Effects of prenatal stress on neuroactive steroid responses to acute stress in adult male and female rats

Abstract Acute swim stress results in the robust production of several neuroactive steroids, which act as mediators of the stress response. These steroids include glucocorticoids, and positive GABAA receptor modulatory steroids such as allopregnanolone and tetrahydrocorticosterone (THDOC), which potentiate inhibitory GABA signalling, thereby playing a role in the negative control of the hypothalamic‐pituitary‐adrenal (HPA) axis. Prenatally stressed (PNS) offspring exhibit increased vulnerability to stress‐related disorders and frequently display exaggerated HPA axis responses to stressors during adulthood, which may be a result of reduced neuroactive steroid production and consequently inhibitory signalling. Here, we investigated whether exposure of rats to prenatal social stress from gestational day 16‐20 altered neuroactive steroid production under non‐stress conditions and in response to an acute stressor (swim stress) in adulthood. Using liquid chromatography‐mass spectrometry, nine neuroactive steroids were quantified (corticosterone, deoxycorticosterone [DOC], dihydrodeoxycorticosterone, THDOC, progesterone, dihydroprogesterone, allopregnanolone, pregnenolone, testosterone) in plasma and in five brain regions (frontal cortex, hypothalamus, amygdala, hippocampus, brainstem) of male and female control and PNS rats. There was no difference in the neuroactive steroid profile between control and PNS rats under basal conditions. The increase in circulating corticosterone induced by acute swim stress was similar in control and PNS offspring. However, greater stress‐induced corticosterone and DOC concentrations were observed in the brainstem of male PNS offspring, whereas DOC concentrations were lower in the hippocampus of PNS females compared to controls, following acute stress. Although PNS rats did not show deficits in allopregnanolone responses to acute stress, there were modest deficits in the production of THDOC in the brainstem, amygdala, and frontal cortex of PNS males and in the frontal cortex of PNS females. The data suggest that neuroactive steroid modulation of GABAergic signalling following stress exposure may be affected in a sex‐ and region‐specific manner in PNS offspring.


