Reorganization of perineuronal nets in the medial Preoptic Area during the reproductive cycle in female rats

Perineuronal nets (PNNs) are aggregations of extracellular matrix associated with specific neuronal populations in the central nervous system, suggested to play key roles in neural development, synaptogenesis and experience-dependent synaptic plasticity. Pregnancy and lactation are characterized by a dramatic increase in neuroplasticity. However, dynamic changes in the extracellular matrix associated with maternal circuits have been mostly overlooked. We analyzed the structure of PNNs in an essential nucleus of the maternal circuit, the medial preoptic area (mPOA), during the reproductive cycle of rats, using the Wisteria floribunda (WFA) label. PNNs associated to neurons in the mPOA start to assemble halfway through gestation and become highly organized prior to parturition, fading through the postpartum period. This high expression of PNNs during pregnancy appears to be mediated by the influence of estrogen, progesterone and prolactin, since a hormonal simulated-gestation treatment induced the expression of PNNs in ovariectomized females. We found that PNNs associated neurons in the mPOA express estrogen receptor α and progesterone receptors, supporting a putative role of reproductive hormones in the signaling mechanisms that trigger the assembly of PNNs in the mPOA. This is the first report of PNNs presence and remodeling in mPOA during adulthood induced by physiological variables.


Results and Discussion
perineuronal nets change along the female reproductive cycle. A profuse WFA-positive signal for difusse ECM and PNNs was found in different brain structures, a fact previously reported 15,17,18,33 . One of the most prominent WFA-positive regions was the neocortex, showing the characteristic layered pattern related to its intrinsic laminar organization (see Supplementary Fig. S1) and a Golgi like staining pattern on a subset of neurons (see Supplementary Fig. S1) 33 . Most of the neocortical PNNs stained neurons showed the stellate morphology that characterizes the GABAergic interneuron phenotype.
Neural plasticity in adult females is strongly influenced by gonadal hormones 34,35 . The mPOA is one of the most sensitive brain regions to this endocrine factors, having the highest density of steroids receptors 36,37 . Therefore we assessed the ECM expression in this area along the female reproductive cycle. Contrary to the high PNNs expression detected in the cortex, no WFA staining was detected at the mPOA of cycling females (see Supplementary Fig. S1).
As shown in Fig. 1a-c, neither PNNs nor diffuse staining were detected in diestrus, proestrus, and estrus phases. A small set of scarce perisomatic PNNs stained neurons, together with some interneuronal diffuse signal, was detected in the mPOA of male rats (Fig. 1d). However, when WFA staining was assessed on GD18, we found a massive expression of ECM at mPOA localized predominantly in the dorsomedial nuclear region, as well as PNNs structures (Fig. 1e).
The results regarding virgin females and males are in accordance with previous studies, as Seeger et al. 31 reported the absence of PNNs in the mPOA of male and non-pregnant-female rats (with no specified reproductive status). More recently, Horii-Hayashi et al. analyzed male mice PNNs during postnatal development excluding the mPOA since the absence of those structures in this area was assumed 28 . Our results, showing the high levels of ECM organization in pregnant rats, constitute the first report of a different pattern of ECM organization during adulthood, induced by a natural physiological condition.
In addition to its key role in the expression of parental and sexual behavior [38][39][40] , the mPOA is involved in neuroendocrine functions as the secretion of gonadotropin-releasing hormone, among other endocrine mediators 41 . As pointed out by Horii-Hayashi et al., in cycling females the mPOA would require high levels of neural plasticity, both structurally and synaptically, to adequately respond to the hormonal levels in the circulation, a fact that might explain the absence of PNNs 26 . Thus, the high organization level of the PNNs during pregnancy, in contrast with its absence in cycling rats, could be the result of a transition from a cyclic pattern of activity to a tonic mode, where the gradually and sustained increase in pregnancy hormones levels allows the maintenance of pregnancy and its adaptive changes. the mpoA WfA staining is associated with a pnns expressing neuronal subpopulation. High magnification images of WFA-DAB stained mPOA (Fig. 2) show that the increase in the ECM staining detected during gestation is mainly due to the appearance of PNNs expressing cells, rather than to an interneuronal diffuse component. In-depth analysis showed a Golgi like staining pattern that is consistent with neuronal cells, showing cellular processes similar to dendrites and axonal initial segments (Fig. 2a). Moreover, the detection of WFA with fluorescent coupled streptavidin showed a complex PNNs structure composed by holes, as those classically described in honeycomb-like neocortical PNNs organization 13,42 , combined with a distributed punctuated pattern (Fig. 2b). This peculiar PNNs organization requires further analysis, exploring the fine structural organization, the molecular components involved, and its relation with the synaptic contacts distribution.
