Transcript
REVIEW Regulation of the stressresponse by thegutmicrobiota: Implications forpsychoneuroendocrinology TimothyG.Dinan*,JohnF.Cryan Alimentary Pharmabiotic Centre, University College Cork, Cork, Ireland Received 1 February 2012; received in revised form 7March 2012; accepted 7March 2012 Contents 1.Introduction .. . .. . . . . .. . . .. . . . . . . . . . . .. . .. . . . . .. . ... .. . .. . .. . ... .. . .. . .. . .. ... . .. . ... . . . 1370 2. Brain—gut axis . .. . . . . .. . . .. . . . . . . . . . . .. . .. . . . . .. . ... .. . .. . .. . ... .. . .. . .. . .. ... . .. . ... . . . 1371 3. ENS and CRF . . .. . . . . .. . . .. . . . . . . . . . . .. . .. . . . . .. . ... .. . .. . .. . ... .. . .. . .. . .. ... . .. . ... . . . 1371 Psychoneuroendocrinology (2012) 37 ,1369—1378 KEYWORDS Brain—gut axis;Microbiota;HPA;Probiotics;Germ-free;Stress;Novel psychotropics Summary There is now an expanding volume ofevidence to support the view that commensalorganisms within the gutplay a role in early programming and later responsivity ofthe stresssystem. The gut is inhabited by10 13 —10 14 micro-organisms, which is ten times the number of cellsinthe human body and contains 150 times as many genes as ourgenome. It has long beenrecognised that gutpathogens such as Escherichia coli ,if they enter the gut can activate the HPA.However, animals raised in a germ-free environment show exaggerated HPA responses topsychological stress, which normalises with monocolonisation by certain bacterial speciesincluding Bifidobacterium infantis . Moreover, increased evidence suggests that animals treatedwith probiotics have a blunted HPA response. Stress induces increased permeability of the gutallowing bacteria and bacterial antigens to cross the epithelial barrier and activate a mucosalimmune response, which in turn alters the composition of the microbiome and leads to enhancedHPAdrive. Increasing data from patients with irritable bowel syndrome and major depressionindicate that in these syndromes alteration of the HPA may be induced by increased gutpermeability. In the case ofirritable bowel syndrome the increased permeability can respondto probiotic therapy. Detailed prospective studies in patients with mood disorders examining thegutmicrobiota, immune parameters andHPA activity are required to throw further light on thisemerging area. It is however clear that the gut microbiota must be taken into account whenconsidering the factors regulating the HPA. # 2012 Elsevier Ltd. All rights reserved. * Corresponding author at: Department of Psychiatry, Cork University Hospital, Wilton, Cork, Ireland. Tel.:+353 214901224. E-mail address:
[email protected] (T.G. Dinan). Available online at www.sciencedirect.com journa lh omepa ge: www.e lsevier.c om/l ocate/psyne ue n0306-4530/$ — see front matter # 2012Elsevier Ltd. All rights reserved.doi:10.1016/j.psyneuen.2012.03.007 4. Content of gut microbiota ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1371 5. Microbiota and the HPA .. ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1372 6. Probiotics and the stress response .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1373 7. Antibiotics and stress .... ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1374 8. Probiotics and glucocorticoids . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1374 9. Influence of stress on themicrobiome ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1374 10. Concluding remarks .. ... ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1375 Acknowledgements .. ... ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1376 References . .. . . . . .. ... ... . .. .. . ... . .. . . . ... ... . ... . . ... . ... . . . . . .... ... . . . . .. ... . .. . . . . 