistical significance atp < 0.05. Principal component analyses (PCA) were conducted onmicrobiome data at the phylum level in Experiment 1 (Table 5 inSupplementary material). Two-way analyses of variance (ANOVAs)

54 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

were used to compare the effects of treatment (control and antibi-otics) and time (Pre-Treatment, Treatment, and Post-Treatment)across groups in Experiment 1. If a two-way ANOVA reported a sig-nificant interaction of treatment and time, Tukey’s Honest Signifi-cant Difference post hoc tests were run to determine pair-wiserelationships. To determine the diversity across groups, we calcu-lated the Shannon-Wiener index and converted values to the effec-tive number of species (Hill, 1973; Jost, 2006), and we alsocalculated Bray-Curtis dissimilarity scores across groups and timepoints. In Experiment 2, two-tailed t-tests were used to comparethe relative abundance of each phylum across treatment groupsfollowing Treatment Period 2.

All behavioral analyses were performed in JMP v. 11.0.0 (SASInstitute, Inc., Cary, NC), and attributed statistical significance atp < 0.05. Data were log or square root transformed to attain nor-mality and equal variances. PCAs were conducted on the socialbehaviors (Tables 1–4 in Supplementary material) to reduce thenumber of variables for analysis, retaining PCs with an eigenvaluegreater than 1. Two-tailed t-tests were used to compare physiolog-ical and behavioral changes across treatment groups. Mixed modelANOVAs were calculated with treatment (control and antibiotics)as a between-groups variable and time (Treatment 1, Recovery 1,Treatment 2, Recovery 2) as a within-subjects variable. If a mixedmodel ANOVA reported a significant effect or interaction of effectspair-wise comparisons were conducted using two-tailed t-tests.Spearman’s rank correlations were run on the relative abundanceof each phylum and number of attacks following Treatment Period2 in both sexes. Analysis on behavioral data was verified using thefalse discovery rate, however, all p-values reported here are rawp-values to maintain consistency.

4. Results

4.1. Experiment 1

4.1.1. Siberian hamsters have a diverse gut microbiomeThe microbial community composition in fecal samples of male

and female hamsters with and without antibiotics is shown in

Fig. 1. The microbial community composition in fecal samples of male and female hmicrobiome is made up of 14 phyla, including one phylum from the Archaea kingdomsequences. Both the male and female gut microbiome are dominated by Firmicutes andand instead most strongly affects phyla in lower abundance in both male and female ha

Fig. 1. The composition of the male gut microbiome was dominatedby Firmicutes (�50%) and Bacteroidetes (�40%). In males, 95unique operational taxonomic units (OTU) were identified from14 phyla, including one phylum from the Archaea kingdom and adesignated ‘Other’ category when there was no specific BLAST hitfor the sequences. Thirteen OTUs were eliminated from this count,because they were only represented in 1 sample.

The composition of the female gut microbiome was dominatedby Firmicutes (�50%) and Bacteroidetes (�30%). In females, 98unique operational taxonomic units were identified from 14 phyla,including one phylum from the Archaea kingdom and a designated‘Other’ category when there was no specific BLAST hit for thesequences. Thirteen OTUs were eliminated from this count,because they were only represented in 1 sample.

4.1.2. Antibiotics differentially affect the gut microbiota across thesexes

Three PCs were extracted for each sex. For males, 92.23% of thetotal variance was explained by PCs 1–3 (Table 5 in Supplementarymaterial). Baseline measures were scattered for PCs 1 and 2 amonggroups, but clustered for PC 3, indicating a variable microbial com-position for males. In contrast, control and antibiotic groups wereclearly separated in opposite directions during both the treatmentand recovery periods, which indicates that antibiotic treatmentalters microbial diversity in the gut (Fig. 2a&b). In males, treatmentaffected 5 phyla: Cyanobacteria (F1,12 = 4.91, p = 0.047), Elusimicro-bia (F1,12 = 6.24, p = 0.03), Euryarchaeota (F1,12 = 5.89, p = 0.03),Proteobacteria (F1,12 = 12.23, p = 0.004), and TM7 (F1,12 = 4.85,p = 0.048). Time affected Proteobacteria (F2,12 = 4.90, p = 0.03),and TM7 (F2,12 = 4.57, p = 0.03). There was an interaction betweentreatment and time for Tenericutes (F2,12 = 4.74, p = 0.03), andTM7 (F2,12 = 5.51, p = 0.02) (Fig. 3a-f). There was no effect of treat-ment or time or their interaction for 8 phyla: ‘Other’, Actinobacte-ria, Bacteroidetes, Deferribacteres, Firmicutes, Fusobacteria,Spirochaetes, and Verrucomicrobia (p > 0.05 in all cases).

For females, 92.28% of the total variance was explained by PCs1–3 (Table 5 in Supplementary material). Baseline measures wereclustered for all 3 PCs among groups, indicating a homogenous

amsters across treatment groups. The composition of the male and female gutand a designated ‘Other’ category where no specific BLAST hit was found for the

Bacteroidetes, yet antibiotic treatment does not affect these highly abundant phyla,msters.

Fig. 2. Principal components analysis (PCA) plots representing PC1-3. In males, baseline measures were scattered for PCs 1 and 2 (a), but clustered for PC 3 (b), indicating avariable microbial composition. In females, baseline measures were clustered for all 3 PCs among groups, indicating a homogenous microbial composition for females (c, d). Inboth males and females, control and antibiotic groups were separated in opposite directions during treatment and recovery, which indicates that antibiotic treatment altersmicrobial composition of the gut.

K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62 55

microbial composition for females. In contrast, control andantibiotic groups were clearly separated in opposite directionsduring both the treatment and recovery periods, indicating thatantibiotic treatment alters the microbial composition of the gut(Fig. 2c&d). In females, treatment affected 3 phyla: Cyanobacteria(F1,12 = 18.88, p = 0.001), Proteobacteria (F1,12 = 11.01, p = 0.006),and Tenericutes (F1,12 = 6.67, p = 0.02). Time affected Proteobacte-ria (F2,12 = 6.99, p = 0.01). There was an interaction between treat-ment and time for Proteobacteria (F2,12 = 4.55, p = 0.03) (Fig. 3g-i).There was no effect of treatment, time, or their interaction for 11phyla: ‘Other’, Actinobacteria, Bacteroidetes, Deferribacteres,Elusimicrobia, Euryarchaeota, Firmicutes, Fusobacteria, Spiro-chaetes, TM7, and Verrucomicrobia (p > 0.05 in all cases).

Analysis of the No-Treatment group validated that we did notintroduce any new, unidentifiable bacteria into the antibiotic or

control groups, therefore analysis and results were completed onthe antibiotic- and water-treated groups only.

