Sexually dimorphic neuronal inputs to the neuroendocrine dopaminergic system governing prolactin release

Abstract Prolactin (PRL) is a pleiotropic hormone that was identified in the context of maternal care and its release from the anterior pituitary is primarily controlled by neuroendocrine dopaminergic (NEDA) neurones of the arcuate nucleus of the hypothalamus. The sexually dimorphic nature of PRL physiology and associated behaviours is evident in mammals, even though the number and density of NEDA neurones is reported as not being sexually dimorphic in rats. However, the underlying circuits controlling NEDA neuronal activity and subsequent PRL release are largely uncharacterised. Thus, we mapped whole‐brain monosynaptic NEDA inputs in male and female mice. Accordingly, we employed a rabies virus based monosynaptic tracing system capable of retrogradely mapping inputs into genetically defined neuronal populations. To gain genetic access to NEDA neurones, we used the dopamine transporter promoter. Here, we unravel 59 brain regions that synapse onto NEDA neurones and reveal that male and female mice, despite monomorphic distribution of NEDA neurones in the arcuate nucleus of the hypothalamus, receive sexually dimorphic amount of inputs from the anterior hypothalamic nucleus, anteroventral periventricular nucleus, medial preoptic nucleus, paraventricular hypothalamic nucleus, posterior periventricular nucleus, supraoptic nucleus, suprachiasmatic nucleus, lateral supramammillary nucleus, tuberal nucleus and periaqueductal grey. Beyond highlighting the importance of considering sex as a biological variable when evaluating connectivity in the brain, these results illustrate a case where a neuronal population with similar anatomical distribution has a subjacent sexually dimorphic connectivity pattern, potentially capable of contributing to the sexually dimorphic nature of PRL release and function.