To further confirm if the WFA Golgi like pattern is associated with neurons, semithin sections of previously WFA stained slices were made using toluidine blue to counterstain sections. As showed in Fig. 2c, WFA-associated DAB stained PNNs envelop neuron shaped cells that exhibit big heterochromatic nuclei with prominent nucleolus and cytoplasm containing Nissl bodies. Interestingly, we found neighbor DAB unstained cells, indicating that the expression of PNNs is associated with a subset of neurons within the regions of the mPOA here analyzed.
This neuronal subtype specificity matches the complexity and structural heterogeneity of the mPOA which contains cells expressing different neurotransmitters and neuropeptides 43,44 . Previous reports showed that 50 to 95% of neurons in the mPOA and ventral bed nucleus of the stria terminalis (vBNST) synthesize the inhibitory neurotransmitter GABA 43,45 . Interestingly, recent data showed that in maternal mice and following a pup exposure procedure, most Fos-expressing neurons in the mPOA and in the vBNST are inhibitory GABAergic neurons, while very few appear to be glutamatergic 43 . On the other hand, while PNNs could be found surrounding  various cell types in different brain regions [46][47][48] , they mainly enwrap a subpopulation of inhibitory neurons, the fast-spiking parvalbumin-expressing (PV+) neurons in the neocortex 46,[49][50][51] . It is then tempting to speculate that the neurons in mPOA expressing PNNs are GABAergic PV+ cells.
We can postulate that PNNs could be mainly associated with GABAergic neurons that are part of the maternal circuit. To further sustain this hypothesis, we are currently working to address the neurochemical identity of the PNNs associated neurons in the mPOA and its projection targets as well as its activations in different behavioral contexts.
Despite the role of the mPOA in the establishment and maintenance of maternal behavior following parturition 5,52 , few data are available regarding whether different subregions within this area might be more relevant 43 , as shown in other behaviors 53 . For this reason, we aimed to deeply characterize the distribution of PNNs expressing neurons by performing a regional analysis of serially sectioned brains. We found that the WFA neuronal subpopulation was not homogeneously distributed along mPOA. Even more, antero-posterior analysis on GD21 showed a particular PNNs spatial distribution starting with few labeled neurons in the anterior region, increasing in medial mPOA with a maximum in the caudal portion of the nucleus (Bregma −0.92 to −1.08 mm) (Fig. 3).
How the population of PNNs expressing neurons is involved in the neural circuit responsible for the emergence of maternal behavior is still unclear. In fact, most of these neurons are located in the medial portion of the dorso-ventral mPOA axis, a location identified by Tsuneoka et al. 43 following excitotoxic lesions as crucial for maternal behavior expression. Surprisingly, the medial mPOA portion is not the region where cFos expression occurs in response to pup exposure 43 . In order to understand this complex organization, it is necessary to perform a detailed anatomical study analyzing the antero-posterior distribution of PNNs together with the labeling of the different neurotransmitters involved and the cFos expression in response to maternal experience.

the pattern of expression of pnns in the mpoA dynamically changes throughout pregnancy.