4.1.3. Antibiotics affect the diversity of the gut microbiomeIn males, the effective number of species before treatment was

11.653 in control animals and 12.290 in Abx animals. Immediatelyfollowing the treatment period, the effective number of specieswas 10.455 in control animals, and in Abx animals, the effectivenumber of species decreased to 5.683, approximately 46% lessdiverse than the control group at this time point. Following therecovery period, the effective number of species was 8.840 in con-trol animals, and in Abx animals, the value increased slightly, to6.531 (Table 6 in Supplementary material). Similarly, in females,the effective number of species before treatment was 10.583 incontrol animals and 9.988 in Abx animals. Immediately following

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Cya

noba

cter

ia

0.000

0.001

0.002

0.003

0.004

0.005

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Pro

teob

acte

ria

0.00

0.02

0.04

0.06

0.08

0.10

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Ten

eric

utes

0.00

0.01

0.02

0.03

0.04

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Elu

sim

icro

bia

0e+00

1e−04

2e−04

3e−04

4e−04

5e−04

6e−04

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Eur

yarc

haeo

ta

0e+00

1e−04

2e−04

3e−04

4e−04

5e−04

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

TM

7

0.00

0.01

0.02

0.03

0.04W

pre

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Cya

noba

cter

ia

0.000

0.001

0.002

0.003

0.004

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Pro

teob

acte

ria0.00

0.02

0.04

0.06

0.08

Wpr

e

Ww

ater

Wpo

st

Xpre

Xabx

Xpos

t

Prop

ortio

n of

Ten

eric

utes

0.00

0.01

0.02

0.03

0.04

(a) (d) (g)

(b) (e) (h)

(c) (f) (i)

aa

a

a a a

a

aa

b

b

b

b

b

b

b

bb

Males Females

Fig. 3. Effects of antibiotics on the relative abundance of bacterial phyla in male (a-f) and female (g-i) hamsters. Although Bacteroidetes and Firmicutes are in the highestrelative abundance in male and female hamster fecal samples, antibiotics did not affect these phyla, and instead those phyla in low relative abundance are most stronglyaffected by treatment in both sexes. In males and females, there was an effect between treatment, time, or their interaction for 3 phyla: Cyanobacteria (a,g), Proteobacteria (b,h), and Tenericutes (c,i). Male microbiome analysis showed an effect between treatment, time, or their interaction for 3 additional phyla: Elusimicrobia (d), Euryarchaeota (e),and TM7 (f). There was no effect of treatment or time or their interaction for 8 phyla in males: ‘Other’, Actinobacteria, Bacteroidetes, Deferribacteres, Firmicutes, Fusobacteria,Spirochaetes, and Verrucomicrobia, and there was no effect of treatment or time or their interaction for 11 phyla in females: ‘Other’, Actinobacteria, Bacteroidetes,Deferribacteres, Elusimicrobia, Euryarchaeota, Firmicutes, Fusobacteria, Spirochaetes, TM7, and Verrucomicrobia. Bar heights represent means ± S.E.M. Means with differentletters are statistically different (p < 0.05).

56 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

the treatment period, the effective number of species was 11.680in control animals, and in Abx animals, the effective numberdecreased to 6.620, approximately 44% less diverse than the con-trol group at this time point. Following the recovery period, theeffective number of species was 12.317 in control animals, and inAbx animals, there was small increase, to 7.547 (Table 7 in Supple-mentary material). In both male and female antibiotic groups,

there was a decrease in alpha diversity following the treatmentperiod, and the same decrease was not seen in control groups.

Further, in males across groups, the pre-treatment Bray-Curtisdissimilarity score was 0.322, the treatment period Bray-Curtis dis-similarity score was 0.374, and the post-treatment Bray-Curtis dis-similarity score 0.190 (Table 8 in Supplementary material). Infemales across groups, the pre-treatment Bray-Curtis dissimilarity

K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62 57

score was 0.111, the treatment period Bray-Curtis dissimilarityscore was 0.277, and the post-treatment Bray-Curtis dissimilarityscore 0.206 (Table 9 in Supplementary material). In both malesand females, the greatest dissimilarity was observed during thetreatment period.

4.1.4. Antibiotics did not affect body mass or organ massAntibiotics did not affect male body mass (F1,14 = 1.09, p = 0.91),

liver mass (F1,14 = 0.91, p = 0.91), or spleen mass (F1,14 = 0.67,p = 0.53), when compared with control males. Similarly, antibioticsdid not affect female body mass (F1,13 = 1.93, p = 0.42), liver mass(F1,13 = 1.31, p = 0.74), or spleen mass (F1,13 = 2.59, p = 0.12), whencompared with control females (Table 10 in Supplementarymaterial).

4.2. Experiment 2

4.2.1. Repeated antibiotics affect the gut microbiotaSimilar to Experiment 1, we assessed the effects of antibiotics

on the male and female microbiome. In males, treatment signifi-cantly affected the relative abundance of Tenericutes (t2 = -5.539,p = 0.031), and it moderately affected Cyanobacteria(t2 = �3.5572, p = 0.071) and Proteobacteria (t2.441 = �3.0512,p = 0.072) (Fig. 4a and Fig. 1 in Supplementary material). Therewas no effect of treatment for 11 phyla: ‘Other’, Actinobacteria,Bacteroidetes, Deferribacteres, Elusimicrobia, Euryarchaeota, Fir-micutes, Fusobacteria, Spirochaetes, TM7, and Verrucomicrobia(p > 0.05 in all cases). In females, treatment affected TM7(t3.620 = -4.102, p = 0.018) and Tenericutes (t2 = -4.092, p = 0.055),and it moderately affected Cyanobacteria (t2 = -3.001, p = 0.095)(Fig. 4b and Fig. 1 in Supplementary material). There was no effectof treatment for 11 phyla: ‘Other’, Actinobacteria, Bacteroidetes,Deferribacteres, Elusimicrobia, Euryarchaeota, Firmicutes,Fusobacteria, Proteobacteria, Spirochaetes, and Verrucomicrobia(p > 0.05 in all cases).

4.2.2. Repeated antibiotics affect the diversity of the gut microbiomeFollowing the second treatment period in males, the effective

number of species was 8.691 in control animals and 3.950 in Abxanimals, approximately 55% less diverse than the control group;

Water

Prop

ortio

n of

Ten

eric

utes

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

Water Abx

Prop

ortio

n of

Ten

eric

utes

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007 (b)(a)

*

Fig. 4. Effect of antibiotics on relative abundance of Tenericutes in male and female hamExperiment 2, in males (a) and females (b), treatment affected the relative abundanceheights represent means ± S.E.M. An asterisk (*) indicates statistically significant differerelative abundance of Tenericutes and the number of attacks in male and female hamstersof Tenericutes in male and female hamsters showed trends towards significant associat

in females, the effective number of species was 8.601 in controlanimals and 5.011 in Abx animals, approximately 42% less diversethan in the control group (Table 11 in Supplementary material). Inboth male and female antibiotic groups, alpha diversity was lowerfollowing the second treatment period, when compared to controlgroups.

In males, the Bray-Curtis dissimilarity score was 0.267 follow-ing treatment period 2, and in females, the Bray-Curtis dissimilar-ity score was 0.156 (Table 12 in Supplementary material).