1376 1.Introduction The hypothalamic—pituitary-adrenal axis (HPA) and otheraspects of the stress response are regulated by exposureboth to psychological stressors and physical stressors such asinfections (Dinan and Scott, 2005).Response of the HPA andthe sympathoadrenal medullary system (SAM) to psychologi-cal stress is mediated by key neurotransmitter systems suchas the serotonergic (5HT) and norepinephrine (NE) systemswith endorphins playing an important inhibitory role (Dinan,1996; Bhatnagar and Dallman, 1998).Infective agents acti-vate the neuroendocrine system via pro-inflammatory cyto-kines, which exert influence on the hypothalamus especiallyon the parvicellular neurons of theparaventricular nucleus(Dinan, 2001; Rivest, 2010). Moreover, theHPA is tightlyregulated torespond efficiently to gut pathogens such as Escherichia coli .In anelegant series of recent studiesZimomra et al. (2011) demonstrated that the initial activa-tion of the HPA axis inresponse to E. coli infection islargelymediated by COX-induced prostanoid synthesis. This riseincirculating corticosterone correlates with the riseincircu-lating PGE2 and administration of indomethacin (non-selec-tive COX inhibitor), abolished the early rise inplasmacorticosterone. Furthermore, corticosterone levels reach apeak before maximum circulating ACTH, indicating thatprostaglandins stimulate corticosterone release through anACTH-independent mechanism. When prostaglandin produc-tion iscompletely inhibited, corticosterone levels increased2h following E. coli challenge, confirming that other factorsstimulate HPA responses atthis later time. Asignificantcorrelation between the rise of pro-inflammatory cytokinesand corticosterone is observed. Administration of rat IL-6antibodies attenuates theelevation incorticosterone 2 hfollowing E. coli challenge. This dataisconsistent withprevious work showing that PGE2 directly mediates corticos-terone release from cultured rat adrenals (Mohn et al., 2005)and cortisol release from human adrenal H295R cells(Vakharia and Hinson, 2005) (Fig. 1). Figure 1 At a hypothalamic level classic neurotransmitters andcytokines regulate corticotrophin releasing hormone (CRH) andvasopressin(AVP) release into the portal vasculature. A series of negative feedback loops controls the forward drive. The adrenalcortex can be directly activated by PGE2 from the immune system stimulated by gut pathogens. 1370 T.G. Dinan, J.F.Cryan However, that non-pathogenic gut microbes might influ-ence the HPA isarelatively novel concept. There isnow anexpanding volume of evidence to support theview thatcommensal organisms within the gut play a rolein earlyprogramming and later responsivity of the stress system(Grenham et al., 2011). This network has been termed thebrain—gut—microbiota and here we will review the datasupporting the role of gut microbes in determining stressreactivity especially through the HPA. 2. Brain—gutaxis Itis well established that the brain regulates gut activity butrecent attention has focused on thereverse pathway and themanner in which gut microbes can influence the brain (Gren-ham et al., 2011). This brain—gut—microbiota axis includesthe central nervous system (CNS), theneuroendocrine andneuroimmune systems, the sympathetic and parasympa-thetic arms of the autonomic nervous system, the entericnervous system (ENS) and of course the intestinal microbiota(Rhee et al., 2009; Cryan and O’Mahony, 2011). These com-ponents interact to form a complex reflex network withafferent fibres that project to integrative CNS structuresand efferent projections to the smooth muscle. This bidirec-tional communication network enables signals from the brainto influence the motor, sensory and secretory modalities of the gastrointestinal tract and conversely, visceral messagesfrom the gut can influence brain function, especially areas of the brain devoted to stress regulation, most notably thehypothalamus.The vagus provides an important line of communicationbetween the gut microbiota and the HPA. Hosoi et al. (2000)measured the expression of CRF mRNA inthe hypothalamusand plasma levels of ACTH and corticosterone after vagalstimulation inrodents. CRF mRNA in the hypothalamus wasincreased 2h after vagal stimulation and plasma levels of ACTH were markedly elevated. They also observed increasesin plasma levels of corticosterone. Ofclinical relevance isthefact that vagal nerve stimulation is associated withclinicalantidepressant benefit (Nemeroff et al., 2006) coupled withnormalisation of HPA parameters in patients with treatmentrefractory depression (O’Keane et al., 2005), indicating thatthe vagus not only impacts on HPA activity but also on thecore pathophysiology of major depression.The top-down and bottom-up perspective of informationflow as well asthe detailed structural integration and func-tioning of the various brain—gut—microbe axis componentshas been reviewed extensively elsewhere (Mayer, 2011). It isbecoming clear that alterations in this communication play afundamental role in a wide variety of disorders. Moreover,the possibility of regulating the gut microbiota is opening upas atractable therapeutic target for a host of stress-relateddisorders. 