To investigate the temporal pattern of PNNs organization in the mPOA during the pregnancy, we analyzed WFA reactivity on different stages of gestation (GD10, GD14, GD18, and GD21). We found that the expression of PNNs in this area was not constant during the gestation, but rather showed a gradual increase over the progression of this period, starting close to GD10. As showed in Fig. 4, on GD10 few PNNs surrounded neurons are observed and no interneuronal diffuse signal was detected. However, on GDs14 and 18 the expression increases, reaching the maximum extension and intensity of PNNs (with a small amount of interneuron diffuse staining) before parturition (GD21, Fig. 4). It is worth noting that the first PNNs expressing neurons appear in the ventromedial edge of its final distribution, progressing later towards the dorsolateral edge, ending in an eccentric ring shape (GD21, Fig. 4).
This variable pattern of the ECM expression could be attributed to the heterogeneity and complexity of the mPOA and its regions, which includes several nuclei with different subpopulations of neurons expressing specific sets of neurotransmitters 49 , regulated by several and diverse factors.

Hormone simulated-pregnancy induces pnns expression in the mpoA of ovariectomized rats.
The dynamic changes in the expression of PNNs observed resemble the temporal profile of the hormonal levels during gestation. In rats, circulating estrogens are low for the first two weeks of gestation gradually increasing until parturition, while plasma progesterone slowly rises throughout the first two weeks of pregnancy, reaching a peak in the third week and declining abruptly before parturition. On the other hand, pituitary prolactin is secreted in two daily surges until mid-pregnancy when its secretion is inhibited by placental lactogens. This inhibition terminates towards the end of gestation with a large prepartum surge of prolactin [54][55][56] .
This suggests that the prolonged exposure to estrogen, progesterone and/or prolactin during this period could be related to the assembly of the ECM in the mPOA. To explore this possibility, we assessed WFA reactivity in ovariectomized (OVX) rats (a) following a hormonal simulated pregnancy treatment with estrogen and progesterone 57 ; (b) treated with estrogen + progesterone + cabergoline (to inhibit prolactin secretion); and (c) to each steroid separately. In accordance with the absence of PNNs labeling in cycling females, OVX oil-treated control rats did not show any label of WFA in mPOA (Fig. 5). Interestingly, following the hormone-simulated pregnancy treatment with estrogen and progesterone, the mPOA of OVX females showed PNNs assemblies around cell bodies (Fig. 5). When OVX rats were treated only with estradiol no signal of PNNs could be detected. Progesterone administration on the other hand, induced a few perisomatic PNNs labeled neurons, suggesting a complementary www.nature.com/scientificreports www.nature.com/scientificreports/ effect of both gonadal hormones. As shown in Fig. 5(e), the inhibition of a putative gonadal steroid-induced prolactin secretion, due to the dopaminergic agonist cabergoline, prevented the complete expression of PNNs induced by the treatment with estrogen and progesterone (Fig. 5b).
These findings indicate that steroid hormones are capable of inducing the assemblage of the PNNs in the mPOA, and that the effects could be partially attributed to an increase in the secretion of prolactin. Interestingly, although the main gonadal steroid hormones of both the estrous cycle and pregnancy are the same, the changes in the assembly of the PNNs were only observed during the pregnancy or following hormone simulated-pregnancy treatment and not during the estrous cycle. Taken together these results show that this structural plasticity is the consequence of the high and sustained hormonal levels observed during the gestation period and suggest that this phenomenon is specific of this reproductive cycle stage.
On the other hand, it is worth noting that the expression of PNNs in the mPOA was higher in pregnant rats than in hormonal-treated OVX females, suggesting that prolactin and possibly other endocrine factors present during gestation (i.e. placental lactogens, growth factors, chorionic gonadotropin) could be involved in the full expression of PNNs seen in the mPOA of pregnant rats before parturition 58 . Moreover, this difference could be  www.nature.com/scientificreports www.nature.com/scientificreports/ explained by the fact that circulating levels of gonadal steroids induced by the pharmacological treatment differed from the natural variations of the hormonal milieu of gestation.
Recently, Fang et al. reported that the optogenetic activation of mPOA GABAergic neurons that express ERα and project to the ventral tegmental area strongly inhibits non-dopaminergic cells in this area and drive pup retrieval behavior in maternal mice (Fang et al. 63 ). It is also known that estrogen can increase GABA release and reuptake as well as GABA A-receptor expression in the mPOA 59,60 , and that steroid manipulation modulates mRNA levels of glutamic acid decarboxylase (GAD) in this area 61,62 .