dimorphic neural circuits as a result of the differential organisational action of gonadal sex hormones. [1][2][3] Nonetheless, sex as a biological variable remains underexplored in neural circuit mapping studies. 4 Prolactin (PRL) is a non-gonadal pleiotropic peptide hormone primarily released by the anterior pituitary gland. 5 Originally named after its role in lactation, 6 PRL is released in response to innumerous external factors and physiological states. 7 Although some, such as stress, 8,9 are shared by both sexes, others are sexually dimorphic.
The latter ones include the release of PRL in response to nipple stimulation in lactating females, 5 the PRL circadian surges in naturally cycling females, 10,11 and PRL release during copulation in males. 12 In females, PRL is fundamental in organising a series of physiological and behavioural programmes that prepare individuals for motherhood: it decreases female receptivity after fertilisation, 13 and also promotes food intake 14 and maternal behaviour. 15,16 All of these programmes are fundamental to ensure the survival of the progeny.
Bycontrast,theroleofPRLinmale-specificbehavioursislesswell understood, although it has been proposed that PRL release during copulation regulates libido. 17 PRL is primarily produced and released into the bloodstream by specialised cells of the anterior pituitary, the lactotrophs. 18 Its receptor, the prolactin receptor (PRLr), can signal through a multitude of second messenger cascades when activated. 19 The PRLr has widespread expression in both the male and female mouse brain. 20,21 Although PRLr expression is concentrated in the rostral and mediobasal hypothalamus, extra-hypothalamic responses to PRL can also be detected in the medial amygdala, bed nucleus of the stria terminalis, lateral septum and others. 19,21 Some of these extra-hypothalamic regions are known to be important for the modulation of sex-specific behaviours. 22 Several inhibitory and stimulatory factors control the release of PRL, 22 although the most important site of regulation resides in the neuroendocrinedopaminergicneurones(NEDA)ofthemedialbasal hypothalamus, the majority of which are located in the arcuate nucleusofthehypothalamus(ARH). 18 NEDAneuronesinhibittheproduction and release of PRL by lactotrophs via dopamine transmission into the blood in response to PRL itself. Suppression of dopamine discharge by NEDA neurones leads to disinhibition of lactotrophs, which quickly release PRL into circulation, thus establishing a self-inhibitory feedback loop. 7,18,23 SeveralhypothesesexistregardingtheregulationofNEDAneural activity based on the behavioural and physiological conditions thatleadtoPRLreleaseandontheneurochemicalsthatNEDAneurones respond to in ex vivo brain slices. 7,24 Nevertheless, the sources ofnon-localbrain-derivedsignalsareyettobeuncovered.Anotable exception is the suprachiasmatic nucleus, whose neurones synapse ontoNEDA 25 neurones and influence their circadian activity in female rats. 26  or the neurohypophysis (tuberhypophyseal dopaminergic neurones). 28 Besides dopamine, NEDA neurones produce and release several other neurotransmitters and modulators, suggesting an even broader biological role for these hypothalamic neurones. 29,30 Given the role of NEDA in prolactin-related sex-specific physiology and behaviours, we set out to characterise: (i) the number anddistributionofNEDAneuronesinthedorso-medialarcuatenucleus of female and male mice and (ii) the brain regions contributing monosynapticinputstoNEDAneuronesinmaleandfemalemice.To genetically access NEDA neurones, we took advantage of the DAT promoter to first label and quantify NEDA neurones in both sexes.
Secondly, using the rabies virus (RV) monosynaptic tracing system, 31,32 we performed a whole-brain survey of regions that harbor neuronesdirectlysynapsingontoNEDAneurones.TheRVmonosynaptic tracing system consists of a recombinant RV and two different adeno-associatedviruses(AAV)thataredeliveredintoatransgenic mouse. The RV is modified in three important ways: (i) to confine the RV spread to a single synaptic jump, the RV envelope glycoprotein (Gprotein) necessary for the virus' transsynaptic spread is deleted and complemented in transbywayofafirsthelperAAVthatexpresses the G-protein in a Cre-dependent manner via the FLEx switch (ie, the G-protein gene is flanked by inverted loxP sites). The Cre recombinase is expressed from the genome of an appropriate transgenic mouse line. This modification renders the G-deleted RV capable of jumping transsynaptically exclusively from cells that were co-infected withtheG-proteincodingAAV.TheRVdoesnotspreadfurtherthan the first-order monosynaptic partners because these cells do not contain the G-protein necessary to assemble functional RV particles; (ii) to make the RV transfection cell type-specific, an envelope protein

| Animals
Animalswerekeptunderareversed12:12hdark/lightcycle(lights on 20.00 h) with access to food and water available ad lib. in temperature-controlled rooms (22-24°C). The animals were group housed until the first surgical procedure and isolated thereafter.
All mice used were sexually naive adults aged between 3-4 months. All procedures were reviewed and performed in accordance with the Champalimaud Welfare Body and the Champalimaud Foundation Ethics Committee guidelines, and were also approved by thePortugueseNationalAuthorityforAnimalHealth.

| Imaging and data analysis
Brain sections were imaged using an automated slide scanner (AxioScanZ1;CarlZeiss,Oberkochen,Germany).Thelocationsand numbers of labelled neurones were manually determined using the Allen Brain Atlas as a reference (http://atlas.brain-map.org/). All procedures were performed using zen, version 2.0 (Carl Zeiss) and fiji (https ://fiji.sc). To generate figures with representative images, brightness and contrast were adjusted using photoshop (Adobe Systems,SanJose,CA,USA).OnlyanimalstargetedtotheARHand displaying a clear GFP signal were considered (see Results). Manual cell counts were stored as csv files and the data analysis was performed using custom Pyhton2.7 scripts (available upon request). Our analysis is focused on brain regions that showed GFP-positive cells in at least three animals. The cell counts for each animal were normalised by dividing the number of neurones found in each region by the number of cells in the arcuate nucleus of the hypothalamus (ARH) (siteofinjection;seeResults).TheARHwassubsequentlyremoved fromanalysis.Givenoursamplesize(n females = 5, n males = 5), a single region sexual dimorphism is considered statistically significant at alpha = 0.05 when a Mann-Whitney U test resulted in a critical U value equal to or smaller than 2 (two-tailed test), and statistically significantatalpha=0.01whenaMann-WhitneyU test resulted in a critical U value equal to 0 (two-tailed test).

