Introduction

Sex hormones have a considerable influence on brain functioning and behaviour. There is a strong relationship between low oestrogen levels and emotional disorders such as mood swings, anxiety, and depression in postmenopausal women1,2, whereas oestrogen replacement therapy reduces these symptoms3. Many reports also indicate that the risk of anxiety and depression is higher in perimenopausal women compared to pre- or postmenopausal women even when considering factors such as age, race, and lifestyle among others4,5. Finally, in rodents, reduced oestrogen signalling due to oestrogen receptor β (ERβ) knock-out increases anxiety in females (but not in males)6,7, while ERβ agonists usually induce severe anxiolytic actions8. It is generally known that one of the most important structure responsible for overall regulation of emotional behaviour, especially in the processing of fear and anxiety is the amygdala9,10 and oestrogens reduce neuronal excitability in this brain structure11,12,13 through their receptors14,15, thereby reduced oestrogen signalling as a result of ERβ knock-out is anxiogenic. It was noted that the function of this brain structure is always impaired in anxiety disorders16. This may be associated with changes in neurotransmission that were well-documented in many reports17,18,19.

Furthermore, electrophysiological studies revealed that in mutant mice enhanced anxiety coincides with a reduced threshold for induction of synaptic plasticity in the medial amygdala which may result in excessive responses to naturally harmless stimuli, a hallmark of fear and anxiety disorders7. Finally, pharmacological manipulation of γ-aminobutyric acid (GABA) type A (GABAA) receptors in the amygdala of wild-type mice mimics the consequences of ERβ knock-out7 suggesting involvement of the inhibitory neurotransmitter GABA in this phenomenon. Indeed, it is plausible that abnormal GABA activity may account for the increased anxiety in ERβ knock-out mice20 as it is associated with fear and anxiety as evidenced by various experiments21,22,23. Furthermore, recent evidence has shown that calbindin expression, a marker of the largest GABAergic subpopulation, is strongly reduced in the amygdala of these mice24 and this decrease has a potential to reduce interneuron firing and threshold for synaptic plasticity25,26. Indeed, oestrogen interacts with neuromodulatory systems including dopaminergic and cholinergic systems20,27,28,29, which are involved in controlling fundamental behaviours by interacting with broad areas of the brain30,31. These systems project to many areas, including the cortex, forebrain, striatum, hippocampus, and also amygdala32,33,34,35,36.

However, to the best of our knowledge, there are no publications to date explaining the relationship between oestrogen and other neurotransmitter systems e.g., dopaminergic, and cholinergic. This is extremely important because the amygdala is densely packed with dopaminergic innervation from the ventral tegmental area and substantia nigra37. It is known that dopamine (DA) release increases in the basolateral amygdala under stressful conditions, and an activation of dopamine receptors (DAs) in this region is crucial to generate fear-related memory38,39,40. Furthermore, some reports have pointed to a role for the DA in anxiety41,42. However, the action of DA on the amygdala circuits is rather unclear and depends on the specific populations of innervated neurons and receptors used. Generally, DAs have been classified in two types, D1-like receptors (DA1) and D2-like receptors (DA2)43. The activation of DA1 increases neuronal activity whereas activation of DA2 restrains neuronal activity44,45, and the interaction between DA1 and DA2 is necessary to express most dopaminergic-related behaviours46,47, e.g., DA1 and DA2 are important in mediating anxiety, especially visible in an elevated plus maze test48,49. It should be emphasized that in natural conditions, oestrogens strongly modulate dopaminergic neurotransmission. It may increase or decrease DA activity e.g., by its degradation, reuptake, and recover, also by upregulating dopaminergic receptors27 or by reducing the affinity of the DA transporter (DAT)50. Furthermore, many of the effects, have been shown to be mediated by ERβ signalling51,52,53,54,55. It is well-known that the amygdala also receives a dense cholinergic innervation from the ventral pallidum and substantia innominata of the basal forebrain56. Muscarinic receptors (AChRM) can inhibit or excite postsynaptic neurons57, whereas nicotinic receptors (AChRn) directly excite postsynaptic neurons58. The acetylcholine (ACh) releases in the amygdala plays a role in anxiety disorders59. Previous studies in rodents have shown an increase in anxiety, specifically in depressed mood and co-morbid anxiety states, as a result of increased AChRn activity60. Especially, α7 antagonist infusion or knock-down has been associated to an anxiolytic behaviour as a conclusion of light–dark box and tail suspension test61, and high concentrations of this receptor in the amygdala can play a significant role in the modulation of changes in amygdala function62. On the other hand, AChRM1 seems to be significant in fear extinction because direct infusion of oxotremerine (the muscarinic agonist) into the basolateral amygdala induces an enhancement in fear extinction learning63. Notably, oestrogen has also been demonstrated to shape cholinergic neurotransmission in the brain, often via ERβ64,65. The oestrogen increases choline and choline acetyltransferase reuptake in the hippocampus and frontal lobe66, but no data are available for the amygdala.

Altogether, the amygdala is strongly modulated by dopaminergic and cholinergic fibres, which activate both glutamatergic and GABAergic neurons and thereby have a considerable influence on the amygdala inhibitory/excitatory balance and processing. Further, in normal conditions dopaminergic and cholinergic neurotransmission is heavily modulated by oestrogen-signalling, so this modulation could be altered in ERβ knock-out mice due to ERβ deficiency. To test this hypothesis, a single immunofluorescent staining was performed to assess the presence/absence of quantitative changes in the expression of tyrosine hydroxylase (TH), DAT, DA1, DA2, acetylcholinesterase (AChE), vesicular acetylcholine transporter (VAChT), AChRM1 and AChRα7 in the amygdala of ERβ knock out female mice. To make the text easier to read, the immunoreactive elements (somata, fibres and neuropil) expressing these markers will be uniformly described as TH+, DAT+, DA1+, DA2+, AChE+, VAChT+, AChRM1+ and AChRα7+, respectively. Additionally, the hierarchical clustering and functional importance of mutual interactions of these proteins was also studied using Gene Ontology67.