Despite the many possible steroid-induced neuronal phenotype alterations, given the central role of GABAergic neurons in maternal behavior 45,63 , its higher number and its association with extracellular matrix components in other regions, we hypothesized that mPOA-PNN+ neurons are GABAergic projecting neurons that express steroid receptors. Exploring this hypothesis will require immunological detection of GAD and ERα and PR, combined with retrograde labeling experiments. In a first attempt to characterize this neuronal phenotype we performed a double-label immunohistochemistry to detect ERα and WFA and PR and WFA in the mPOA of GD21rats. pnns associated neurons in the mpoA express estrogen receptor α and progesterone receptors.
As shown in Fig. 6, cell nuclei of PNNs associated neurons show different intensities of ERα label, suggesting variations in protein expression levels ( Fig. 6-left, arrowheads) which are also evident in the nuclei of PNNs negative cells (Fig. 6-left, arrow). Additionally, we observed an abundant dotted neuropilic label probably associated to neuronal or astrocytic membrane processes 64,65 . This neuropilic label was not found in the neocortex where a conspicuous nuclear label was clearly detected (Supplementary Fig. S2).
PNNs associated neurons also express the PR label localized on the cell nucleus, however, neither mPOA neuropilic nor nuclear neocortical label was detected. Interestingly, there was a positive PR nuclear label in the PNNs negative cells (Fig. 6-left, arrow), lacking the variability in intensity observed for the ERα label ( Fig. 6-left, arrowheads). As this profile of expression was observed at GD21, the analysis of temporal dynamic of the expression of both receptors along the gestation period is of great interest.

the pnns expression in the mpoA persists after parturition, gradually fading throughout lactation.
In order to advance in the characterization of the possible role of mPOA PNNs in the establishment and maintenance of maternal behavior and lactation, we aimed at understanding if the high organization of the ECM persists after the birth of the young. Following parturition, the densely packed PNNs observed in mPOA during late gestation begin to fade. As shown in Fig. 7, PNNs are present after parturition (LD2) but with a decreased Figure 6. PNNs associated neurons express estrogen receptor alpha and progesterone receptor. Double labeling using WFA and anti-estrogen-receptor-alpha antibody (ERα + WFA, left) or WFA and anti-progesteronereceptor antibody (PR + WFA, right). In both cases PNN-WFA label was tagged using streptavidin-594 (red), while hormone receptors were tagged using secondary antibodies conjugated with alexafluor-488 (green). In the left panel notice the presence of cell nucleus label for ERα and its variability between different PNNs expressing neurons (arrowheads). Positive nuclei belonging to PNNs negative cells (arrows) are also evident. Interestingly, there is a conspicuous doted label in the neuropilic region that is absent in neocortical neuropilic region ( Supplementary Fig. 2

). Note the presence of PR label nuclei in PNNs associated neurons in the right panel. PR label intensity is less variable compared to ERα label (left panel) and lacks the neuropilic dotted component. A population of PR labeled nucleus colocalizes with PNNs negative cells (arrows).
expression intensity. After that, PNNs undergoe a complex dynamic degradation process along lactation. The WFA expression pattern as well as the reactivity level change gradually, declining from the second day of lactation (LD2), transiently increasing halfway through lactation (Fig. 7, LD7) and resulting in a very diffuse label at the time of postpartum day 22-23 ( Fig. 7, LD22). The transient increase during lactation (LD7) seems to be due to an increase in the diffuse interneuronal extracellular matrix component, rather than an increase of PNNs expressing neurons, explaining how global WFA staining increase occurs even when PNNs are fading. By the end of the lactation period (PPD 22) only a few perisomatic PNNs positive cells are still present with a surrounding interneuronal diffuse staining. These results suggest a complex PNNs degradation process probably involving the action of matrix metalloproteinases 66 , an underlying mechanism for a programmed PNNs degradation, or a passive mechanism of disassembly combined with transient synthesis events.