4.2.3. Antibiotics affect aggressive behavior differentially across thesexes

Male hamsters given a single, 7-day antibiotic treatment dis-played no change in overall aggression (PC-AGG, F1,14 = 4.80,p = 0.79) when compared with males given a one-time, 7-day con-trol treatment (Fig. 5c). Specifically, males given a one-time week-long antibiotic treatment displayed no change in number(F1,14 = 4.20, p = 0.66, Fig. 5a) or duration (F1,14 = 4.26, p = 0.62) ofattacks and no change in number (F1,14 = 4.15, p = 0.70) or duration(F1,14 = 3.62, p = 0.81) of chases. In addition, latency to first attackdid not differ across groups (F1,14 = 2.64, p = 0.44). In contrast tomales, female hamsters given a single 7-day antibiotic treatmentdisplayed a decrease in overall aggression (PC-AGG, F1,16 = 7.80,p = 0.01) when compared with females given a single 7-day controltreatment (Fig. 5d). Specifically, females given a one-time week-long antibiotic treatment displayed fewer (F1,16 = 7.48, p = 0.01,Fig. 5b) and shorter attacks (F1,16 = 7.04, p = 0.02), as well as fewer(F1,16 = 6.41, p = 0.02) and shorter (F1,16 = 4.23, p = 0.04) chases. Inaddition, females given 7 days of antibiotic treatment exhibited alonger latency to the first attack (F1,16 = 8.16, p = 0.01).

Male hamsters given two 7-day treatments of antibiotics sepa-rated by a 7-day recovery period displayed a decrease in overallaggression (PC-AGG, F1,14 = 12.78, p = 0.003) when compared withmales given two control treatments following the same timeline(Fig. 5c). Specifically, males given two week-long antibiotic treat-ments displayed decreases in number (F1,14 = 7.18, p = 0.02,Fig. 5a) and duration (F1,14 = 2.44, p = 0.03) of attacks and decreasesin number (F1,14 = 10.94, p = 0.005) and duration (F1,14 = 22.92,p = 0.0003) of chases. However, latency to first attack did not differacross groups (F1,14 = 12.49, p = 0.49). Similar to males, female

Abx

(c) ρ = 0.764p = 0.007

*

sters following treatment period 2. Similar to what was found in Experiment 1, inof Tenericutes (males: t2 = �5.539, p = 0.0310; females: t2 = �4.092, p = 0.055). Barnces between group means at p < 0.05. There was also an association between theat the end of treatment period 2 (c). When separated by sex, the relative abundance

ions with number of attacks (see Table 13 in Supplementary material).

Fig. 5. Effects of antibiotics on aggressive behaviors across the sexes. Antibiotics affect frequency of attacks in male (a) and female (b) hamsters and overall aggression scoresin male (c) and female (d) hamsters. Bar heights represent means ± S.E.M. An asterisk (*) indicates statistically significant differences between group means at p < 0.05.

58 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

hamsters given two week-long antibiotic treatments displayed adecrease in overall aggression (PC-AGG, F1,16 = 24.35, p = 0.0001)when compared with females given two week-long control treat-ments (Fig. 5d). Specifically, females given two week-long antibi-otic treatments displayed fewer (F1,16 = 22.73, p = 0.0002, Fig. 5b)and shorter (F1,16 = 20.14, p = 0.0004) attacks and fewer(F1,16 = 24.63, p = 0.0001) and shorter (F1,16 = 22.24, p = 0.0002)chases. In addition, the latency to the first attack was longer(F1,16 = 6.38, p = 0.02) in repeated antibiotic-treated females.

4.2.4. Aggression is associated with relative abundance of bacteria inthe gut

The number of attacks in male and female hamsters followingtreatment period 2 was associated with the relative abundanceof Tenericutes (rs = 0.764, n = 12, p = 0.007) (Fig. 4c), Cyanobacteria(rs = 0.717, n = 12, p = 0.014), and TM7 (rs = 0.667, n = 12,p = 0.046). In contrast, the number of attacks was not associatedwith the 11 other phyla: ‘Other’, Actinobacteria, Bacteroidetes,Deferribacteres, Elusimicrobia, Euryarchaeota, Firmicutes,Fusobacteria, Proteobacteria, Spirochaetes, and Verrucomicrobia(p > 0.05 in all cases). For complete list of correlations and for sep-arate correlations by sex, see Table 13 in Supplementary material.

4.2.5. Antibiotics did not affect investigative behaviorMale hamsters given antibiotic treatment displayed no change

in overall investigative behavior (PC-INV, one-time: F1,14 = 2.17,p = 0.69; PC-INV, two-time: F1,14 = 2.04, p = 0.18) when comparedwith males given control treatment (Fig. 6a). Specifically, malesgiven antibiotic treatment displayed no change in number (one-time: F1,14 = 4.70, p = 0.80; two-time: F1,14 = 4.38, p = 0.85) or dura-tion (one-time: F1,14 = 4.49, p = 0.50; two-time: F1,14 = 4.97,p = 0.92) of nose-to-anogenital investigation and no change innumber (one-time: F1,14 = 4.37, p = 0.85; two-time: F1,14 = 2.40,p = 0.14) or duration (one-time: F1,14 = 4.21, p = 0.65; two-time:F1,14 = 3.20, p = 0.10) of nose-to-nose investigation (Table 1). Simi-lar to males, female hamsters given antibiotic treatment displayedno change in overall investigative behavior (PC-INV, one-time:F1,16 = 6.99, p = 0.42; PC-INV, two-time: F1,16 = 1.73, p = 0.21) whencompared with females given control treatment (Fig. 6b). Specifi-cally, females given antibiotic treatment displayed no change in

number (one-time: F1,16 = 1.53, p = 0.70; two-time: F1,16 = 2.46,p = 0.88) or duration (one-time: F1,16 = 4.94, p = 0.49; two-time:F1,16 = 7.90, p = 0.93) of nose-to-anogenital investigation, andfemales treated with antibiotics exhibited no change in number(one-time: F1,16 = 1.61, p = 0.69; two-time: F1,16 = 2.39, p = 0.14)or duration (one-time: F1,16 = 1.23, p = 0.28; two-time:F1,16 = 3.38, p = 0.08) of nose-to-nose investigation (Table 2).

4.2.6. Antibiotics did not affect scent-marking behaviorMale hamsters given antibiotic treatment displayed no change

in overall scent-marking behavior (PC-SMK, one-time:F1,14 = 2.44, p = 0.14; PC-SMK, two-time: F1,14 = 2.35, p = 0.15)when compared with males given control treatment. Specifically,antibiotic-treated males displayed no change in number (one-time: F1,14 = 1.81, p = 0.20; two-time: F1,14 = 2.46, p = 0.14) or dura-tion (one-time: F1,14 = 2.04, p = 0.18; two-time: F1,14 = 2.18,p = 0.92) of ventral gland depositing (Table 1). Similar to males,female hamsters given antibiotic treatment displayed no changein overall scent-marking behavior (PC-SMK, one-time:F1,16 = 0.80, p = 0.38; PC-SMK, two-time: F1,16 = 1.33, p = 0.27)when compared with females given control treatment. Specifically,antibiotic-treated females displayed no change in number (one-time: F1,16 = 0.88, p = 0.36; two-time: F1,16 = 1.37, p = 0.55) or dura-tion (one-time: F1,16 = 0.81, p = 0.38; two-time: F1,16 = 1.43,p = 0.25) of ventral gland depositing (Table 2).