3.ENS andCRF The ENS is acomplex neuronal network with multiple neuro-transmitters and neuromodulators including 5-HT, acetylcho-line and CRF. Although prominence has been given to thecentral srcins of CRF-mediated changes in gastrointestinalfunction, the presence of theCRFligand and receptors withinthe ENS indicate the likelihood that peripheral pathways alsoplay a role inthe local regulation of gut function during timesof stress (O’Malley et al.,2010). Indeed, activation of gutCRFR1 contributes to stress-induced increases in colonicmotility, defecation, permeability and visceral pain sensa-tion (Larauche et al., 2009). Activation of peripheral CRFR2inhibits gastric emptying, suppresses stimulated colonicmotor function and prevents hypersensitivity to repeatedcolorectal distension. CRFR2 has also been proposed to havea roleinstress-induced permeability dysfunction and themodulation of mucosal immune andinflammatory responsesin the colon (Gareau et al., 2008). Furthermore, CRF candirectly activate myenteric neurons to increase colonic moti-lity and permeability and stimulate diarrhoea in rodents(Tache´, 2004). As the ENS iscomprised of sensory, motorand interneurons which are bidirectionally linked to theCNSvia sympathetic and parasympathetic pathways, the expres-sion of CRFR1 and CRFR2 on these neurons and nerve fibresplaces them inan ideal position to act assignalling peptidesin thebrain—gut axis (Tache etal.,1999).The contrasting actionsof CRFR1 andCRFR2 are furtherunderlined by differential expressionpatterns. CRFR2 ispre-valent inupper regionsof theguttract (Wuet al., 2008),whereas CRFR1 is more widespreadin thecolon. CRFR1 isstrongly expressed inmucosal cells (O’Malley et al., 2010),where it mayregulatestress-inducedionsecretion andpara-cellular permeability leading to bacterial—hostinteractionsand mucosal inflammation. CRFR2 hasalso been detectedinthe colonicmucosa andhas similarly been proposed tohave arole instress-inducedpermeability dysfunctionand the mod-ulation of mucosalimmune and inflammatory responses in thecolon (Teitelbaum et al., 2008),but isalsoimportant inexerting anti-nociceptiveeffects on visceral pain (Millionet al., 2005).Evidence isnowmounting that stressresultsinthe recruitment and activationof CRF receptorsin the colontoinduce thestress-related changes in gutfunctionand that aheightened stresssusceptibility results in altered expression of CRF receptors. 4.Contentofgut microbiota The gut isinhabited by 10 13 —10 14 micro-organisms, which isten times thenumber of human cells in our bodies andcontains 150 timesas many genes asour genome (Qinet al.,2010). The estimated species number varies greatlybut it is generally accepted that the adult microbiome con-sists of greater than 1000 species and more than 7000 strains(Ley etal., 2006). It is an environment dominated by bac-teria, mainly strict anaerobes, but also including viruses,protozoa, archae and fungi. The microbiome islargelydefined by 2 bacterial phylotypes, Bacteroidetes and Firmi-cutes with Proteobacteria, Actinobacteria, Fusobacteria,and Verrucomicrobia phyla present in relatively low abun-dance (Xu et al.,2007).Traditional culture-based analysis used to define theenteric flora, is only adequate for a minority of the gutmicrobiota that isamenable to cultivation (Xuet al.,2007;Eckburg et al.,2005). This methodological problem has beencircumvented by the use of culture-independent techniques, aportfolio consisting ofsequencing based methods, geneticfingerprinting, fluorescently labelled oligonucleotide probesRegulation of the stress response by the gut microbiota 1371 (FISH), quantitative PCR aswellas metagenomic approaches(Archie and Theis, 2011). Therealisation thatthe secretoryand metabolic capability of the microbiome was likely asimportant as phylotype composition has also led to the useof metabolomic andmetaproteomic approaches to improveour understanding of gut microbial—host interactions (O’Haraand Shanahan, 2006). Unfortunately, advances