To further explore the possible role of WFA stained neurons, it would be interesting to address if the early gene expression in mPOA neurons of females exposed to maternal experience is colocalized within the PNNs+ neuronal population here described.
The quantification of WFA staining confirmed the qualitative analysis of PNNs expression throughout the female reproductive period, showing two different processes: (1) the increase in the area occupied by the PNNs  The quantification of WFA staining waxes and wanes along the female cycle. An increase in the area occupied by the PNNs along gestation reaches a maximum immediately before parturition (GD21) followed by a reduction of staining after parturition (LD2) with a subsequent increase towards a maximum (LD7), and a final reduction of WFA staining at the end of lactation period (LD22, Fig. 8). Data are expressed as means ± SEM. Different letters indicate significant differences between groups (p < 0.05, Tukey post hoc test).
Interestingly, a trace of extracellular matrix expression on late postpartum remains up two weeks after weaning (Fig. 8, Post-W), suggesting that once a female undergoes a reproductive cycle a molecular label is generated. How these remanent extracellular components could facilitate subsequent reproductive structural and physiological changes is an interesting question to be answered.
The complexity of PNNs dynamics observed during gestation and the postpartum period suggests a fine regulatory role of the extracellular matrix on neural circuit properties. Experimental evidence points toward a profound remodeling of neuronal morphology and synaptic connections regulated by gonadal steroids 9 . On the other hand, the hormonal changes taking place during pregnancy initiate complex behavioral changes that promote maternal care of the offspring after birth. We also know that maternal behavior dynamically changes along the postpartum period, adjusting to the physiological needs of the pups 1,67 . This adaptation has been attributed to functional modifications in the maternal circuitry, mainly the mPOA, during lactation 1,68 . Although the mechanisms by which this adaptation takes place are not known, a role of the sensory stimulation provided by the pups as well as endocrine factors has been postulated 69 .
We hypothesize that if the high organization of PNNs before parturition is related to the stabilization of the maternal circuitry, the waning in its expression observed during lactation could be related to the dynamically changing expression of the maternal behavior characteristic of this period.
The phenomenon here described opens a new opportunity to explore the cellular and molecular mechanisms involved in PNNs assembly and disassembly during adulthood, raising many interesting questions: What are the cellular components involved in the secretion of the extracellular matrix? Which are the molecules responsible for catalyzing the PNNs assemblage around specific neuron populations and, which signals are driving this process? What is the role of metalloproteinases? What is the relation between PNNs and the functional role of mPOA during maternal behavior?

Methods
Animals and housing. Wistar rats (Rattus norvegicus) of both sexes three to four-month-old were used.
Animals were housed under controlled temperature (22 ± 1 °C) and humidity (65%) in a 12 h light-dark cycle (lights on at 0400 h) with free access to food and water. All animals were housed in groups of three to four per cage (48 × 33 × 16 cm), except lactating females or mating couples. Mating was achieved by placing a receptive female rat with a sexually active male overnight and that day was considered as GD 0.
Lactating females were individually housed at pregnancy day 21. Following the parturition they were maintained in individual cages with their litters (adjusted to four male and four females on the delivery day). Animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee of the Facultad de Ciencias, Universidad de la República (reference number: 240011-001541-17) and were in accordance with Uruguayan Law N°18,611 for the Care and Use of Laboratory Animals. experimental groups. Experimental groups were designed to reveal the pattern of expression of the PNNs during the different phases of the reproductive cycle: Virgin females at late proestrus (n = 6), estrus (n = 6) and diestrus (n = 6); pregnant females at gestation day (GD) 10 (n = 4), GD14 females (n = 5), GD18 females (n = 5) and GD21 females (n = 7); postpartum females on lactation day (LD) 2 (n = 7), LD7 (n = 6); LD14 (n = 4), LD 22 (n = 6) and two weeks following weaning (Post-W, n = 4). An additional group of males (n = 5) was included.