4.2.7. Antibiotics did not affect grooming behaviorAntibiotic-treated males displayed no change in overall groom-

ing behavior (PC-GRM, one-time: F1,14 = 0.16, p = 0.70; PC-GRM,two-time: F1,14 = 2.45, p = 0.14) when compared with males givencontrol treatment. Specifically, antibiotic-treated males displayedno change in number (one-time: F1,14 = 0.51, p = 0.49; two-time:F1,14 = 1.49, p = 0.12) or duration (one-time: F1,14 = 1.57, p = 0.46;two-time: F1,14 = 4.85, p = 0.14) of self-grooming (Table 1). Similarto males, antibiotic-treated females displayed no change in overallgrooming behavior (PC-GRM, one-time: F1,16 = 0.85, p = 0.37; PC-GRM, two-time: F1,16 = 1.80, p = 0.39) when compared withfemales given control treatment. Specifically, antibiotic-treatedfemales displayed no change in number (one-time: F1,16 = 0.54,p = 0.43; two-time: F1,16 = 1.35, p = 0.26) or duration (one-time:

Fig. 6. Effects of antibiotics on investigative behavior across the sexes. Antibiotic treatment does not affect frequency of investigation in male (a) or female (b) hamsters. Barheights represent means ± S.E.M. An asterisk (*) indicates statistically significant differences between group means at p < 0.05.

Table 1Experiment 2: Means ± SEM of investigation, scent-marking, and grooming behaviors in male hamsters across treatment groups. No values were significantly different acrosstreatment groups.

Male D7 D14 D21 D28

Control Abx Control Abx Control Abx Control Abx

Frequency of Nose-to-Nose Investigation 17.63 ± 2.97 20.63 ± 4.25 13.13 ± 2.08 11.50 ± 2.19 7.88 ± 2.63 3.88 ± 0.91 11.88 ± 3.57 6.63 ± 1.21Duration of Nose-to-Nose Investigation 28.93 ± 3.90 27.59 ± 7.95 19.56 ± 9.59 14.04 ± 4.63 11.85 ± 5.88 2.24 ± 0.43 10.20 ± 3.32 5.53 ± 2.73Frequency of Nose-to-Anogenital

Investigation9.63 ± 3.03 9.25 ± 2.35 7.75 ± 2.38 6.63 ± 1.86 2.25 ± 0.80 2.75 ± 1.33 3.13 ± 1.04 3.13 ± 0.83

Duration of Nose-to-AnogenitalInvestigation

17.46 ± 5.43 13.90 ± 4.31 9.94 ± 3.72 6.89 ± 1.27 2.95 ± 1.30 1.83 ± 0.79 3.4 ± 1.26 4.84 ± 1.74

Total Frequency of Investigation 27.25 ± 5.26 29.88 ± 5.20 20.88 ± 3.46 18.13 ± 2.99 10.13 ± 3.18 6.63 ± 2.17 15.00 ± 3.87 9.75 ± 1.40Total Duration of Investigation 46.39 ± 7.56 41.49 ± 8.55 29.50 ± 11.88 20.93 ± 5.50 14.80 ± 6.09 4.06 ± 1.11 13.60 ± 3.81 10.36 ± 3.09Frequency of Scent-marking 1.00 ± 0.58 0.22 ± 0.15 2.00 ± 1.04 0.22 ± 0.15 1.00 ± 0.69 0.00 ± 0.00 1.13 ± 0.63 0.00 ± 0.00Duration of Scent-marking 1.02 ± 0.59 0.13 ± 0.09 2.41 ± 1.39 0.12 ± 0.08 1.28 ± 1.03 0.00 ± 0.00 1.64 ± 0.91 0.00 ± 0.00Frequency of Grooming 3.44 ± 0.96 8.89 ± 2.25 10.00 ± 2.57 9.89 ± 3.33 11.00 ± 3.39 2.88 ± 1.02 10.00 ± 2.95 3.75 ± 1.17Duration of Grooming 10.12 ± 4.69 20.47 ± 4.57 36.36 ± 11.37 25.80 ± 13.82 21.73 ± 6.21 4.29 ± 1.65 32.75 ± 13.03 6.85 ± 2.21

Table 2Experiment 2: Means ± SEM of investigation, scent-marking, and grooming behaviors in female hamsters across treatment groups. No values were significantly different acrosstreatment groups.

Female D7 D14 D21 D28

Control Abx Control Abx Control Abx Control Abx

Frequency of Nose-to-Nose Investigation 16.33 ± 1.99 18.22 ± 2.44 18.44 ± 4.29 19.67 ± 3.10 11.44 ± 2.28 17.00 ± 2.79 10.00 ± 2.36 14.56 ± 2.49Duration of Nose-to-Nose Investigation 17.58 ± 3.29 22.09 ± 4.57 14.69 ± 4.42 23.64 ± 6.01 10.04 ± 2.80 18.49 ± 4.45 6.19 ± 1.67 11.34 ± 2.12Frequency of Nose-to-Anogenital Investigation 9.00 ± 2.85 9.89 ± 2.09 7.33 ± 1.94 8.33 ± 1.25 8.78 ± 1.59 6.11 ± 1.37 6.00 ± 0.85 9.22 ± 1.77Duration of Nose-to-Anogenital Investigation 11.52 ± 4.50 17.97 ± 5.27 12.16 ± 2.08 12.61 ± 1.89 14.06 ± 3.28 9.03 ± 2.15 8.46 ± 1.97 14.00 ± 2.66Frequency of Scent-marking 0.00 ± 0.00 0.22 ± 0.22 0.11 ± 0.11 0.22 ± 0.15 0.33 ± 0.24 0.11 ± 0.11 0.44 ± 0.44 0.33 ± 0.24Duration of Scent-marking 0.00 ± 0.00 0.26 ± 0.26 0.09 ± 0.09 0.20 ± 0.15 0.32 ± 0.23 0.03 ± 0.03 0.56 ± 0.56 0.16 ± 0.10Frequency of Grooming 5.67 ± 1.12 5.56 ± 1.08 3.56 ± 0.91 5.56 ± 0.91 4.00 ± 0.69 3.33 ± 0 69 3.67 ± 0.73 2.44 ± 0.47Duration of Grooming 11.40 ± 2.88 11.06 ± 2.13 7.90 ± 1.31 12.78 ± 3.12 5.92 ± 1.02 5.98 ± 0.93 8.58 ± 2.07 4.21 ± 0.70

K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62 59

F1,16 = 1.07, p = 0.32; two-time: F1,16 = 1.09, p = 0.31) of self-grooming (Table 2).

4.2.8. Antibiotics did not affect body mass or reproductive physiologyAntibiotics did not affect male body mass (F1,16 = 1.35, p = 0.56),

testes mass (t16 = �1.68, p = 0.52), liver mass (t16 = 1.78, p = 0.10),or spleen mass (t16 = 1.40, p = 0.70), when compared with controlmales. Similarly, antibiotics did not affect female body mass(F1,16 = 3.19, p = 0.86), ovaries mass (t16 = �1.13, p = 0.90), uterinehorn mass (t16 = �1.14, p = 0.89), parametrial white adipose tissuemass (t16 = �0.98, p = 0.35), liver mass (t16 = �1.05, p = 0.96), orspleen mass (t16 = �1. 64, p = 0.54), when compared with controlfemales (Table 14 in Supplementary material).