Results

The results of this present study show that the expression of the main dopaminergic markers, including TH, DAT and DA2, is significantly elevated in the amygdala of ERβ knock-out mice compared to matched controls, and the only exception constitutes DA1 which is not affected by ERβ loss. Furthermore, although the expression of cholinergic markers such as AChE and VAChT does not differ in ERβ knock-out mice compared to matched controls, the density of cholinergic receptors such as AChRM1 and AChRα7 is increased in mutant mice.

Dopaminergic markers in the amygdala of ERβ knock-out mice

Both, the characteristics and distribution of TH (Figs. 1A and 2) immunoreactivity were comparable in ERβ−/− and ERβ+/+ mice, similar for the characteristics and distribution of DAT (Figs. 1B and 3), and they were consistent with previous results in the rodent amygdala37,68,69,70. For example, as in the mouse and rat, the immunoreactivity of TH (Fig. 2) and DAT (Fig. 3) were observed in fibres and puncta (Figs. 2A–B′ and 3A–B′). Furthermore, in both, ERβ−/− and ERβ+/+ mice, the density of TH+ elements was very high in the CE, high in the BL, and from moderate to low in the rest of the amygdala (Figs. 1A and 2A,A′). In the case of DAT, a high density of immunoreactive elements was observed in the CE, moderate in the BL, while in the remaining nuclei it was at a very low level (Figs. 1B and 3A,A′). Detailed densitometric studies revealed that the volume density of TH+ and DAT+ elements was heavily elevated in all amygdala regions of ERβ−/− mice when compared to matched controls (Fig. 1A,B and Table 1). The increase in TH density was generally in the range of 32–35%. The only exception was the CE with a value of ~ 25%. The increase in DAT density ranged from 28 to 30%, except for the LA with a value of ~ 25%.

Figure 1
figure 1

The densitometry of dopaminergic markers such as tyrosine hydroxylase (TH), dopamine transporter (DAT), dopamine D1-like receptor (DA1) and dopamine D2-like receptor (DA2) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. Note that the volume density of TH (A), DAT (B) is significantly elevated in ERβ−/− mice when compared to ERβ+/+ mice whereas values for DA1 (C) do not differ in both mice lines but for DA2 (D) is significantly elevated in ERβ−/− mice. Additionally, note that the automated cells counting for DA1 (E) are not affected but DA2 (F) is significantly increased due to ERβ deficiency. Data are expressed as a box-and-whiskers plots, with the "box" depicting the median and the 25th and 75th quartiles, and "whiskers" showing 5th and 95th percentile. ** (p ≤ 0.01) and *** (p ≤ 0.001) indicates statistically significant differences between the studied groups of mice (ERβ−/− and ERβ+/+ mice) analysed by Student's t-test. Elements—immunoreactive somata, fibres and neuropil. LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus, CO—cortical nucleus, ME—medial nucleus and CE—central nucleus.

Full size image
Figure 2
figure 2

Representative colour photomicrographs illustrating the staining pattern of tyrosine hydroxylase (TH) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note increased density and signal intensity of TH in ERβ−/− mice (A,B) when compared to ERβ+/+ littermates (A′,B′). LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus and CE—central nucleus. The scale bar applies to all microphotographs, and it corresponds to the length of 200 µm in (A,A′), and 50 µm in (B,B′).

Full size image
Figure 3
figure 3

Representative colour photomicrographs illustrating the staining pattern of dopamine transporter (DAT) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note increased density and signal intensity of DAT in ERβ−/− mice (A,B) when compared to ERβ+/+ littermates (A′,B′). The arrows indicate areas where fiber swellings are located. LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus. The scale bar applies to all microphotographs, and it corresponds to the length of 200 µm in (A,A′), and 50 µm in (B,B′).

Full size image
Table 1 Statistical analysis of studied markers between ERβ−/− and ERβ+/+ mice in the amygdala.
Full size table

As was observed in previous studies in the rodent amygdala 15,49,71,72 in both mouse lines, DA1+ and DA2+ signal was mainly observed in soma and immunoreactive puncta, although some clearly depicted fibers were also present (Fig. 4). The density of DA1+ elements was heterogenous in the amygdala, and the highest it was in the BL, while in the CE it was the lowest (Fig. 1C). In the case of DA2, the highest density of immunoreactive elements was observed in the BL, whereas in the BM it was the lowest (Fig. 1D). Although the statistical analysis showed no significant differences in the volume density of DA1+ elements between ERβ knock-out and wild-type mice (Fig. 1C and Table 1), the volume density of DA2+ elements was elevated in the mutant mice (Fig. 1D and Table 1). Furthermore, higher number of immunoreactive neurons was present in DA1 than DA2 preparations (Fig. 1E,F). The DA2+ density increase was in the range of 11–13%. It is worth mentioning that apart from densitometric analyses, the neurons endowed with DA1 and DA2 were also additionally counted. This analysis revealed that the number of DA1+ neurons did not differ between ERβ knock-out (Fig. 4A) and wild-type mice (Figs. 1E, 4A′ and Table 1) while the number of DA2+ cells was elevated in mutant mice (Figs. 1F, 4B and Table 1).