| INTRODUC TI ON
Several steroids are rapidly produced as a physiological response to acute stress, where they act as allostatic mediators to allow the body to adapt and cope with the stressor. 1 As well as the production of glucocorticoids (ie, cortisol in humans, corticosterone in rats andmice)followingactivationofthehypothalamic-pituitary-adrenal (HPA)axis,othersteroidsarealsoproduced,suchasdeoxycorticosterone (DOC), progesterone and their metabolites, with elevated levels detected in both the periphery and in the brain following acute stress. [2][3][4] Steroids that exert rapid non-genomic effects in the brain are collectively known as neuroactive steroids, 5 and they can originate from peripheral endocrine organs or are produced de novo via the action of steroidogenic enzymes such as 5α-reductase and 3α-hydroxysteroid dehydrogenase (3α-HSD) expressed within the brain. 6,7 Neuroactivesteroidsexertrapideffectsbybindingto their membrane-bound cognate receptors 8 or to ion channel-associated membrane receptors such as GABA 9 or glutamate NMDA receptors, 10 and contribute to the rapid modulation of the stress response by fine-tuning neuronal signalling 11 (Figure 1). For example, although glucocorticoids are known for their peripheral actions on glucose metabolism, they also bind to central glucocorticoid (GR) and mineralocorticoid (MR) receptors to modulate synaptic transmission,resultinginnegative-feedbackregulationoftheHPA axis. 12,13 Neuroactive steroids that positively modulate GABA A receptors, such as the progesterone metabolite, allopregnanolone, and the corticosterone metabolite, tetrahydrodeoxycorticosterone (THDOC), are of particular interest because their increase in the brain following acute stress has been proposed to be necessary for both mounting the stress response, 14 as well as for terminating the F I G U R E 1 Steroid interconversion and their receptors. The steroids investigated in the present study are underlined, shown alongside thereceptorsthroughwhichtheyexerttheirrapideffects.Allneuroactivesteroidsareproducedfrompregnenolone,whichcanbe converted to progesterone by 3β-hydroxysteroidhydrogenase(3β-HSD),orsulphatedtopregnenolonesulphate,whichpositively modulatesNMDAreceptor(NMDA-R)signalling.Progesteronecansubsequentlybeconvertedtodeoxycorticosterone(DOC)then corticosterone,andbothtargetglucocorticoidandmineralocorticoidreceptors(GRandMR,respectively).ProgesteroneandDOCcan also undergo 5α-reductiontorespectivelyformdihydroprogesterone(DHP),whichhasagreateraffinityfortheprogesteronereceptor (PR),ordihydrodeoxycorticosterone(DHDOC),whichmodulatesGABA A receptors (GABA A R).Followingafurther3α-hydroxysteroid dehydrogenase (3α-HSD)reduction,the5α,3α-reducedsteroids,allopregnanoloneandtetrahydrodeoxycorticosterone(THDOC)are formed, both of which are positive modulators of the GABA A R. Progesterone can be converted to androgens by a two-step pathway, catalysedbythe17-hydroxylaseand17,20-lyaseactivitiesofasingleenzyme,cytochromeP450c17.Dihydrotestosterone(DHT),whichhas morepotenteffectsontheandrogenreceptor(AR)comparedtotestosterone,isproducedfromtestosteronevia5α-reduction or from a 'backdoorpathway'withouttheneedfortestosteroneproduction.DHTissubsequentlyreducedtoform3α-androstanediol (3α-diol)and 3β-androstanediol (3β-diol).Although3α-diol potentiates GABA A receptors, 3β-diolislikelytoexertitsrapideffectsthroughtheoestrogen receptor β (ERβ).Lastly,testosteronecanbeconvertedto17β-oestradiol via the actions of aromatase HPAaxisresponsewhenitisnolongerneeded,byincreasinginhibitory signalling. 15,16 Evidencefromhumanandanimalstudiesdemonstratethatexposure to early-life stress can alter the ability of an individual to cope with challenging events later in life. [17][18][19] The vulnerability to stress-related affective disorders, such as anxiety and depression, is also increased and is frequently accompanied by HPA axis dysregulation. 20,21 Inourmodelofprenatalsocialstress,themaleprenatallystressed(PNS)offspringexhibitananxiousphenotype,and bothsexesdisplaygreaterHPAaxisactivationinresponsetoacute stressorssuchassystemicinterleukin(IL)-1β administration and restraint. 22,23 These phenotypes can be considered as a manifestation of "allostatic overload", where physiological systems fail to adapt to perturbations, resulting in vulnerability instead of resilience when faced with stressors. 1 Allostatic overload arises from an imbalance in the production or action of allostatic mediators. 1 Therefore, it is possible that the steroidal milieu of PNS offspring differs from that of non-stressed offspring, particularly so for positive GABA A modulatory neuroactive steroids, given their role in ameliorating anxiousbehaviourandHPAaxisdysregulationassociatedwithprenatal stress. 22,24,25 The reduced production of these steroids, would beexpectedtoleadtoagenerallossofinhibitorytone,andinturn greater and/or more prolonged HPA axis activation that may increase vulnerability to stress-related disorders. 16,26 Indeed, there is evidence that supports the concept of deficits in neuroactive steroid production in PNS offspring. 22,27 The capacity for neurosteroid production is evidently reduced, as reflected by lower levels of 5α-reductasegeneexpressionandactivityinthe brain. 22

| Generation of prenatally stressed rats using the modified resident-intruder paradigm
Female rats (n =20)werehousedwithasexuallyexperiencedmale, and mating was confirmed by the presence of a semen plug in the breedingcagethenextmorning.Femaleratswerereturnedtosamesexgroup-housingaftermating(designatedasgestationalday(GD) 1) until GD16, after which time they were housed singly in IVCs.
Lactating dams (ie, 'residents' for the resident-intruder paradigm; n =10)werealsogenerated,bymatingthem1weekaheadoftheexperimental dams. Lactating 'resident' rats were group-housed until GD20, then transferred to individual open top cages in a separate room in the animal facility.