in culturemethods havenotkeptpace with the riseof these alternativetechnologies and adual-pronged line of attack may berequired to complete the circle,anotinconsiderable logisticalchallenge.While the gut shows enormous microbial diversity it isnowclear that there are important developmental and longitu-dinal variations which impact on the functioning of theHPAand the wider stress axis. Colonisation of the infant gutcommences atbirth when delivery exposes the infant to acomplex microflora and its initial microbiome has amaternalsignature (Ma¨ndar and Mikelsaar, 1996; Mackie et al., 1999;Adlerberth and Wold, 2009). The microbiome of unweanedinfants is simple withhigh inter-individual variability (Kur-okawa et al., 2007; McCracken and Lorenz, 2001).Thenumbers and diversity of strict anaerobes increase as a resultof diet and environment, and after 1 year of age a complexadult-like microbiome is evident. Despite asignificant inter-personal variation in the enteric microbiota, there seems tobe abalance that confers health benefits and an alteration inbeneficial bacteria can negatively influence the wellbeing of the individual (Cryan and O’Mahony, 2011). Several factorsmay alter the microbiome such as infection, disease, diet andantibiotics, but as ageneral rule, ittends to revert to thestable diversity established in infancy once the threat of theinitial distorting factor has subsided (Forsythe et al., 2010).Interestingly, ithas been shown that the core microbiota of an aged individual is distinct from that of younger adults(Claesson etal.,2011) and that age related shifts inthecomposition of theintestinal microbiota are linked toadverse health effects in the elderly host (Woodmansey,2007). The number of Bifidobacteria decreases withage(Claesson etal.,2011) and parallels changes in health statusand decreased plasticity within the HPA. It is important tonote that the HPA has been strongly implicated in suscept-ibility to the development of obesity and the metabolicsyndrome (Dallman, 2010; Finger et al., 2011, 2012). Intandem increasing evidence points to a role of gut microbiotain obesity (Turnbaugh and Gordon, 2009; Davey et al.,2012).Whether HPA and gut microbiome interact directly in obesityremains an area for future investigation. 5.MicrobiotaandtheHPA The use of germ free (GF) animals (with no bacterial expo-sure) has provided one of themost significant insights into therole of the microbiota inregulating the development of theHPA. The germ free paradigm is based on thefact that theuterine environment issterile during prenatal development(Adlerberth and Wold, 2009) and with surgical delivery repla-cing the normal vaginal delivery, the opportunity for post-natal colonisation of the gut iseliminated once animals aremaintained ina sterile environment. Subsequent comparisonwith their conventionally colonised counterparts allowsinferences to bedrawn regarding the morphological andphysiological parameters that may beunder the influenceof the developing microbiota. An alternative approach is theinduction of dysbiosis of the enteric flora, either throughadministration of antibiotics or deliberate infection in pre-clinical studies (Bennet et al.,2002).Broad spectrum anti-biotics inparticular are known to perturb the microbiome byreducing biodiversity and delaying colonisation and arewidely used asa method to intentionally alterthemicro-biome ina reproducible manner (Donskey et al., 2003).The morphological consequences of growing up germ freewere evidenced by thegreatly enlarged cecum, reducedintestinal surface area, increased enterochromaffin cellarea, smaller Peyer’s Patches and smaller villous thicknessinthese animals compared to conventional controls (Abramset al.,1963). It was not surprising, given these gross struc-tural aberrations, that multiple facets of normal functionincluding that of the stress response would also be affected.Toll-like receptors (TLRs) are present on cells of theinnateimmune system and recognise characteristic moleculestermed pathogen associated molecular patterns (Akira andHemmi, 2003).These receptors are the gateway to animmune response. Pathogen recognition by aparticularTLR results in a cascade of events leading to the activationof the NF- k B signalling system, production and release of cytokines and activation