Figure 4
figure 4

Representative colour photomicrographs illustrating the staining patterns of dopamine D1-like receptor (DA1) and dopamine D2-like receptor (DA2) in the amygdala basolateral nucleus (BL) of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note a lack of differences in DA1 density between ERβ−/− mice (A) and ERβ+/+ littermates (A′). Note also increased density of DA2 in ERβ−/− mice (B) when compared to ERβ+/+ littermates (B′). The scale bar applies to all microphotographs, and it corresponds to the length of 50 µm.

Full size image

Cholinergic markers in the amygdala of ERβ knock-out mice

The staining patterns for AChE (Figs. 5A and 6) and VAChT (Figs. 5B and 7) were quite similar in ERβ−/− and ERβ+/+ mice, and they did not differ from the observations reported in the adult mouse and rat brain 69,73,74,75,76. For example, the immunoreactivity of both AChE (Fig. 6) and VAChT (Fig. 7) was present in fibres and puncta, and many of these fibres were thin and characterised by swellings (Figs. 6A–B′ and 7A–B′). However, in AChE preparations, the fibres were tightly packed while those in VAChT preparations were much less densely arranged (Figs. 6A–B′ and 7A–B′). Furthermore, in both, ERβ−/− and ERβ+/+ mice, the density of AChE+ and VAChT+ elements was high in the BL, moderate in the LA, BM and CO, while in the ME and CE it was at a quite low level (Figs. 5A,B, 6A,A′ and 7A,A′). Finally, all these similarities in the staining patterns were additionally confirmed by densitometric analysis, which showed no significant differences in the volume density of AChE+ and VAChT+ elements between ERβ−/− and ERβ+/+ mice (Fig. 5A,B and Table 1).

Figure 5
figure 5

The densitometry of cholinergic markers such as acetylcholinesterase (AChE), vesicular acetylcholine transporter (VAChT), muscarinic acetylcholine type 1 receptor (AChRM1) and alpha-7 nicotinic acetylcholine receptor (AChRα7) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. Note that the volume density of AChE (A) and VAChT (B) do not differ in both mice lines, but for AChRM1 (C) and AChRα7 (D) is significantly elevated in ERβ−/− mice when compared to ERβ+/+ mice. Additionally, note that the automated cells counting for AChRM1 (E) and AChRα7 (F) are not affected due to ERβ deficiency. Data are expressed as a box-and-whiskers plots, with the "box" depicting the median and the 25th and 75th quartiles, and "whiskers" showing 5th and 95th percentile. ** (p ≤ 0.01) and *** (p ≤ 0.001) indicates statistically significant differences between the studied groups of mice (ERβ−/− and ERβ+/+ mice) analysed by Student's t-test. Elements—immunoreactive somata, fibres and neuropil. LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus, CO—cortical nucleus, ME—medial nucleus and CE—central nucleus.

Full size image
Figure 6
figure 6

Representative colour photomicrographs illustrating the staining pattern of acetylcholinesterase (AChE) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note similar density and signal intensity of AChE in ERβ−/− (A,B) and ERβ+/+ mice (A′,B′). LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus and CE—central nucleus. The scale bar applies to all microphotographs, and it corresponds to the length of 100 µm in (A,A′), and 50 µm in (B,B′).

Full size image
Figure 7
figure 7

Representative colour photomicrographs illustrating the staining pattern of vesicular acetylcholine transporter (VAChT) in the amygdala of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note similar density and signal intensity of VAChT in ERβ−/− (A,B) and ERβ+/+ mice (A′,B′). LA—lateral nucleus, BL—basolateral nucleus, BM—basomedial nucleus and CE—central nucleus. The scale bar applies to all microphotographs, and it corresponds to the length of 100 µm in (A,A′), and 50 µm in (B,B′).

Full size image

AChRM1 and AChRα7 (Fig. 8) immunoreactivity was much more abundant in neuropil (immunoreactive puncta) than in DA1+ and DA2+ preparations, however, neuronal somata endowed with these proteins were also frequently present in both ERβ−/− and ERβ+/+ mice. Similar characteristics were also previously described in some reports on the rodent amygdala15,77,78,79. It is worth noting that, AChRM1 immunoreactivity was much higher than the immunoreactivity for AChRα7 (Figs. 5C,D and 8). However, individual AChRM1+ (Fig. 8A,A′) perikarya were quite difficult to discern, as the level of immunoreactivity in neurons was often equal to that of the surrounding neuropil. On the other hand, on the surface of AChRα7+ (Fig. 8B,B′) perikarya there were frequently densely placed punctate structures. These punctate structures were also present within the neuropil, which was less strongly stained than in AChRM1 preparations. The densitometric studies showed that the density of AChRM1+ immunoreactive elements was the highest in the ME and BM, while the LA and BL showed the lowest density (Fig. 5C). For AChRα7, the ME and CO had the highest density while the LA, BL and BM had the lowest (Fig. 5D). These studies also revealed that the volume density of AChRM1+ and AChRα7+ immunoreactive elements was significantly elevated in all amygdala regions of ERβ−/− mice when compared to matched controls (Fig. 5C,D and Table 1). The density increase was in the range of 7–8% for both AChRM1 and AChRα7. Additionally, a slightly higher number of immunoreactive neurons was present in AChRM1 than in AChRα7 preparations (Figs. 5E,F and 8). Interestingly, the automated counting of neurons endowed with AChRM1 and AChRα7 revealed that the number of the cells did not differ between ERβ−/− and ERβ+/+ mice (Fig. 5E,F and Table 1).