Halfoftheexperimentalpregnantdams(n=10)wereexposed
to social stress from GD16 to GD20 using the modified resident-intruder paradigm, as described previously. 23

| Acute swim stress
Swim stress was selected as the acute stressor because it has previously been shown to robustly increase the concentrations of neuroactive steroids such as allopregnanolone and pregnenolone. 3

| Tissue collection
Trunkblood(approximately5mL)wascollectedintoachilledtube containing0.5mLof5%(w/v)ethylenediaminetetracaeticacidand keptoniceuntilcentrifugationat4°Ctoseparatetheplasma.Plasma was stored at −20°C until the steroid assays were performed. The brain was rapidly removed and one hemisphere was subjected to gross dissection to remove the frontal cortex, hypothalamus, hippocampus, amygdala and brainstem, as described previously. 4 Dissected brain tissue was frozen immediately on dry ice, and stored at−80°Cuntilsampleprocessingforliquidchromatography-tandem massspectroscopy(LC-MS/MS). Frozen brain samples were weighed prior to sample processing, and all samples from the same brain region were processed and analysed in the same day. The mean ± SD weights of tissues were: 29.9 ±0.9mgforfrontalcortex;46.4± 1.2 mg for hypothalamus; 21.6 ± 0.7 mg for amygdala; 53.6 ± 0.9 mg for hippocampus; and 99.5 ± 2.0 mg for brainstem. Each batch of samples was processed with seven standard calibrants and a zero sample containing only 4% BSA. For standard calibrants and plasma, 100 μL was used. Samples were homogenised in 500 µL of methanol/1% formic acid(FA)andsonicatedonice.Forstandardcalibrantsandplasma, 400 µL of methanol/1% FA was added. Next, 20 μL of deuterated internal standard mix (25 ng mL -1 of each internal standard) was added and homogenates were briefly sonicated. After incubation on dry ice for 30 minutes to aid protein precipitation, homogenates were centrifuged for 10 min (13 000 gat4°C).Thesupernatantwas  Table S2). Values obtained using this method were comparable to those reported by us previously 4 and other studies utilising gas chromatography-MS 32,34 or LC-MS. [35][36][37] Concentrations of samples were extrapolated and converted to ng mL -1 (forplasma)ornormalisedto the wet weight of the tissues (ng g -1 forbraintissues).

| Statistical analysis
A two-way ANOVA was used to analyse the effects of prenatal stressandacuteswimstresswithineachsex.Statisticswerecarried out using prism, 6.0(GraphPadSoftwareInc.,SanDiego,CA,USA).
Theresultsofthetwo-wayANOVAarereportedaboveeachgraph, andindividualdatapointsforeachexperimentalgroupareoverlaid on the bar graphs representing the group mean ± SEM. Post-hoc testing using Fisher's least significance difference test was carried ou. P < 0.05 was considered statistically significant. Post-hoc test results are also reported. To further investigate the presence of any sexdifferences,three-wayANOVAs(withsex,acutestressandprenatal stress as main factors) were also carried out (see Supporting information,TableS3).

| Corticosterone
There was a main effect of acute stress on plasma corticosterone concentrations, with significantly greater levels in both control and