5. Discussion

The gut microbiome is a diverse symbiotic community that actsas a vital part of the body’s ecosystem, and its disruption can have

lasting effects. Previous studies have shown that disrupting the gutmicrobiome can influence anxiety- and depression-like behaviors,as well as learning and memory (Foster and Neufeld, 2013;Mayer et al., 2015), yet the effects of gut dysbiosis via antibioticson aggressive behaviors had yet to be explored. Here we testedthe effects of antibiotic use on social behaviors, including aggres-sion, investigation, scent-marking, and grooming. In Experiment1, we administered antibiotics on a short-term basis, and assessedthe effects on the microbial communities in the gut; we found thatantibiotic treatment altered the gut microbiome in both male andfemale Siberian hamsters. In Experiment 2, we administeredrepeated doses of antibiotics and assessed social behaviors, anddiscovered an interesting sex difference in response to antibiotictreatment. Specifically, we found that male aggressive behaviordecreased after two antibiotic treatments, whereas investigation,scent-marking, and grooming remained unaffected. Aggressionreturned to normal following a ‘‘wash-out” period. In contrast tomales, females given a single antibiotic treatment showeddecreased aggression, with all other behaviors remaining the

60 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

same; females were unable to recover behaviorally from the initialtreatment. Here we provide data suggesting that antibiotic treat-ment with a fluoroquinolone antimicrobial agent that results inmarked disruption of the microbiome, may also have lasting, sex-specific effects on social behavior in hamsters, and likely otherrodent species. These robust antibiotic-induced changes in behav-ioral patterns across sexes may have potential implications for sex-specific treatment for many physiological as well as psychologicaldisorders.

5.1. The symbiotic community in the gut is vital to mammalian health

Despite inter-individual variability, short-term treatment withantibiotics, as well as repeated doses of antibiotics, altered the bac-terial composition of the gut in male and female Siberian hamsters,at the phylum level. Although Bacteroidetes and Firmicutes were inthe highest relative abundance in the microbiome of male andfemale hamsters, antibiotics did not affect these phyla; instead,those phyla in relatively low abundance (e.g., Cyanobacteria, Pro-teobacteria, and Tenericutes) were most strongly affected by treat-ment in both sexes. Much of the current research focuses explicitlyon phyla that are present in high abundance in the microbiome(e.g., Bacteroidetes and Firmicutes) and not those in low relativeabundance (Ley et al., 2006). In our studies, we have shown thatwhen Bacteroidetes and Firmicutes remain unchanged, other phylain the gut are altered, and we provide evidence suggesting that thegut microbiota may influence the brain and behavior.

In particular, in male and female hamsters in Experiment 1,there was a significant change in relative abundance of Cyanobac-teria, Proteobacteria, and Tenericutes following antibiotic treat-ment. Cyanobacteria, often found in the normal gut flora ofmammals (Sukenik et al., 2015), disappeared from the microbiotain both males and females. In both sexes, the relative abundanceof Cyanobacteria began to increase after 7 days of recovery, thoughnever reaching Pre-Treatment levels. Following antibiotic treat-ment in males and females, the relative abundance of Proteobacte-ria also decreased, however, Proteobacteria showed greaterrecovery in both male and female microbiota 7 days post-treatment. In contrast, the relative abundance of Tenericutesdecreased after antibiotic treatment in both males and females,and it did not recover in either sex after the wash-out period, sug-gesting that this phylum is less capable of recovering afterperturbation.

In addition to the three phyla discussed above, the male micro-biota also showed significant changes in relative abundance ofElusimicrobia, Euryarchaeota, and TM7 in Experiment 1. Followingantibiotic treatment in males, both Elusimicrobia and Eur-yarchaeota disappeared from the gut, and neither phylum reap-peared following recovery. In contrast, TM7, a recently describedsubgroup of gram-positive bacteria (Kuehbacher et al., 2008),decreased in relative abundance after antibiotic treatment, andrecovered following the 7-day recovery period. TM7 has beenshown to play a possible role in inflammatory bowel disease(IBD) and other similar diseases (Kuehbacher et al., 2008).

Although there were some shifts in other phyla in male andfemale microbiota, we found no significant change in the relativeabundance of seven of the thirteen phyla in males and ten of thethirteen phyla in females. Most phyla identified were comprisedof common genera often seen in other studies. Within the Firmi-cutes, we observed Ruminococcus, Clostridium, Lactobacillus,Eubacteria and Roseburia, often critical in gut motility and boweldisorders, yet the mechanisms behind these relationships are notwell understood (Machiels et al., 2013; Tremaroli and Bäckhed,2012). Within the Bacteroidetes, Actinobacteria, Proteobacteria,and Euryarchaeota phyla, we identified common genera, includingPrevotella, Bifidobacterium, Desulfovibrio, and Methanobrevibacter,

respectively, which aid in degradation of complex glycans, reduc-tion of sulfate, and production of intestinal methane (Tremaroliand Bäckhed, 2012). Similar to Experiment 1, following the secondtreatment period in Experiment 2, in males, Tenericutes wereaffected by antibiotic treatment and Cyanobacteria and Proteobac-teria, though not significant, were affected by the second treatmentof antibiotics as well. In females, treatment affected TM7 andTenericutes, and though not significant, the second treatment withantibiotics also affected Cyanobacteria.

Further, analysis of alpha and beta diversity in both experi-ments suggests that, in addition to the changes in relative abun-dance of specific phyla, antibiotic treatment also produced adecrease in diversity in both sexes, though to varying degrees.Maintaining a diverse microbial composition in the gut may be avital part of eliciting appropriate, species-typical behavior. Theprecise role that each bacterium plays in the community may notbe the most important factor in maintaining homeostasis withinthe body. Instead, the way in which each organism interacts withothers within the community may be most critical. We provideevidence that suggests bacteria in low relative abundance in thegut microbiome may have potential long-term consequences onphysiology and behavior.

5.2. Antibiotics affect behavior in a sex-dependent manner

Previous studies have suggested that what is lacking in much ofthe microbiome work to date is consideration of the role of sex inthe gut-brain axis (Foster and Neufeld, 2013). We demonstratehere that there is a strong sex difference in response to the sameantibiotic treatment, and that females were more strongly affectedby a single antibiotic-treatment period, with seemingly long-lasting effects on behavior. We also present data suggesting thatspecific behaviors (i.e., attacking) may be associated with the rela-tive abundance of particular bacteria in the gut. These findings sug-gest that females may be more sensitive to stressors, and thereforewith further insult, they will exhibit an even more robust change inbehavior.

The sex-specific results that we present here provide evidencethat while antibiotics are an important therapy for many infectiousdiseases in both sexes, the physiological and behavioral conse-quences of antibiotic-use in males and females should be furtherinvestigated. It is of further importance to focus on these sex differ-ences because, in humans, women are twice as likely to suffer fromanxiety and depression as men, and exactly why we see this is yetto be determined (Klein and Corwin, 2002; Kornstein et al., 2000).Notably, sex hormones (e.g., estradiol and testosterone) affect reg-ulation of motor and sensory function in the gastrointestinal (GI)tract, and estrogens plays a particularly important role in thesephysiological changes. Further, bowel disorders and mental dis-eases related to these disorders are more common in women thanin men as well (Mulak et al., 2014), suggesting the role that sexplays in regulation of the gut-brain axis. In a recent study, menwho self-reported feeling depressed also exhibited higher rates ofaggression, substance abuse, and risk taking when compared withwomen reporting the same feelings of depression (Martin et al.,2013). This study suggests that the behavioral symptoms oftenseen associated with depression and potentially other psychiatricdiseases, may vary across sexes.