Figure 8
figure 8

Representative colour photomicrographs illustrating the staining patterns of muscarinic acetylcholine type 1 receptor (AChRM1) and alpha-7 nicotinic acetylcholine receptor (AChRα7) in the amygdala basolateral nucleus (BL) of ERβ knock-out (ERβ−/−) and wild-type (ERβ+/+) mice, n = 6 per group. (A,B) ERβ−/− mice. (A′,B′) ERβ+/+ mice. Note increased density of AChRM1 and AChRα7 in ERβ−/− mice (A,B) when compared to ERβ+/+ littermates (A′,B′). Note also strong neuropil staining in AChRM1 preparations (A,A′) which makes it difficult to discern cells (arrows) from the background. The scale bar applies to all microphotographs, and it corresponds to the length of 50 µm.

Full size image

Protein–protein interaction network analysis

The protein–protein interaction (PPI) network contained 9 nodes and 20 edges, each node represented a target (dopaminergic, cholinergic and ERβ markers), each edge represented correlation evidence between 2 targets, and edges of different width represented different types of confidence from the low 0.150 to the highest 0.900. Disconnected nodes were not found, indicating there were direct or indirect interactions among the 9 potential targets with 129 biological processes (BP) significantly enriched, only 7 important BP connected with emotional behaviour processing were selected (p < 0.05, Table 2, Fig. 9). The entire analysis detected 5784 reference publications (PubMed), 3 clusters of local networks (STRING) and 3 Reactome pathways.

Table 2 Network statistics and biological process (Gene Ontology) of the interactions.
Full size table
Figure 9
figure 9

Protein–protein interaction (PPI) networks of potential targets by STRING database (Mus musculus). (A) PPI network of 9 potential targets, contains 9 nodes (proteins) and 20 edges (protein–protein associations). Dotted lines represent edges among 3 clusters (dopaminergic, cholinergic, and oestrogenic). The thickness of the line is proportional to the edge confidence. (B) Pie chart of term associated with emotional behaviour. Each colour represents a particular biological process; please note that studied markers are involved in these processes (A).

Full size image

Discussion

This is the first study to describe changes in the expression of main markers of dopaminergic and cholinergic systems in the amygdala of female mice lacking ERβ which exhibit increased anxiety in relation to their wild-type counterparts as reported in many papers7,52,80. The studies showed that among dopaminergic markers the expression of TH, DAT and DA2 was significantly increased in the amygdala of mice with ERβ deprivation in relation to matched controls whereas an ERβ deficiency did not affect DA1 expression. Analysis of the expression of cholinergic markers showed that, AChRM1 and AChRα7 were overexpressed, while the content of AChE and VAChT remain unaffected. The available data from various studies suggest that increased levels of TH, DA2 and AChRM1 may be an anxiolytic mechanism to reduce anxiety, whereas DAT overexpression appears to be anxiogenic. AChRα7 overexpression can both reduce and promote anxiety, so the role of AChRα7 is not obvious.

ERβ is considered as a potent anxiolytic and antidepressant factor55. It was reported that in female mice lacking ERβ, anxiety and depressed mood are also accompanied by alternations in neurotransmission and there is a strong link between oestrogen signalling and dopaminergic signalling54,55. Oestrogens modulate the activity of the catecholaminergic systems53 through increasing or decreasing DA activity, e.g., by its degradation, reuptake, and recover, also by upregulating dopaminergic27 or by reducing the affinity of the DAT50. Furthermore, many of these effects, have been shown to be mediated by ERβ signalling51,52,53,54,55. Our results showed that TH, DAT and DA2 expression was significantly increased in the amygdala of female ERβ deficient mice, suggesting a role for ERβ in monoamine neurotransmission and anxiety behaviour in mice and corresponds well with the results of other authors. Namely, stimulating as well as inhibiting effects of oestrogen on dopaminergic neurotransmission have been documented in the caudate putamen of ERβ knock-out mice52. Thus, in the absence of functional ERβ, regardless of the presence or absence of circulating oestradiol (E2) in plasma, female mice exhibited enhanced anxiety and decreased concentrations of DA and serotonin (5-HT) in the caudate putamen in comparison to the WT animals52. On the other hand, after E2 treatment in the culture, TH activity in the medial basal hypothalamus as well as DA content in the diencephalic but not in the mesencephalic from both adult male and female rats were reduced81. Finally, Jacome et al.82 found that ERβ activation altered the levels of DA’s metabolite (homovanillic acid, HVA) in several areas of the brain, including the prefrontal cortex, ventral hippocampus, and dentate gyrus, but not in the striatum or medial septum. To summarize, oestrogens appear to modulate DA neurotransmission in a manner that is region-specific.

The present analysis of dopaminergic markers showed that the expression of TH, DAT, and DA2 was significantly elevated in the amygdala of female mice lacking ERβ and which show increased anxiety-like behaviour compared to matched controls, while the expression of DA1 remains unchanged in these animals. This is consistent with the fact that increased TH and DAT levels were observed in young infant rats in anxiogenic conditions83, and DAT itself was significantly higher in the amygdala of patients with social anxiety disorder18. Interestingly, the DAT content in the amygdala of these patients positively correlated with the severity of symptoms18. Importantly, the loss of TH in the mice basolateral amygdala induces anxiety-like behaviour17 whereas in humans, increasing catecholamine levels (via tyrosine administration) before fear conditioning reduces fear expression84. Actually, tyrosine itself did not reduce anxiety, but rather, it weakened the responses of fear expression84. On the other hand, studies in mice with DAT knockout demonstrate that reduced DAT levels, which is one of the most important proteins engaged in dopaminergic tone regulation85, correlate with intensified turnover of DA and increased DA content in the synaptic cleft86. Mice with DAT deprivation also exhibit less anxiety in the elevated plus maze test and other anxiety-related paradigms87. These studies were in line with the earlier evidence indicating that exposure of mice to bisphenol A (BPA) during development disrupts the hormonal balance, and reduces expression of DAT and anxiety-like behaviours88. Oestrogens may up- or down-regulate DAT expression by both direct and indirect (kinase-mediated) interactions. Subsequently, these changes in DAT expression lead to modifications in the ability of the neurons to transport DA89. Interestingly, neurodegenerative diseases often appear first in women in life stages when physiological oestrogen levels start to fluctuate and then drop90. Thus, it seems that oestrogens that which previously maintained synaptic DA concentration by promoting outflow cease to regulate this process, e.g., in the case of ERβ deficiency89. The level of DA in the synapse thus falls, and then might promote the development of diseases characterized by low DA stimulation89. It should also be kept in mind that clear gender differences have been observed in dopaminergic neurotransmission, i.e. in the mesocortical organization91,92, which may have functional consequences for DA signalling processes. For example, sex differences and different endocrine states in women can alter response to DA medications, which has been well documented90,93.