of the HPA. However intheabsenceof the resident enteric flora, key members of the TLR familyhave low orabsent expression profiles inthe gut,thuscompromising appropriate immune and neuroendocrineresponses to pathogenic threats (O’Hara and Shanahan,2006). For example theTLR4 knockout mouse does notrespond to Gram negative bacteria with an activation of the HPA (Gosselin and Rivest, 2008)Seminal studies by Sudo et al. (2004) provide insight intothe role of theintestinal microbiota inthe development of the HPA axis. In germ free mice a mild restraint stress inducesan exaggerated release of corticosterone and ACTH com-pared to thespecific pathogen free (SPF) controls. The stressresponse intheGF miceis partially reversed by colonisationwith faecal matter from SPF animals and fully reversed bymonoassociation with Bifidobacterium infantis in atimedependant manner (Bailey and Coe, 1999). This study clearlydemonstrated that themicrobial content of the gut is criticalto the development of anappropriate stress response laterinlife and also that there is anarrow window in early life wherecolonisation must occur to ensure normal development of theHPA axis.The question emerges whether the gut flora can have aninfluence over neural circuits and behaviour associated withthe stress response? Sudo et al.(2004) reported adecrease inbrain derived neurotrophic factor (BDNF), akey neurotrophininvolved in neuronal growth and survival, and expression of the NMDA receptor subunit 2a (NR2a) inthe cortex andhippocampus of male GF animals compared to SPF controls.On the other hand, Neufeld et al.(2011) actually found anincrease inhippocampal BDNF mRNA in female mice that wascontrary to the protein decreases observed in the earlierstudy. We have recently also found decreases in hippocampalBDNF mRNA levels as well as adistinct changes intheserotonergic system in male but not female mice (Clarkeet al.,submitted for publication). This suggests that areg-ulation of microbiome—gut—brain axis may besex depen-dent. Alterations inhippocampal NMDA and 5HT1A receptor1372 T.G. Dinan, J.F.Cryan expression have been shown ina number of studies (Neufeldet al., 2011). Both of these receptors areknown to influenceCRH release fromthe hypothalamus and changes inexpres-sion may explain altered HPA function in germ free animals.This decreased anxiety in germ free animals has been repro-duced in other laboratories in both the elevated plusmazeand the light dark box (where germ free animals spent moretime in the light compartment) (Heijtz et al., 2011).It islong knownthatstress andHPAcaninfluence thecomposition ofgutmicrobiome(Tannock andSavage,1974). However,thefunctional consequences ofsuchchanges arenowonly beingunderstood. Maternal separa-tion, an early lifestressor which can resultinlong-termHPA changes (O’Mahonyetal., 2011) has beenshown to,cause asignificantdecrease infaecal lactobacillion day 3post separation, which returnstobaseline by day 7 asassessedby enumeration oftotalandGram-negative aero-bic and facultativeanaerobic bacterialspecies(Bailey andCoe, 1999).However, earlylifestress can also havelongtermeffectsonthemicrobiome. Analysis of the16S rRNAdiversity inadult rats exposed to maternal separationfor3 h perday from post natal days 2—12 revealed a signifi-cantlyaltered faecalmicrobiomewhencomparedtothenon-separated controlanimals(O’Mahonyetal.,2009).Astudy usingbacterial tagencodedFLXamplicon pyrose-quencing (bTEFAP) demonstrated that thecommunitystructure ofmicrobiota frommiceexposedtoa prolongedrestraintstressorwas significantlydifferent thanthecom-munity structurefoundinnon-stressed control mice (Bai-ley et al.,2010). Morerecently usingthesame approachrepeated socialstressor hasbeen shown todecrease therelative abundanceofbacteriainthe genusBacteroides,while increasingthe relative abundance of bacteria in thegenusClostridiuminthececum.Thestressor alsoincreased circulatinglevelsof IL-6 and MCP-1, which weresignificantlycorrelated withstressor-induced changestothreebacterial genera(i.e.,Coprococcus, Pseudobutyrivi-brio, and Dorea).Interestingly,antibiotic exposure blockedthe increase IL-6and MCP-1. Thesedata showthatexpo-sure torepeated stressaffectsgutbacterial populationsina cytokinedependentmanner (Bailey et al., 2011). 