| Mouse arcuate nucleus NEDA populations are sexually monomorphic
The sexually dimorphic functions of PRL are exemplified by lactation as-  Table S1). These 10 animals were selected not only because the RV injection site was located near the ARH, but also because thesebrainsdisplayedrobustARHsomaandMEaxonterminalssignal as well as consistent GFP signal in several other brain regions.  Therefore, in our analysis, we only included brain regions where we observed GFP-labelled neurones in at least three animals ( Figure 3C).

Weidentified59regionsreliablycontributinginputstotheARH.
The cell counts are reported as average percentage input contribution

| Monosynaptic input patterns to NEDA are sexually dimorphic
To investigate the possibility that the distribution of monosynaptic inputs observed for male and female dat-cre mice is sexually dimorphic, we performed a Mann-Whitney U test on each region using

| D ISCUSS I ON
The present study reports for the first time whole-brain monosyn-apticinputstotheNEDAoftheARHthatcontrolthereleaseofthe pituitary hormone PRL.

| Distribution of NEDA neurones in mice is monomorphic
Recently, much effort has been dedicated to unravelling the neuronal circuitry underlying sexual dimorphic behaviours, as a result of the importance of these for the reproductive success of individuals and survival of the progeny. Dimorphisms in behaviour have been mostly associated with quantitative differences in cell numbers. 39 Bycontrast,here,wehaveshownthatthenumberanddistribution ofDat-positiveNEDAneuronesinmiceissimilarbetweenmalesand females, in accordance with previous studies, 37 implying that the dimorphism in PRL physiology, such as higher PRL concentration in the blood of females, 5 and associated behaviours must arise from other sources.Still,itislikelythattheoutputofmouseNEDAneuronesis sexuallydimorphic:intherat,forexample,femaleNEDAneurones produce higher levels of dopamine than their male counterparts. 34

| Identification of long-range input regions to NEDA of the ARH of male and female mice
TheARHisanextensivelystudiedareaofthebrainindiversecontexts and has a highly diverse cell population. 40 In our study, the ARH had consistently high numbers of GFP-positive only (nonstarter) cells, thus attesting to the high degree of intrinsic connectiv-itywithintheARH. 29 Wefocusedouranalysisoninputsoriginating outsidetheARHbecausetheRVsystememployedhereisnotadequate to perform local circuitry input mapping. To perfoem a local circuitry study, a different version of the RV system that employs a mutantTVAwithreducedtransfectionefficiencyisrecommended 41 to allow labelling of more sparse local inputs.
In female mice, the area contributing the most inputs to the ARHistheDMH,whereas,inmales,itisthePVH.Inbothsexes, both of these areas contribute a substantial amount of inputs to NEDA neurones. Neuronal activity in the DMH and the PVH has been implicated in lactation. 42 Weproposethatconcomitant activation of these brain regions might result in direct signalling ontoNEDAneuronesandthusaffectPRLrelease;thenatureof the neurotransmitter released by the DMH and PVH remains to be clarified. Regarding males, the role of these brain areas in the context of the regulation of NEDA neurones is currently a mystery.
In rodents, dimorphisms have been identified in the number of neurones within brain regions or their projections that are relevant for dimorphic social behaviours. 39 Indeed, in our study, we detected significant sexual dimorphisms in four of the six areas classically defined as the mammalian social brain network 22  in the present study are also PRL-responsive. 21 This suggests the existence of a feedback mechanism in the brain influencing the control of PRLreleasebyPRL-responsiveregionsthatcontactNEDAneurones,in additiontodirectPRLactionontoNEDAneurones.