It is worth noting that intracellular DA accumulation may result in oxidative stress and neurotoxicity, and then subsequently negatively affect the survival of neurons94,95. Furthermore, moderately intensified activity of the DAT leads to spontaneous loss of dopaminergic neurons that may be reversed by L-3,4 dihydroxyphenylalanine (L-DOPA, as a precursor to DA) application, suggesting that the integrity of DA neurons is mainly dependent on the DAT potential to maintain proper presynaptic DA homeostasis96. Interestingly, in 2-year-old ERβ knock-out female mice there is a remarkable amount of neurodegeneration of dopaminergic neurons, particularly in the substantia nigra97. Elevated DA2 content in the amygdala of ERβ knock-out subjects coincides with elevated expression of DA2 (short and long) isoforms in mice characterized by anxiety-like behaviours and depression19. Moreover, the dopaminergic mechanisms mediated by DA2 receptors are predominantly associated with fear expression rather than with fear acquisition98. Although, the mechanisms underlying the elevated DA2 content in the amygdala of ERβ knock-out mice is as yet obscure, but, as evidenced, conditioned fear intensifies DA release in the amygdala99 which may be reflected in the present study by TH up-regulation. Further, in response to the DA2 agonist, levels of fear-potentiated startle and freezing decreased, whereas in response to DA2 antagonist they increased98,100. On the other hand, assuming that released DA more probably will bind to DA1, which are more numerous in the amygdala than DA2—present study and Rouge-Pont et al.101, it could result in overactivation of adenylyl cyclase. Subsequently, the elevated levels of DA2 may be induced to manage with the overactivation of adenylyl cyclase as DA2 activation down-regulates adenylyl cyclase43. Interestingly, a low level of DA would predominantly activate DA1 receptors, disinhibiting and facilitating the amygdala function, whereas higher DA levels would also activate DA2 receptors102. A similar situation was also observed in the present study, as TH and DA2 were significantly elevated in the amygdala of mice with the lack of ERβ, whereas simultaneously no change in DA1 expression was observed in these animals. Taken together, these data suggest that elevated levels of TH and DA2 provide increased DA content and facilitate DA2 recruitment in the amygdala. Such conditions may constitute a kind of anxiolytic mechanism to reduce fear and anxiety-like behaviour. In contrast, DAT overexpression in the amygdala decreases DA availability in the synaptic cleft, promoting anxiety.

Amygdala cholinergic modulation in female mice with ERβ knock-out is altered. This observation corresponds with results of other authors, which reported that oestrogens play an important role in the cholinergic neurotransmission. For instance, ovariectomy decreased the high affinity choline uptake as well as choline acetyltransferase activity103,104,105, whereas oestrogen replacement therapy reversed these effects105,106. In addition, an increase, and a decrease of the AChRMs density in the hypothalamus and in the preoptic area, respectively, in oestrogen-treated ovariectomized rats, when compared with ovariectomized rats have been shown107. In the rat hippocampus, ovariectomy up-regulated AChRM4108,109 and the E2 treatment reversed this effect108. Furthermore, adult female rats, showed greater variation in density and affinity of AChRMs in cortical tissues than male rats110. Thus, interaction between oestrogens and cholinergic system shows discrepancies that could be the result of species related differences, sex differences, and might be associated with brain areas and hormonal status of the animals, resulting in differential expression of cholinergic markers.

There is evidence indicating that AChRα7 is involved in circuits controlling mood and anxiety, when activated, can induce both anxiolytic111 and anxiogenic effects112. For example, studies in rats proved that in the basolateral amygdala, AChRα7 increase primarily GABAergic inhibitory tone, and when they are further exogenously activated they enhance spontaneous inhibitory postsynaptic currents in glutamatergic excitatory neurons, significantly diminishing overall excitability of these cells79. This may be one of the potential pathways allowing nicotine to suppress excitability of the amygdala and in this way reduce anxiety113 and alleviate depression114. This also suggests that the elevated content of AChRα7 in the amygdala of ERβ knock-out mice could be a mechanism to counteract increased anxiety-like behaviour in these animals. On the other hand, mecamylamine infusion into the mouse amygdala (non-competitive and non-selective nicotinic receptors’ antagonist), viral-mediated downregulation of the AChRα7 or knock-out AChRα7 all resulted in strong anxiolytic- and antidepressant-like effects61. Moreover, negative allosteric modulation of AChRα7 via BNC210 decreases reactivity of the amygdala in patients with generalized anxiety disorder115, while the amygdala’s hyperactivity to stimuli related to threat is a hallmark of anxiety116. These phenomena may be explained by the fact that AChRα7 are present not only in inhibitory but also in excitatory neurons15,79, and can enhance both GABAergic79 and glutamatergic117,118 neurotransmission in the amygdala. Furthermore, AChRα7 are especially crucial for synaptic plasticity in this brain region119. Interestingly, rats with anxiety disorder due to traumatic brain injury have significant surface expression of AChRα7 and current mediated by these receptors on principal neurons which are one of the important contributors of anxiety120. Considering that oestrogens reduce neuronal excitability in the amygdala13 through their receptors14, thereby reduced oestrogen signalling as a result of ERβ knock-out would be anxiogenic. Thus, AChRα7 antagonism can mediate an anxiolytic- and antidepressant-like response, and increased expression of these proteins in the ERβ knock-out mice could be one factor in increasing anxiety-like behaviour. Taken together, overexpression of AChRα7 in ERβ knock-out female mice may reflect two opposite mechanisms, and which of these is actually present in these animals requires further study.