5.3. Potential mechanisms mediating the influence of the gutmicrobiota on behavior

There are many potential neuroendocrine-immune mecha-nisms mediating microbiome–CNS cross-talk, and we are begin-ning to understand the role that each of those play in regulationof physiology and the resulting behavioral phenotype. Antibiotics

K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62 61

are a suitable tool to understand the cause of the microbiota-dependent changes we see in physiology and behavior (Fröhlichet al., 2016). A wide range of antibiotics has been used in recentmicrobiome research, often administered in cocktails of three ormore antibiotics (e.g., ampicillin, neomycin, vancomycin)(Fröhlich et al., 2016), forcefully fighting off microbial activity.Though this is successful in knocking down many of the bacterialcommunities in the gut, the effects of an individual antibiotic onthe microbiome and downstream effects on behavior in rodentspecies, is lacking. Previous studies have shown that applicationof many common antibiotics (e.g., clindamycin, ampicillin, van-comycin, erythromycin, and gentamicin) cause lethal enterocolitisin hamster species; the fluoroquinolone antibiotic used in the pre-sent study, enrofloxacin, does not produce the same lethal effects,and unlike some other antibiotics, it does not cross the blood brainbarrier (Alvarez et al., 2010; Bartlett et al., 2015; Ooie et al., 1997a,b). Thus, the effects of enrofloxacin on aggressive behavior in bothmales and females in this study may be due to the communicationbetween the gut microbiota and the brain, rather than a directeffect of antimicrobial use on behavior, yet we still do not knowthe specific mechanisms underlying this effect. As with previousstudies utilizing a range of antibiotics, it is not known if the antibi-otic’s effects on behavior are due to changes in the gut microbiotaalone. Many commonly used antibiotics have the ability to thin themucosal layer of the epithelium, increasing permeability of drugs,other molecules of the immune system, and even infectious mole-cules (reviewed in Ubeda and Pamer, 2012). The current studiesshed light on the potential role that the microbiota play in thebehavioral phenotype and provide evidence of the correlationbetween microbes and behavior. However, further studies couldfocus on more direct assessments of the influence of microbialcommunities on the brain and behavior.

Behavior may be strongly influenced by the gut microbiota viadirect communication with the CNS. For example, mice challengedwith the common gastrointestinal pathogen, Campylobacter jejuni(C. jejuni), exhibit reduced exploration in the open arms of the plusmaze, and the brains of these mice show increased expression of c-Fos labeling in autonomic brain regions, without activation of animmune response (Goehler et al., 2008). The lack of immuneresponse suggests that the vagus nerve may play an important rolein sending information from the gut microbiota to particularregions of the brain (Goehler et al., 2008). Further studies haveshown that challenging the gut microbiota with non-infectiousbacteria and probiotics, increases c-Fos mRNA levels in specificregions of the brain without activation of the systemic immuneresponse as well (Sudo et al., 2004). It is also possible that theeffects of antibiotics on behavior are due to changes in the micro-biota on other surfaces in addition to the gut. Future work will helpdetermine whether the effects on behavior seen here are directlyrelated to gut microbial changes or if the effects are indirect, actingvia other organ systems of the body.

Alternatively, the immune system, and particularly cytokines,may play a role in the cross-talk between the microbiota and thebrain, and this may influence the changes we see in behavior. Stud-ies suggest that gut dysbiosis is related to disturbances in adaptiveimmune cells and that some inflammatory bowel disorders may bethe consequence these changes (Lee and Mazmanian, 2010). Forexample, germ-free mice exhibit altered development of manyaspects of the immune system, including fewer, smaller and inac-tive lymph nodes, spleens, and Peyer’s patches, which thereby pro-hibit the immune system from responding appropriately (Hoshiet al., 1992; Pollard and Sharon, 1970). Germ-free mice are alsodeficient in particular cytokines, including IL-17, which aids inthe production of T helper cells during an immune response; how-ever, microbiota disturbance does not affect the levels of someother cytokines, suggesting that there are specific features of the

immune response that may be sensitive to these changes (Ivanovet al., 2008).

6. Conclusion

The gut microbiome is a diverse, host unique, and symbioticbacterial environment that works in concert with many otherorganisms in the body to influence the brain and behavior. Ourwork here suggests that antibiotic treatment alters the microbialcommunities in the gut in a sex-dependent manner, and this dis-ruption is correlated with social behavior. Determining the rolethat the microbiota play in communication with the brain acrossthe sexes is vital to our understanding of psychological disordersand the behavioral changes that are present in these diseases. Morebroadly, these data further our understanding of the behavioralconsequences of antibiotic use in a non-model species and demon-strate robust, sex-specific effects on social behavior. Our data pro-vide insight into a feature of the behavioral phenotype not yetdiscussed in this area of literature, and it may provide vast impli-cations for human psychological disease and treatment.

Author contributions

K.E.S. and G.E.D. designed the research; K.E.S., N.M.R., and E.A.S.performed the research; C.P.J. and K.E.S. analyzed the microbiomedata; N.M.R., K.E.S., and E.A.S. analyzed the behavioral and physio-logical data; and K.E.S., N.M.R., E.A.S., C.P.J., and G.E.D. wrote thepaper.

Conflict of interest

The authors declare no conflict of interest, financial orotherwise.

Acknowledgments

The authors thank Jim Pottebaum, DVM for his assistance inantibiotic treatment choice, and L.C. Beck, D.L. Boyes, and K.J.O’Malley for assistance in behavioral filming, necropsies, and gen-eral animal procedures. This work was supported by NationalScience Foundation Graduate Research Fellowship Program to N.M.R., Faculty Research Support Program Grant, and IndianaUniversity.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.bbi.2016.10.023.

References

Alvarez, A.I., Perez, M., Prieto, J.G., Molina, A.J., Real, R., Merino, G., 2010.Fluoroquinolone efflux mediated by ABC transporters. J. Pharm. Sci. 99, 4215–4227. http://dx.doi.org/10.1002/jps.

Archie, E.A, Theis, K.R., 2011. Animal behaviour meets microbial ecology. Anim.Behav. 82, 425–436. http://dx.doi.org/10.1016/j.anbehav.2011.05.029.

Bartlett, J.G., Chang, T., Moon, N., Onderdonk, A.B., 2015. Antibiotic-induced lethalenterocolitis in hamsters: studies with eleven agents and evidence to supportthe pathogenic role of toxin-producing clostridia. Am. J. Vet. Res. 39, 1525–1530.

Bercik, P., Denou, E., Collins, J., Jackson, W., Lu, J., Jury, J., Deng, Y., Blennerhassett, P.,MacRi, J., McCoy, K.D., Verdu, E.F., Collins, S.M., 2011. The intestinal microbiotaaffect central levels of brain-derived neurotropic factor and behavior in mice.Gastroenterology 141, 599–609. http://dx.doi.org/10.1053/j.gastro.2011.04.052.