To determine whether reproductive hormones were responsible for the high organization of PNNs found in GD21 females we compared the expression of PNNs in the following groups: ovariectomized (OVX) vehicle-treated (n = 3), OVX estrogen + progesterone treated (n = 3), OVX estrogen-treated (n = 3), OVX progesterone treated (n = 3) and OVX estrogen + progesterone + cabergoline treated (n = 3) females. ovariectomies and hormonal treatments. Females were anesthetized with 2.0 ml/kg of a solution that contained ketamine HCl (75.0 mg/ml), xylazine (7.5 mg/ml) and acepromazine maleate (1.5 mg/ml) and ovariectomized through a ventral incision. Following a recovery period of two weeks, they were divided into two groups receiving either hormonal or vehicle (corn oil) treatments via subcutaneous injection (between 8:00 and 10:00 h for 21 days). The OVX vehicle-treated group was administered 0.2 ml of corn oil daily for 16 days and then 0.1 ml for 4 days. The OVX hormone-treated groups were treated as follow: OVX estrogen + progesterone group received a low dose of estradiol benzoate (EB, 2.5 μg) combined with a high dose of progesterone (4 mg) dissolved in 0.2 ml corn oil daily for 16 days and on days 17-21 received a high dose of EB (50 μg) dissolved in 0.1 ml of corn oil. The OVX estrogen group received a low dose of EB (2.5 μg) dissolved in 0.1 ml corn oil daily for 16 days and on days 17-21 received a high dose of EB (50 μg) dissolved in 0.1 ml of corn oil. The OVX progesterone-treated group received a high dose of progesterone (4 mg) dissolved in 0.2 ml corn oil daily for 16 days and on days 17-21 received 0.1 ml of corn oil. The OVX estrogen + progesterone + cabergoline group received a low dose of estradiol benzoate (EB, 2.5 μg) combined with a high dose of progesterone (4 mg) dissolved in 0.2 ml corn oil tissue preparation for immunohistochemical studies. Animals were anesthetized with sodium thiopental (100 mg/kg, i.p.) and transcardially perfused with heparinized phosphate buffer saline (PBS), followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB) (pH 7.4). Brains were removed and post-fixed with the same fixative overnight, then cryoprotected with 15% followed by 30% sucrose (in 0.1 M PBS) until they sank, before being stored at −80 °C. Coronal sections (40 µm thick) were obtained using a Leica CM1850 UV cryostat according to the Paxinos Stereotaxic Atlas 70 . Sections were collected and stored in a cryoprotective solution (30% glycerol, 30% ethylene glycol in PB) at −20 °C until use. Slices were rinsed three times PB (pH 7.4) to remove cryoprotective solution.
For PNNs fluorescent labeling, endogenous peroxidase blockade step was suppressed and WFA was detected using Streptavidin AlexaFluor 488 conjugated (1/500 diluted) for 2 hours at room temperature.
For double labeling immunohistochemistry (ERα + WFA and PR + WFA), 30 µm thick slices containing the mPOA region were processed according to the following procedure: (1)  Fluorescent stained sections were mounted on slides and cover slipped using glycerol 80% in PB.
Image acquisition and PNNs signal quantification. Figure images were taken using a Nikon Eclipse E4000 microscope coupled to a Micrometrics 319CU CMOS 3.2 Megapixel Camera. Fluorescent labeled WFA preparations images were obtained using a spectral confocal microscopy (Leica TCS SP5 II). For quantification, images were acquired using Nikon SMZ1000 binocular microscope (3x, NA 0.3) and analyzed using ImageJ software (NIH, https://imagej.nih.gov/ij/). Images were transformed to 8bits in gray scale for quantification purposes and transformed from pixels to micrometers. Regions of interest (ROI) were defined according to the maximum extension area of PNNs positive staining inside the mPOA at GD21 in one slice per brain (−0.92 ~ −1.08 mm Bregma), using a maximum entropy threshold protocol (Image-J). The settled area of 802 μm2 was applied to all images in mPOA region bilaterally. PNNs density was calculated as the mean bilateral PNNs signal density inside the ROI. Statistical analysis was made by two-way ANOVA followed by Tukey post hoc test for independent measures (statistical significance, p = 0.05).

Data availability
The datasets generated during the current study are available from the corresponding author on reasonable request.