Increased expression of AChRM1 in the amygdala of mice deprived of ERβ is quite prominent as AChRM1 are mostly present on projection neurons78 while there is a significant loss among these cells in ERβ knock-out mice24. Although AChRM1 are not directly linked with the expression of anxiety-like behaviours, they are, however, involved in several types of emotional/motivational learning including contextual fear conditioning and fear extinction121; and impaired fear extinction is a hallmark of anxiety-related disorders122. The reduced expression of AChRM1 within the amygdala was correlated with deficits in fear extinction learning, however, the increase in AChRM1 expression in this brain region may be related to intact and properly functioning fear extinction learning123. Further, direct injection of the muscarinic cholinergic agonist oxotremerine into the basolateral amygdala elicited an enhancement in fear extinction learning63. In addition, systemic AChRM1 antagonism reduced contextual fear extinction, while positive allosteric modulation of AChRM1 increased consolidation of contextual fear extinction in mouse model of posttraumatic stress disorder124. Finally, analyses using cevimeline (an AChRM1 agonist) and telenzepine (an AChRM1 antagonist), as well as AChRM1 knock-out mice showed that AChRM1 regulates the cued fear memory consolidation by redundant activation of phospholipase C in the basolateral amygdala125. Taken together, AChRM1 are required for fear extinction mediated by the amygdala, and overexpression of these proteins in mice with increased anxiety and cellular deficits may be a compensatory mechanism to alleviate fear extinction deficits and reduce anxiety.

The STRING database generated a protein–protein interaction (PPI) network for all proteins tested in this study and showed strong interactions between them. Moreover, the Gene Ontology (GO) molecular function enrichment analysis identified seven terms (biological processes) associated with behaviour and cognitive functions of Mus musculus, influenced by ERβ, as well as by the dopaminergic and cholinergic markers studied. Based on the conducted STRING of PPI analysis, it can be assumed that the changes in behaviour7 and cognitive functions impairments126 observed in ERβ−/− female mice may be related to the changes in dopaminergic and cholinergic transmission observed in the present study. This assumption needs to be verified by experimental data, especially since the available literature lacks any data on this subject.

It should be kept in mind that future studies are required to elucidate in detail the exact role of these markers in the formation of anxiety. For example, measurements of DA and ACh concentrations as well as the content of other markers in the amygdala by Western blot, ELISA and/or HPLC quantifications would probably allow for more precise functional conclusions.

Materials and methods

All details information about materials and small equipment were added to Supplementary 1.

The animals

Adult female mice, i.e., ERβ−/− (homozygous B6.129P2-Esr2tm1Unc/J, common name: ERβ KO, n = 6, aged 6–8 weeks) and ERβ+/+ (C57BL/6J, common name: B6, n = 6, aged 6–8 weeks) were provided by the Jackson Laboratory (Bar Harbor, ME, USA). The number of animals for this study was estimated based on a pilot study using studied markers in 3 brains in each group (n = 3 /ERβ−/−and n = 3 /ERβ+/+). This study indicated that the individual differences were small, the increase in the number of n did not significantly affect the obtained results, and n = 6 was sufficient. After acclimatization for 1 week housed in the animal facilities at the Faculty of Veterinary Medicine of the University of Warmia and Mazury (Olsztyn, Poland) under standardized conditions: 21 ± 1 °C, 12–20 air exchanges/h, 12-h shift of the light–dark cycle, free access to feedstuff devoid of phytoestrogens (LabDiet® JL Rat and Mouse/Auto 6F 5K52) and tap water (ad libitum). To avoid stress from isolation, animals were kept separately in sanitized polypropylene cages in a group of three. The care and handling of animals was strictly in line with the European Union Directive for animal experiments (2010/63/EU) and principle of the 3Rs: replacement, reduction, and refinement. Thus, all animals used were registered, and the staff was properly trained (Certificate No 1267/2015). This study is reported in accordance with ARRIVE guidelines. All efforts were made to minimize animal suffering and use the minimum number of animals necessary to generate reliable scientific data. It is worth noting that the strains of mice used in the present study were selected cautiously. We chose to use ERβ−/− females rather than ERβ−/− males, as ERβ−/− females are the best validated animal model of reduced oestrogen signalling and anxiety disorders based on genetic, behavioural, and neurobiological studies7,52,80. For example, results of anxiety-related tests such as the open field and the elevated plus maze (routinely used tests to study anxiety-related behaviour in mice) conducted in female and male ERβ−/−, and their wild-type (WT) control mice showed significant genotype differences and consistent behavioural abnormalities that are sensitive to enhanced anxiety. In the open field test, ERβ−/− females showed a significant increase in thigmotaxis and latency to move from the centre of the field to the wall comparing to both WT control and male ERβ−/− mice. On the other hand, total locomotion was significantly reduced in these mice7,127,128. Furthermore, in the elevated plus maze test, ERβ−/− mice spent less time in the open arm compared to other study groups7. The C57BL/6J mice were used to serve as a wild-type control according to the recommendations of the Jackson Laboratory (http://jaxmice.jax.org/strain/004745). The B6.129P2-Esr2tm1Unc/J mice were derived directly from C57BL/6J substrain mice (wild-type littermates), which prevented any genetic differences in the studied case.