| The role of PRL in male sexual behaviour
Even though PRL is paramount for the regulation of sexual behaviour and sex-specific behaviours, little is known about the role of this hormone in male physiology. Curiously, we detected inputs that consistently appeared in males but not in females. One of these input areas, the suprafascicular nucleus (SPF), has been reported as having direct projections to dynorphin-positive neurones of the ARH in non-lactating females 48 but not dopaminergic neurones.
Because the SPF has been implicated in the control of ejaculation, 44 it is tempting to speculate that the projection from the SPF onto NEDA neurones controls the release of PRL during copulation. PRL release during sexual behaviour could be involved in the priming of the male brain for paternity, as appears to be the case in females. 15 This view is in accordance with the observation that sexually experienced males have decreased rates of infanticide. 49

| Caveats of the monosynaptic rabies tracing technique and further work
Despite the specificity and potential of the rabies monosynaptic tracing system, 31 there are some limitations to this methodology, in particular, the completeness of input coverage to a given population of starter cells, as well as in the types of synapses the RV can traverse. 32 Therefore, we do not claim to have uncovered all the brain regionssynapsingontoDat-positiveNEDAneurones,butratherthat the regions we have uncovered are reliable monosynaptic inputs outsidetheARH.
The main challenge of the present study is the heterogeneity ob- Average input distribution of green fluorescent protein (GFP)-positive cells by sex. Each brain region was averaged across sex and converted to a percentage value of total inputs by sex. To detect significant differences between males and females in each brain region, we performed a two-tailedMann-WhitneyU test and accepted significance when U≤2(*α = 0.05) and U=0(**α = 0.01). Ten regions were found to significantly contribute more inputs in males than in females: the anterior hypothalamic nucleus, the anteroventral periventricular nucleus, the medial preoptic nucleus, the paraventricular hypothalamic nucleus, the posterior periventricular nucleus, the supraoptic nucleus, the suprachiasmatic nucleus, the lateral supramammillary nucleus, the tuberal nucleus and the periaqueductal grey  To mitigate this issue, we only considered brain regions where we observed GFP-positive neurones in at least three of the animals.
In addition, the choice of Dat as marker might result in incomplete coverage of all dopaminergic neurones participating in PRL regulation. For example, a brain area reported in female rats that synapsesontotyrosinehydroxylase-positiveneuronesoftheARHis the intergeniculate leaflet of the lateral geniculate, 38 an area related to circadian regulation. However, in our study, we did not reliably find GFP-positive neurones in this area in females, only in two male animals (see Supporting information, Table S1). It is possible that this particulardiscrepancyisasaresultofthefactthatnotalloftheARH Th-positive neurones are Dat-positive. 37 Several experiments can be performed as a follow-up to ensure that the regions reported as monosynaptic inputs are indeed synaps-ingontoNEDAneurones,startingwiththeinjectionofanterograde traveling viral particles in the brain regions identified in this study.

| CON CLUS IONS
The present study has laid the necessary grounds for the future detailed investigation of PRL function in a broad variety of contexts, ranging from sexual behaviour to maternal care and drug development efforts adequate to both male and female physiology. Importantly, as far as we know, this is the first instance where the rabies monosynaptic tracing system was used to unveil the sexual dimorphism of synaptic inputs into a neuronal population, emphasising the importance of considering sex as a variable in studies of neuronal connectivity. Finally, despite the clear existence of dimorphisms in behaviour, there are very few cases where the underlying circuitry has been identified and examples of dimorphisms in connectivity onto a monomorphic population are rare. Therefore, besides the contribution to our knowledge regarding the regulation of PRL physiology and associated in behaviours, the present study also puts forward an interesting candidate circuit for interrogating basic questions related to sex-specific development and wiring of neuronal circuits.

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