6.Probioticsandthestressresponse A probiotic isgenerallydefined as a livemicro-organismwhich whenadministeredinadequate amountsconfersahealth benefit onthe host (Quigley,2008).Probioticsareemerging as potential therapeuticsfor stress-related gas-trointestinal disorders such as irritablebowelsyndrome.Overstated and exaggeratedclaimsfor thehealth benefitsof probiotics have beenmade.In realitymany of theseclaimsarebased on weak or non-existent data.Morerecently,probiotic administrationstudiessupport a roleforthemicrobiotainanxiety like behaviours (Bercik et al.,2011a). Administration of L.helveticus R0052 and B.longum R0175 taken incombinationproduces anxiolytic-likeactivityinrats (Messaoudi etal.,2011). This findingmustbe viewed cautiously giventhefact thatmany smallmolecules which apparentlyhave anxiolytic impact inrodentslack efficacy whensubjectedto rigorousevalua-tion in man.Arecent studyfound that chronic treatment with theprobiotic Lactobacillus rhamnosus over 28 days producedanimals with lower levels of corticosterone andreduceddepressive behaviours in the forced swim test in additionto a less anxious phenotype in the elevated plusmaze (Bravoet al., 2011). Alterations inGABA the maininhibitory neuro-transmitter system were also observed. L. rhamnosus treatedanimals showed alterations of GABAB1b mRNA inthe brainwith increased expression incortical regions and decreasedexpression in thehippocampus, amygdala, and locus coer-uleus aswell as reduced GABAA a 2mRNA expression intheprefrontal cortex and amygdala and increased GABAA a 2inthe hippocampus. The mechanism behind these changes waspartially elucidated by studies in vagotimised animals. Theneurochemical and behavioural effects of this bacterium, donot occur following vagotomy, indicating that the vagus is akey route of communication between probiotic bacteria andthe brain. A rolefor the gut microbiota inpain perception hasalso been indicated, with, one study demonstrating thatspecific Lactobacillus strains could induce the expressionof m -opioid and cannabinoid receptors in intestinal epithelialcells and mimic the effects of morphine inpromoting analge-sia (Rousseaux etal.,2007). HPA parameters were not mea-sured in the study.Bercik et al.(2011a) have shown that infection-inducedbehavioural changes were associated with decreased hippo-campal BDNF mRNA which could bereversed by a B. longum without affecting cytokine ortryptophan metabolism. Inter-estingly, these authors find that manyeffects of probioticsoccur independent of vagus nerve activation (Bercik et al.,2010, 2011b).Other potential mechanisms through which probiotics mayinfluence the HPA and stress responsivity include neurotrans-mitter modulation. B. infantis 35624, for example, has beenshown inSprague-Dawley rats to induce an elevation inplasma tryptophan levels, a precursor to serotonin (5-HT)which isakey neurotransmitter within the brain—gut axis(Desbonnet etal., 2008).Since CNStryptophan concentra-tions are largely dependent on peripheral availability and theenzymatic machinery responsible for theproduction of 5-HTis not saturated at normal tryptophan concentrations (Rud-dick et al., 2006),the implication here is that the microbiotamight play some rolein theregulation of CNS as well asenteric nervous system 5-HTsynthesis. This effect ispoten-tially mediated by the impact of themicrobiota on theexpression of indoleamine-2,3-dioxygenase, a key enzymein the physiologically dominant kynurenine pathway of tryp-tophan degradation (Forsythe etal., 2010) but of coursemultiple mechanisms are possible and indeed likely, giventhe strain specific effects that have been observed in manyprobiotic studies to date.Diet plays an important rolein relation to the compositionof the microbiota and alterations indiet are known to changethe microbial content of thegut. Aclinical trial wasper-formed on apopulation of 30 human subjects, whowereclassified in low and high anxiety traits (Martin et al., 2009).Biological fluids (urine and blood plasma) were collectedduring 3 test days atthe beginning, midway and at theend of a2 week study. NMR and mass spectroscopy basedmetabonomics were employed to study global changes inmetabolism. Human subjects with higher anxiety traitsshowed adistinct metabolic profile indicative of adifferentRegulation of the stress response by the gut microbiota 1373