Anaesthesia and tissue processing

After a 2-week-long habituation phase, mice were exposed to the same laboratory procedures. Before anaesthesia, monitoring of the oestrous cycle was performed to determine the stage of the oestrous cycle based on vaginal smear13,129. To avoid the oestrogen-induced anxiolytic effect, all mice were anaesthetized in metaestrous; a phase associated with a decline in oestrogen level130. Intraperitoneal injection anaesthesia in mice was done with pentobarbital (Morbital, Biowet, Poland; 2 mL/kg) according to Humane Society Veterinary Medical Association guidelines, and each mouse, after cessation of breathing, was transcardially perfused with sodium chloride (0.9%) followed by 4% paraformaldehyde (PFA) diluted in 0.1 M phosphate-buffered saline (PBS). The brains were then removed and were fixed in 4% PFA for overnight, washed three times in 0.1 M PBS (pH 7.4, 4 °C) and then transferred (for 3–5 days) to sucrose in 1xPBS at 4 °C (10%, 20% and 30%) as cryoprotectant. Lastly, the brains were frozen and then cut coronally into 10 μm thick sections using a cryostat. The sections were mounted on glass slides and stored at − 80 °C until use.

Immunohistochemistry

Representative sections of the amygdala from both ERβ−/− and ERβ+/+ mice were stained using a routine two-way immunofluorescent procedure. The details of this procedure have been described previously by Równiak131. All immunostaining steps were performed using humid chambers and at room temperature (RT). Briefly, to visualize all markers, sections were triple-washed in PBS and then incubated for 1 h at RT with a blocking buffer (0.1 M PBS, 10% normal donkey serum, 0.1% bovine serum albumin, 0.05% thimerosal and 1% Tween-20). These tissue samples were then incubated with primary antibodies diluted in a blocking buffer at RT overnight (Table 3). After that, sections were rinsed three times in PBS and then incubated for 1 h with a solution of secondary antibodies (Table 3). Finally, slides were coverslipped in carbonate-buffered glycerol (pH 8.6).

Table 3 Specification of antigens.
Full size table

It is worth noting that to define the location and boundaries of individual amygdala nuclei in these sections, a series of sections from our previous study24 and from the same mice which were stained using antibodies directed against a neuron-specific nuclear protein (NeuN, pan-neuronal marker) were used. These sections (15 slides in space 70–80 μm) were subjected according to the immunoperoxidase labelling with 3.3-diaminobenzidine (DAB) as a substrate-chromogen. Briefly, these sections were preincubated for 30 min in 0.3% H2O2 diluted in 99.85% methanol and then blocked for 60 min with a solution of 10% normal donkey serum (diluted in PBS). After primary antibodies’ incubation, the sections were triple-washed in PBS, treated with the solution of peroxidase-conjugated secondary antibodies for one hour (Table 3), and finally incubated with a 3% DAB solution. Then, sections were dehydrated, cleaned in xylene and mounted in slide mounting medium, a mixture of distyrene, plasticizer, and xylene (DPX). After this, on all DAB-stained slices, the coordinates for individual amygdala nuclei were determined according to the method previously described by Sterrenburg et al.132.

Controls

The specificity of the primary antisera used in this study was shown by various researchers using these products in multiple previous studies. Furthermore, all antibodies were positively validated by immunoblots (more in Supplementary 2). The specificity of the secondary antibodies was assessed by omitting the primary antibody or replacing it with non-immune sera or PBS. The specificity of the secondary antibody was confirmed by the lack of any immunosignal.

Counts and measurements

The density of the dopaminergic and cholinergic markers in the amygdala nuclei, was quantified on immunofluorescence-labelled sections and analysed using an Olympus BX51 microscope (Olympus GmbH, Germany) equipped with a digital camera (CC-12, Soft Imaging System, Münster, Germany) and Cell-F software (Olympus, Hamburg, Germany). The lateral (LA), basolateral (BL), basomedial (BM), medial (ME), central (CE) and cortical (CO) nuclei were tested and the delineation pattern of these nuclei and the amygdala nomenclature have been adopted (without any modification) from recent version of rodent atlas of Paxinos and Franklin74. In each animal of both ERβ−/− and ERβ+/+ mice, fifteen analyses for evenly spaced sections (per antigen and per nucleus) arranged from the rostral (bregma =  − 0.82) to the caudal extent (bregma =  − 2.46) of the amygdala74 were done. Additionally, density analyses performed in the present study (line scan analysis for volume density and automated cells counting) on the single section were always done with a 40× lens and with the use of 347.6 µm × 260.7 µm regions (test frames). These frames had an area of the computer screen, and they were always arranged to cover the full area of the analysed nucleus. Firstly, we measured the volume density of immunoreactive elements (somata, fibres, neuropil) according to the line scan analysis described and validated by Sathyanesan et al.133. Next, we measured in the same test frame immunoreactive neurons using automated cell counting. The density values calculated from individual test frames on a section were always reduced to the section average. As this average only referred to the area of the test frame, it had to always be converted to a value corresponding to the area of 0.04 mm2 or 0.04 mm3. To calculate antigen density in the particular nucleus, the results were averaged across all sections for each animal. Finally, the values obtained from each amygdala nucleus were also averaged to each group of mice and saved in the format: mean ± standard deviation (SD).