Clarke, G., O’Mahony, S.M., Dinan, T.G., Cryan, J.F., 2014. Priming for health: gutmicrobiota acquired in early life regulates physiology, brain and behaviour. ActaPaediatr. Int. J. Paediatr. 103, 812–819. http://dx.doi.org/10.1111/apa.12674.

62 K.E. Sylvia et al. / Brain, Behavior, and Immunity 60 (2017) 51–62

Collins, S.M., Surette, M., Bercik, P., 2012. The interplay between the intestinalmicrobiota and the brain. Nat. Rev. Microbiol. 10, 735–742. http://dx.doi.org/10.1038/nrmicro2876.

Dinan, T.G., Cryan, J.F., 2012. Regulation of the stress response by the gutmicrobiota: implications for psychoneuroendocrinology.Psychoneuroendocrinology 37, 1369–1378. http://dx.doi.org/10.1016/j.psyneuen.2012.03.007.

Ericsson, A.C., Davis, J.W., Spollen, W., Bivens, N., Givan, S., Hagan, C.E., McIntosh, M.,Franklin, C.L., 2015. Effects of vendor and genetic background on thecomposition of the fecal microbiota of inbred mice. PLoS ONE 10, 1–19.http://dx.doi.org/10.1371/journal.pone.0116704.

Foster, J.A., Neufeld, K.M., 2013. Gut-brain axis: how the microbiome influencesanxiety and depression. Trends Neurosci. 36, 305–312. http://dx.doi.org/10.1016/j.tins.2013.01.005.

Fröhlich, E.E., Farzi, A., Mayerhofer, R., Reichmann, F., Jacan, A., Wagner, B., Zinser, E.,Bordag, N., Magnes, C., Fröhlich, E., Kashofer, K., Gorkiewicz, G., Holzer, P., 2016.Cognitive impairment by antibiotic-induced gut dysbiosis: analysis of gutmicrobiota-brain communication. Brain Behav. Immun. 56, 140–155.

Goehler, L.E., Park, S.M., Opitz, N., Lyte, M., Gaykema, R.P.A., 2008. Campylobacterjejuni infection increases anxiety-like behavior in the holeboard: possibleanatomical substrates for viscerosensory modulation of exploratory behavior.Brain Behav. Immun. 22, 354–366. http://dx.doi.org/10.1016/j.bbi.2007.08.009.

Gould, L., Zeigler, T., 2007. Variation in fecal testosterone levels, inter-maleaggression, dominance rank and age during mating and post-mating periodsin wild adult male ring tailed lemurs (Lemuar catta). Am. J. Primatol. 69, 1325–1339. http://dx.doi.org/10.1002/ajp.

Hill, M.O., 1973. Diversity and evenness: a unifying notation and its consequences.Ecology 54, 427–432.

Hoshi, H., Aijima, H., Horie, K., Nagata, H., Kaneko, T., Ikeda, T., 1992. Lymph folliclesand germinal center in popliteal lymph nodes and other lymphoid tissues ofgerm-free and conventional rats. Tohoku J. Exp. Med. 166, 297–307.

Ivanov, I.I., de Llanos Frutos, R., Manel, N., Yoshinaga, K., Rifkin, D.B., Sartor, R.B.,Finlay, B., Littman, D.R., 2008. Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine. Cell HostMicrobe 4, 337–349.

Jasnow, A.M., Huhman, K.L., Bartness, T.J., Demas, G.E., 2000. Short-day increases inaggression are inversely related to circulating testosterone concentrations inmale Siberian hamsters (Phodopus sungorus). Horm. Behav. 38, 102–110. http://dx.doi.org/10.1006/hbeh.2000.1604.

Jasnow, A.M., Huhman, K.L., Bartness, T.J., Demas, G.E., 2002. Short days andexogenous melatonin increase aggression of male Syrian hamsters(Mesocricetus auratus). Horm. Behav. 42, 13–20. http://dx.doi.org/10.1006/hbeh.2002.1797.

Johnston, R.E., 1993. Memory for individual scent in hamsters (Mesocricetus auratus)as assessed by habituation methods. J. Comp. Psychol. 107, 201–207. http://dx.doi.org/10.1037/0735-7036.107.2.201.

Jost, L., 2006. Entropy and diversity. Oikos 113, 363–375. http://dx.doi.org/10.1111/j.2006.0030-1299.14714.x.

Klein, L.C., Corwin, E.J., 2002. Seeing the unexpected: how sex differences in stressresponses may provide a new perspective on the manifestation of psychiatricdisorders. Curr. Psychiatry Rep. 4, 441–448. http://dx.doi.org/10.1007/s11920-002-0072-z.

Kornstein, S.G., Schatzberg, A.F., Thase, M.E., Yonkers, K.A., McCullough, J.P., Keitner,G.I., Gelenberg, A.J., Ryan, C.E., Hess, A.L., Harrison, W., Davis, S.M., Keller, M.B.,2000. Gender differences in chronic major and double depression. J. Affect.Disord. 60, 1–11. http://dx.doi.org/10.1016/S0165-0327(99)00158-5.

Kuehbacher, T., Rehman, A., Lepage, P., Hellmig, S., Folsch, U.R., Schreiber, S., Ott, S.J.,2008. Intestinal TM7 bacterial phylogenies in active inflammatory boweldisease. J. Med. Microbiol. 57, 1569–1576. http://dx.doi.org/10.1099/jmm.0.47719-0.

Lai, S.C., Johnston, R.E., 1994. Individual odors in Djungarian hamsters (Phodopuscampbelli). Ethology 96, 117–126.

Lee, Y.K., Mazmanian, S.K., 2010. Has the microbiota played a critical role in theevolution of the adaptive immune system? Science 330, 1768–1773. http://dx.doi.org/10.1126/science.1195568.Has (80-.).

Ley, R., Turnbaugh, P., Klein, S., Gordon, J., 2006. Microbial ecology: human gutmicrobes associated with obesity. Nature 444, 1022–1023. http://dx.doi.org/10.1038/nature4441021a.

Machiels, K., Joossens, M., Sabino, J., De Preter, V., Arijs, I., Eeckhaut, V., Ballet, V.,Claes, K., Van Immerseel, F., Verbeke, K., Ferrante, M., Verhaegen, J., Rutgeerts,P., Vermeire, S., 2013. A decrease of the butyrate-producing species Roseburiahominis and Faecalibacterium prausnitzii defines dysbiosis in patients withulcerative colitis. Gut 1–9. http://dx.doi.org/10.1136/gutjnl-2013-304833.

Martin, L.A., Neighbors, H.W., Griffith, D.M., 2013. The experience of symptoms ofdepression in men vs women. JAMA Psychiatry 70, 1100. http://dx.doi.org/10.1001/jamapsychiatry.2013.1985.

Martorell, J., Martínez, A., Soto, S., 2010. Complete ablation of vertical auditiveconduct and ear pinna in a dwarf hamster (Phodopus sungorus) with an auralspontaneous squamous cell carcinoma. J. Exot. Pet Med. 19, 96–100. http://dx.doi.org/10.1053/j.jepm.2009.06.015.

Mayer, E.A., Tillisch, K., Gupta, A., 2015. Gut/brain axis and the microbiota. J. Clin.Invest. 125, 926–938.