| DOC
TherewasamaineffectofacutestressonplasmaDOCconcentrations, with significantly greater levels in the swim-stressed groups compared to the basal groups in both the control and PNS groups, in offspring.Datawereanalysedbytwo-wayANOVAwithF values and P values reported above each graph. A significant main effect of acute stresswasobservedinallregionsinvestigated,forbothmalesandfemales.Alloftheswim-stressexposedgroupshadsignificantlygreater corticosterone concentrations in the plasma and in the brain regions analysed compared to basal groups (*P < 0.05, **P < 0.01, ***P <0.001). Therewasamaineffectofprenatalstressobservedonlyinthemalebrainstem(K),wherecorticosteroneconcentrationsweresignificantly greater in male PNS swim-stressed rats compared to male control swim-stressed rats (##P <0.01).Therewerenosignificantacute stress × prenatal stress interactions in the plasma or in any of the brain regions investigated. Note the difference in the scale of the yaxes betweenthesexes.NS,notsignificant;n= 9 or 10 rats per group Prenatal stress: n.s. Prenatal stress: n.s.   In the male brainstem, an additional main effect of prenatal stress F I G U R E 4 Deoxycorticosterone(DOC)concentrationsintheplasmaandbrainregionsofmaleandfemalecontrol(Con)andprenatally stressed(PNS)offspring.Asignificantmaineffectofacutestresswasobservedintheplasmaandallbrainregionsinvestigated.Allswim-stressedgroupshadsignificantlygreaterDOCconcentrationscomparedtobasalgroups(*P < 0.05, **P < 0.01, ***P <0.001),exceptforthe malebrainstem(K),whereonlyPNSgroupsshowedsignificantincreasesinDOCconcentrationswithswimstress.Amaineffectofprenatal stresswasdetectedforthemalebrainstem(K)andfemalehippocampus(J).AlthoughPNSswim-stressedmaleshadgreaterbrainstemDOC concentrations compared to male control swim-stressed rats (##P <0.01),hippocampalDOCconcentrationsinfemaleswim-stressedrats were lower compared to female control swim-stressed rats (#P <0.05).Therewerenosignificantacutestress× prenatal stress interactions in the plasma or in any of the brain regions investigated. Note the difference in the scale of the yaxesbetweenthesexes.NS,notsignificant. n = 9 or 10 rats per group was observed, with the PNS group having significantly greater swimstress-inducedDOCconcentrationscomparedtothecontrol swim-stressed group ( Figure 4K). An effect of prenatal stress was also observed in the female hippocampus ( Figure 4J), where PNS swim-stressedfemaleshadsignificantlylowerDOCconcentrations compared to the swim-stressed control group. There were no differ-encesinDOCconcentrationsbetweencontrolandPNSanimalsin anyoftheotherbrainregionsexamined(Figure4).

| DHDOC and THDOC
There was no effect of acute stress nor prenatal stress on plasma DHDOC concentrations in either males ( Figure 5A) or females (Figure5B).Amaineffectofacutestresswasobservedinallbrain regions; however, there was no effect of prenatal stress for any of the brain regions ( Figure 5C-L). Post-hoc testing revealed that, in most brain areas, both control and PNS rats had significantly greater F I G U R E 5 Dihydrodeoxycorticosterone(DHDOC)concentrationsintheplasmaandbrainregionsofmaleandfemalecontrol(Con)and prenatallystressed(PNS)offspring.TherewerenosignificantdifferencesinplasmaDHDOCconcentrationsbetweenthefourtreatment groupsineithermales(A)orfemales(B).Significantmaineffectsofacutestresswereobservedinallbrainregionsinvestigated,inboth sexes.Post-hoctestingrevealedthat,inmostbrainregions,swim-stressedratshadgreaterDOCconcentrationsthanbasalrats(*P < 0.05, **P < 0.01, ***P <0.001),exceptforinthefemalefrontalcortex(D)andinthemalehypothalamus(E),whereswimstressdidnotresultin differences in the PNS groups. There were no significant main effects of prenatal stress, nor any acute stress × prenatal stress interactions in the plasma or in any of the brain regions investigated. Note the difference in the scale of the yaxesbetweenthesexes.NS,notsignificant. n = 9 or 10 rats per group Acute stress: n.s.