All analyses (volume density as well as manual and automated cell counting) were carried out on coded slides. These analyses were done by the principal investigator, and then they were repeated by two independent researchers in a blinded manner (ERβ−/− and ERβ+/+ mice). The obtained results showed a high degree of inter-rater reliability (Pearson R = 0.79).

Volume density counting

To evaluate the volume density of TH, DAT, DA1 and DA2 as well as AChE, VAChT, AChRM1 and AChRα7 immunoreactive elements in the amygdala nuclei of ERβ knock-out and wild type mice, stained sections were analysed according to the automated line scan analysis described and validated by Sathyanesan et al.133. The same parameters were used for capturing the image, including exposure time and camera gain, which provide a good grayscale dynamic range for all images collected from different sections and animals. All images were saved in 8-bit TIF grayscale format. Mathematically, fluorescently labelled structures can be represented as curvilinear structures with local intensity variations (minima or maxima)133, these local intensity extremes can be detected using a Hessian matrix134. In this way, filtering the image based on the Hessian matrix extracts line-like information from the input image. Consequently, all analysed fluorescence images were extracted by the Hessian-based filter included in the plugin called FeatureJ135 for NIH ImageJ software (version 1.53e). As the plugin offers several parameters to be chosen from, according to validation by Sathyanesan et al.133 the following parameter options were selected: “largest eigenvalue of Hessian tensor” option, and “absolute eigenvalue comparison” option and set the “smoothing scale” factor to 0.5. Next, the evaluation of these images by line scan profile analysis was carried out using five lines oriented horizontally (parallel scans) and the same number of lines oriented vertically (perpendicular scans). Lines were drawn using ImageJ’s “line tool” through the region of interest (ROI). Either straight or segmented line were drawn depending on the morphology of the amygdala nuclei. In the following step, the line scans were baseline-adjusted and then processed with a peak detection algorithm. The estimation of the baseline is conditioned by the manner of distinguishing peaks (signal) from the background. To estimate the background noise for the line scans, these scans were always collected from several places of the section which characterized by absence of immune signal. Each mean was then calculated based on these measurements and constituted the ultimate background value for the single section (threshold) for a peak detection algorithm.

Manual and automated cells counting

Before taking an image, the number of DA1+, DA2+, AChRM1+ or AChRα7+ neurons with a clearly visible labelling in the cell soma was manually counted using a high-resolution microscope (Olympus GmbH, Germany). For automated cell counting, the number of cells was counted on previously captured images using the NIH ImageJ software (version 1.53e). In the first step, after selecting the "dark background" option, which separates cells from the dark background, the analysed image was always converted to a binary image in order to extract objects based on their circularity. Then the particle size was set with the parameters 1800–2500 pixels and the circularity discrimination with the parameters 0.2–1.0. The number of test frames for each nucleus was the same in both methods: LA: 1–4, BL: 2–6, BM: 2–3, CO: 1–2, ME: 4–6, and CE: 2–3.

Protein–protein interaction (PPI) network

Additionally, the search tool for the retrieval of interacting genes (STRING) was used as an online tool for the protein–protein interaction (PPI) network analysis67. In this study, selected PPI networks of mice (Mus musculus) were constructed by STRING to access interactions among studied markers. The cut-off criterion p < 0.05 and Markov Clustering (MCL) of nodes were chosen. The results were filtered by the following criteria: p-adjusted ≤ 0.05 (Benjamini–Hochberg correction), query protein only, full STRING network, confidence of network edges, confidence equal to 0.400, and three group of clusters. Next, Gene Ontology (GO, Gene Ontology Resource) enrichment analysis was performed to visualize and indicate what biological functions the studied dopaminergic and cholinergic markers correspond to.

Statistic

The present research used a mixed model, and the parametric method was utilized for the whole analysis. The calculations were made with the use of Statistica 13.3 software (TIBCO Software Inc., Palo Alto, CA, USA). Statistically significant (α = 0.05) differences between experimental and control groups (ERβ−/− and ERβ+/+ mice) were detected by using the appropriate test statistics assuming the statistical power (1-β) was 0.95, and the allocation ratio was 1:1. Since the sample size of the ERβ−/− and ERβ+/+ mice was six individuals per group, fifteen sections were analysed per animal. The data from immunohistochemical studies were averaged and then examined using Shapiro–Wilk tests. While an independent-sample t-test with multiple comparisons was used to determine characteristic differences between the two groups. The examined homogeneity of variance was used, and then the Cochran Q test correction was performed. The significance level was set at p ≤ 0.05. All statistical graphs presented in the study were made in GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA). Data are presented with box-and-whisker plots. The “box” representing the median as well as 25th and 75th quartiles, and “whiskers” showing the 5th and 95th percentile (n = 6 per group).

Ethical approval

This study is reported in accordance with ARRIVE guidelines.

Conclusions

This study describes abnormalities in the expression of the main markers of the dopaminergic and cholinergic systems in the amygdala of female mice lacking ERβ, which are characterized by increased anxiety. The results show that the expression of TH, DAT, DA2, AChRM1, and AChRα7 is significantly elevated in the amygdala of mice with ERβ knock-out in relation to wild-type control subjects whereas the content of DA1, AChE, and VAChT is not affected by ERβ deficiency. Available data from multiple studies suggest that increased levels of TH, DA2, and AChRM1 may constitute an anxiolytic mechanism to alleviate anxiety, whereas DAT overexpression seems to be anxiogenic. The role of AChRα7 overexpression is not obvious, as increased levels of these proteins may reduce or promote anxiety.