Mueller, N.T., Whyatt, R., Hoepner, L., Oberfield, S., Dominguez-Bello, M.G., Widen,E.M., Hassoun, A., Perera, F., Rundle, A., 2015. Prenatal exposure to antibiotics,cesarean section and risk of childhood obesity. Int. J. Obes. 39, 665–670. http://dx.doi.org/10.1038/ijo.2014.180.Prenatal.

Mulak, A., Taché, Y., Larauche, M., 2014. Sex hormones in the modulation of irritablebowel syndrome. World J. Gastroenterol. 20, 2433–2448. http://dx.doi.org/10.3748/wjg.v20.i10.2433.

Nelson, R.J., Demas, G.E., Huang, P.L., Fishman, M.C., Dawson, V.L., Dawson, T.M.,Snyder, S.H., 1995. Behavioural abnormalities in male mice lacking neuronalnitric oxide synthase. Nature 378, 383–386.

O’Mahony, S.M., Marchesi, J.R., Scully, P., Codling, C., Ceolho, A.M., Quigley, E.M.M.,Cryan, J.F., Dinan, T.G., 2009. Early life stress alters behavior, immunity, andmicrobiota in rats: implications for irritable bowel syndrome and psychiatricillnesses. Biol. Psychiatry 65, 263–267. http://dx.doi.org/10.1016/j.biopsych.2008.06.026.

Ooie, T., Terasaki, T., Suzuki, H., Sugiyama, Y., 1997a. Kinetic evidence for activeefflux transport across the blood-brain barrier of quinolone antibiotics. J.Pharmacol. Exp. Ther. 283, 293–304.

Ooie, T., Terasaki, T., Suzuki, H., Sugiyama, Y., 1997b. Quantitative brainmicrodialysis study on the mechanism of quinolones distribution in thecentral nervous system. Drug Metab. Dispos. 25, 784–789.

Pellis, S.M., Pellis, V.C., 1988. Identification of the possible origin of the body targetthat differentiates play fighting from serious fighting in syrian golden hamsters(Mesocricetus auratus). Aggress. Behav. 14, 437–449. http://dx.doi.org/10.1002/1098-2337(1988) 14:6<437::AID-AB2480140605>3.0.CO;2-Q.

Pollard, M., Sharon, N., 1970. Responses of the Peyer’s patches in germ-free mice toantigenic stimulation. Infect. Immun. 2, 96–100.

Reasner, D.S., Johnston, R.E., 1987. Scent marking by male dwarf hamsters(Phodopus sungorus campbelli) in response to conspecific odors. Behav. NeuralBiol. 48, 43–48.

Rendon, N.M., Rudolph, L.M., Sengelaub, D.R., Demas, G.E., 2015. The agonisticadrenal: melatonin elicits female aggression via regulation of adrenalandrogens. Proc. R. Soc. B Biol. Sci. 282 (1819). http://dx.doi.org/10.1098/rspb.2015.2080.

Rendon, N.M., Soini, H.A., Scotti, M.A.L., Novotny, M.V., Demas, G.E., 2016a. Urinaryvolatile compounds differ across reproductive phenotypes and followingaggression in male Siberian hamsters. Physiol. Behav. 164, 58–67. http://dx.doi.org/10.1016/j.physbeh.2016.05.034.

Rendon, N.M., Soini, H.A., Scotti, M.A.L., Weigel, E.R., Novotny, M.V., Demas, G.E.,2016b. Photoperiod and aggression induce changes in ventral gland compoundsexclusively in male Siberian hamsters. Horm. Behav. 81, 1–11. http://dx.doi.org/10.1016/j.yhbeh.2016.02.005.

Romick-Rosendale, L.E., Goodpaster, A.M., Hanwright, P.J., Patel, N.B., Wheeler, E.T.,Chona, D.L., Kennedy, M.A., 2009. NMR-based metabolomics analysis of mouseurine and fecal extracts following oral treatment with the broad-spectrumantibiotic enrofloxacin (Baytril). Magn. Reson. Chem. 47, S36–S46. http://dx.doi.org/10.1002/mrc.2511.

Scattoni, M.L., Ricceri, L., Crawley, J.N., 2011. Unusual repertoire of vocalizations inadult BTBR T+tf/J mice during three types of social encounters. Genes BrainBehav. 10, 44–56. http://dx.doi.org/10.1111/j.1601-183X.2010.00623.x.

Slate, A.R., Bandyopadhyay, S., Francis, K.P., Papich, M.G., Karolewski, B., Hod, E.A.,Prestia, K.A., 2014. Efficacy of enrofloxacin in a mouse model of sepsis. J. Am.Assoc. Lab. Anim. Sci. 53, 381–386.

Soma, K.K., Rendon, N.M., Boonstra, R., Albers, H.E., Demas, G.E., 2015. DHEA effectson brain and behavior: insights from comparative studies of aggression. J.Steroid Biochem. Mol. Biol. 145, 261–272. http://dx.doi.org/10.1016/j.jsbmb.2014.05.011.

Sudo, N., Chida, Y., Aiba, Y., Sonoda, J., Oyama, N., Yu, X.-N., Kubo, C., Koga, Y., 2004.Postnatal microbial colonization programs the hypothalamic–pituitary–adrenalsystem for stress response in mice. J. Physiol. 5581, 263–275. http://dx.doi.org/10.1113/jphysiol.2004.063388.

Sukenik, A., Quesada, A., Salmaso, N., 2015. Global expansion of toxic and non-toxiccyanobacteria: effect on ecosystem functioning. Biodivers. Conserv. 24, 889–908. http://dx.doi.org/10.1007/s10531-015-0905-9.

Thomas, M.A., Spencer, J.F., Toth, K., Sagartz, J.E., Phillips, N.J., Wold, W.S.M., 2008.Immunosuppression enhances oncolytic adenovirus replication and antitumorefficacy in the Syrian hamster model. Mol. Ther. 16, 1665–1673. http://dx.doi.org/10.1038/mt.2008.162.

Tremaroli, V., Bäckhed, F., 2012. Functional interactions between the gut microbiotaand host metabolism. Nature 489, 242–249. http://dx.doi.org/10.1038/nature11552.

Ubeda, C., Pamer, E.G., 2012. Antibiotics, microbiota, and immune defense. TrendsImmunol. 33, 459–466. http://dx.doi.org/10.1016/j.it.2012.05.003.

Wallace, T.C., Guarner, F., Madsen, K., Cabana, M.D., Gibson, G., Hentges, E., Sanders,M.E., 2011. Human gut microbiota and its relationship to health and disease.Nutr. Rev. 69, 392–403. http://dx.doi.org/10.1111/j.1753-4887.2011.00402.x.

Wynne-Edwards, K.E., 2003. From dwarf hamster to daddy: the intersection ofecology, evolution, and physiology that produces paternal behavior. Adv. StudyBehav. 32, 207–261. http://dx.doi.org/10.1016/S0065-3454(03)01005-2.

Wynne-Edwards, K.E., Lisk, R.D., 1987. Behavioral interactions differentiateDjungarian (Phodopus campbelli) and Siberian (Phodopus sungorus) hamsters.Can. J. Zool. 65, 2229–2235. http://dx.doi.org/10.1139/z87-337.