| Progesterone, DHP and allopregnanolone
In males, there was a significant main effect of acute stress on plasma progesterone ( Figure 7A), DHP ( Figure 8A) and allopregnanolone ( Figure 9A) concentrations. Post-hoc testing revealed that, in males, both the control and PNS swim-stressed groups had significantly greater plasma progesterone ( Figure 7A) and allopregnanolone ( Figure 8A) concentrations compared to F I G U R E 7 Progesteroneconcentrationsintheplasmaandbrainregionsofmaleandfemalecontrol(Con)andprenatallystressed(PNS) offspring.Inmales,asignificantmaineffectofacutestresswasobservedinprogesteroneconcentrationsintheplasmaandineachofthe five brain regions, with both control and PNS swim-stressed males having greater progesterone concentrations compared to the basal groups (post-hoc tests: *P < 0.05, **P < 0.01, ***P <0.001).Infemales,althoughthetwo-wayANOVArevealedasignificantmaineffectof acute stress for the plasma and several brain regions, post-hoc testing revealed that pairwise comparisons were not significant for either controlorPNSgroupsintheplasma(B),amygdala(G)orhippocampus(J).Inthefemalefrontalcortex(D),hypothalamus(F)andbrainstem (L),onlythePNSswim-stressedgroupsshowedsignificantlygreaterconcentrationscomparedtothebasalgroups.Additionally,inthefrontal cortexoffemales(D),anacutestress× prenatal stress interaction was detected, and PNS swim-stressed rats had greater progesterone concentrations compared to control swim-stressed rats (#P <0.05).Nomaineffectofprenatalstresswasobservedintheplasmaorinany ofthefivebrainregions.Notethedifferenceinthescaleoftheyaxesbetweenthesexes.NS,notsignificant.n= 9 or 10 rats per group    In females, there were main effects of acute stress in most brain regions for progesterone (Figure 7), although only in    basal PNS groups and progesterone was significantly greater in the PNS swim-stressed group than in the control swim-stressed group ( Figure 7D). In the hypothalamus and brainstem, proges- levels were also significantly increased by swim stress in the PNS groups, but not in the control groups.

| Pregnenolone
There was a significant main effect of acute stress on pregnenolone

| Testosterone
There were no main effects of acute stress, prenatal stress, nor any interaction effect on circulating or central testosterone concentrations,ineithermalesorfemales(Figure11).

| Sex differences
Three-wayANOVArevealedsignificantmaineffectsofsexonneuroactive steroid concentrations in the plasma and brain for virtually all samples examined (see Supporting information, Table S3).

| D ISCUSS I ON
The present study aimed to determine whether the steroidal profile of prenatally stressed offspring differed from that of control offspring, both under basal conditions and following acute swim stress.
Overall, there were no differences between PNS and control offspring with respect to basal levels of neuroactive steroids in either the plasma or the brain. Following acute swim stress, concentrations ofmostoftheneuroactivesteroidsexaminedincreasedinPNSoffspring in a manner largely similar to the control offspring; however, there were a few exceptions ( Figure 12). In particular, swim stress forexample,alongerdurationfortheresolutionofthecorticosterone response, as described previously, 23   previously reported. 52 Although the central allopregnanolone response to stress does not appear to be compromised in the PNS offspring, there is evidence for modest deficits in DHDOC and THDOC production in both the male and female PNS offspring ( Figure 12). Because DHDOC and THDOC also act as positive allosteric modulators on GABA A R receptor, 48 this observation suggests that inhibitory signalling in PNS offspring may be affected. Although this again would depend on GABA A R receptor dynamics, it is of interest given that dysfunctional GABAergic signalling is implicated in the pathophysiology of psychiatric disorders, for which prenatal stress is a risk factor. 50,53 DespitearoleforTHDOCinregulating HPA axis responses to stress being well established, 14,54 further studies are necessary to determine whether the differences observed in the present study, which are in relatively small, would have any significant physiological impact for HPA axis function.
Nonetheless, a role for THDOC in stress-related disorders warrants further investigation because there is a relative paucity of studiesavailable,despitetheobservationthatTHDOCconcentrations are elevated following acute stress, 3 and that its administration can rescue deficits associated with prenatal stress. 31  is sexually dimorphic, 59 which may contribute to sex differences in neuroactive steroid levels in the brain. The impact of these sex differences on function are unclear; however, given the well established roles of neuroactive steroids in modulating stress responses and mood, 16,26,27 differences in neuroactive steroid concentrations in the brain between males and female may contribute to the sex differences in HPA axis regulation and susceptibility to psychiatric disorders. 29,[60][61][62] Additionally, although sex differences in central and peripheral allopregnanolone levels have been described previously, both under basal conditions and following acute stress, 29 this is the first study toreportsexdifferencesintheTHDOCresponsetostress.Inthe main, females had greater basal and stress-induced THDOC concentrations not only in the brain, but also in the plasma, a finding notseeninmales.Itisthereforelikelythat,similartoobservations for allopregnanolone, 63,64 THDOC sensitivity may be different for males and females. Indeed, differences in the maximal GABA A re-ceptorpotentiationofTHDOCbetweenmaleandfemaleratshave been reported previously. 65 Although it is not possible to determine the origin of the steroids measured in the brain in response to stress in the present study, both changes in secretion by peripheral steroidogenic organs, (such as the adrenal glands and gonads) for some steroids, aswellaslocalsynthesisinthebrainforothers,areexpectedto contribute. For example, DOC is synthesised from progesterone predominantlyintheadrenalcortexandissecretedwithcorticosteroneandprogesteronefollowingstressexposure. 2,48 Moreover, we have previously reported that central concentrations of corticosterone, progesterone and DOC following acute swim stress are positively correlated with circulating levels, 4 indicating that changes in the brain likely reflect changes in production in the periphery. On the other hand, we have also demonstrated that DHP, DHDOC and allopregnanolone concentrations in the brain following acute swim stress do not correlate with circulating levels, 4 supporting the concept of local synthesis of these neuroactive steroids in the brain in response to stress. Indeed, in the present study, although acute stress failed to increase circulating DHDOCconcentrationsinbothsexesandTHDOCinmales,cen-tralDHDOCandTHDOCconcentrationswereincreasedinthese animals. Similarly, allopregnanolone concentrations in the brain were elevated after stress in females, despite no corresponding increase in plasma allopregnanolone concentrations. Taken together, these data support the concept of production of these steroids within the brain, through the actions of 5α-reductase and 3α-hydroxysteroid dehydrogenase. 7 Indeed, acute swim stress up-regulates central 5α-reductase (the rate-limiting enzyme) expression, at least in naïve male rats. 66 which would be expected topromoteDHDOC/DHPproductionandsubsequentlyTHDOC/ allopregnanolone synthesis. Although PNS affects baseline 5α-reductasegeneexpressioninasex-andregion-dependentmanner, 22 it remains to be established whether prior PNS exposure and/or sexaltersthestress-inducedincreasein5α-reductase in the brain, whichcouldpotentiallyexplainthedifferentsteroidresponsesto stress in male and female PNS animals.
In summary, the present study provides a comprehensive examination of nine neuroactive steroids in five brain regions in male and female control and prenatally stressed (PNS) rats. We conclude that, under basal conditions, there are no major differences in circulating and central neuroactive steroid concentrations be-tweencontrolandPNSoffspring.However,therearemodestdifferences in how PNS and control offspring respond to acute swim stress, particularly in terms of DHDOC and THDOC production in the brain, which may indicate deficits in GABAergic signalling inPNSoffspring,andthiscouldcontributetodifferencesinHPA axis responses to stress and altered stress-related behaviours in PNS rats. 22,23 Nevertheless, it is worthwhile highlighting again that plasma corticosterone concentrations were not significantly different between control and PNS offspring at the time point studied; therefore, it remains unclear whether more striking differences in neuroactivesteroidconcentrationsrelatedtoHPAaxisregulation may emerge at different time points, or in response to different stressors. On the other hand, the lack of robust differences may reflect the complicated interaction between prenatal programming and the acute stress response, and also underlines the complex natureofsteroidactioninthebrain,forexamplethefindingthat THDOC is both necessary for mounting and for terminating HPA axisresponsestostress. 14 We propose a more integrative approach to investigating the contribution of steroid action following both prenatal and acute stress, aiming to establish how various allostatic mediatorsmayinteracttoregulatestressresponses.TheTHDOC response to stress is likely to play an adaptive role; hence, differences in production and/or action in the brain may contribute to sex differences in HPA axis function and in the dysregulation induced by prenatal stress and warrants further investigation.

ACK N OWLED G EM ENTS
We thank Daniela Schnitzler (University of Edinburgh, UK) for assistance with the swim stress experiments and Dr Andrew Gill

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflicts of interest.