Sex and estrogenic effects on coexpression of mRNAs in single ventromedial hypothalamic neurons

Abstract

Regulated gene expression in single neurons can be linked to biophysical events and behavior in the case of estrogen-regulated gene expression in neurons in the ventrolateral portion of the ventromedial nucleus (VMN) of the hypothalamus. These cells are essential for lordosis behavior. What genes are coexpressed in neurons that have high levels of mRNAs for estrogen receptors (ERs)? We have been able to isolate and measure certain mRNAs from individual VMN neurons collected from rat hypothalamus. Large numbers of neurons express mRNA for ERα, but these neurons are not identical with the population of VMN neurons expressing the likely gene duplication product, ERβ. An extremely high proportion of neurons expressing either ER also coexpress mRNA for the oxytocin receptor (OTR). This fact matches the known participation of oxytocin binding and signaling in sexual and affiliative behaviors. In view of data that ER and OTR can signal through PKCs, we looked at coexpression of selected PKCs in the same individual neurons. The most discriminating analysis was for triple coexpression of ERs, OTR, and each selected PKC isoform. These patterns of triple coexpression were significantly different for male vs. female VMN neurons. Further, individual neurons expressing ERα could distribute their signaling across the various PKC isoforms differently in different cells, whereas the reverse was not true. These findings and this methodology establish the basis for systematic linkage of the brain's hormone-sensitive signaling pathways to biophysical and behavioral mechanisms in a well studied mammalian system.

Estrogens, steroid sex hormones, play important roles in the development and establishment of reproductive and sexual behaviors as well as many other nonreproductive functions (1-4). Although estrogens' actions in females are relatively well understood, its role in male CNS function is less studied (5). Estrogens are regulated by binding to specific nuclear estrogen receptors (ERs), ERα and ERβ, ligand-activated transcription factors (6). ERα and ERβ belong to a large group of six subfamilies of transcription factors (7), one of which comprises the steroid receptors (8). ER expression in the brain varies by region (9) and can affect both female and male sexual behaviors (10-12). Both ERα and ERβ are expressed in the hypothalamus, and particularly in ventromedial nucleus (VMN), which is well known to regulate female sexual behavior (13, 14).

In addition to regulating sexual behaviors, estrogens affect affiliative social behaviors by affecting mRNA levels for the oxytocin receptors (OTR) (15), which are expressed in VMN (4, 16-19). Regulation of OTR by estrogens in VMN has been shown in females as well as males (16, 18, 20-25). A four-gene micronet theory featuring ERα, ERβ, oxytocin, and OTR and has been proposed to explain how neurons in hypothalamus and amygdala cooperate to facilitate social behaviors in female rodents (26, 27). OTRs are important for sex and anxiety behaviors (4). Regulation of OTR also requires second messenger-mediated kinases (28), one of which involves PKCs (29). Evidence shows that PKCα associates with OTR within 2 min of agonist stimulation (30).

It has been demonstrated that estrogens have two types of actions: genomic and membrane (31-33). Although ERs play a major role in genomic actions of estrogens, nongenomic actions in neurons involve several signal transduction pathways (34). And although lordosis behavior requires genomic actions of estrogens in VMN (35), estrogen-induced membrane actions in VMN can potentiate these genomic actions in a manner involving PKC (14). Several in vitro studies show that activation of PKC is involved in the potentiating effect of the rapid membrane actions of estrogens (34, 36-39). PKCs comprise a family of 12 related isozymes (40) that are divided into three groups: classical or calcium-dependent (PKCα, -βI, -βII, and -γ); the novel or calcium-independent (PKCδ, -θ, -η, -μ, and -ε); and atypical PKCs (PKCζ and -λ/ι) (40, 41). Different PKC isozymes can serve different functions in single cells (42). PKC involvement in many different kinds of signaling, including estrogens' rapid action at the membrane (43), have been demonstrated. We chose to test members of calcium-dependent and calcium-independent groups because we know that, in order for estrogen's membrane actions to potentiate its genomic actions, it is necessary to increase PKC activation, which, in turn, increases intracellular calcium.

Despite all this information on neuronal roles for ERs, OTR, and PKCs, there is no evidence about the manner in which these genes are expressed together in individual neurons. Amplification of RNA can be achieved by using RT-PCR (44) or the amplified antisense RNA (aRNA) (45) procedure. Both techniques have been used to amplify transcripts from single living cells (46, 47). Low copy numbers (48) make it very difficult to achieve quantitative analyses of such PCR products. The newest real-time PCR technology has made quantitation more reliable (49), so we chose this method to demonstrate gene coexpression in individual hypothalamic neurons.

Materials and Methods

Animals and Housing. Female (n = 27) and male (n = 9) Sprague-Dawley rats (Charles River Laboratories) were used at 3 weeks of age because they yield good cells for recording, and rats at this age can respond to estrogens behaviorally. The rats were housed on 12:12-h light/dark cycle (lights on at 11 a.m.), and food and water were available ad libitum. All animal procedures were performed according to the National Institutes of Health and The Rockefeller University Institutional Animal Care and Use Committee guidelines.

Experimental Procedures. Each experimental group consisted of nine rats yielding n = 82 neurons per group. Males were studied gonadally intact. Female rats were split into three experimental groups: (i) intact rats whose slices were perfused in a recording chamber with regular artificial cerebrospinal fluid (10 mM glucose/26 mM NaHCO₃/124 mM NaCl/2.5 mM KCl/1.2 mM KH₂PO₄/1.3 mM MgSO₄-7H₂O/2.4 mM CaCl₂-2H₂O/10 mM sucrose, pH 7.4; Sigma); (ii) ovariectomized (OVX) rats treated with 17β-estradiol (10 μg/0.1 ml; two injections 24 h apart); and (iii) OVX rats treated with sesame oil (two injections 24 h apart). Before decapitation, the animals were anesthetized with 40% urethane solution (4 g/100 ml). The brains were removed, and 300-μm slices were prepared on a sectioning system (Vibratome 1000 Plus, Vibratome, St. Louis). The slices were incubated at room temperature for 30 min in oxygenized regular artificial cerebrospinal fluid. After 30 min, the slices were placed in a patch-clamp apparatus chamber and perfused with oxygenized regular artificial cerebrospinal fluid. The collection criterion was based on cell size and location. We attempted to collect neurons whose cells bodies were >10 μm, and they were collected from the ventrolateral VMN of the hypothalamus (Fig. 1). To collect cells, patch glass capillary tubing was used (G8515OT-4, Warner Instruments, Hamden, CT), which was previously cleaned and fire-polished and filled with 15 μl of nuclease-free water (Ambion, Austin, TX). Collection was very fast; from suction to freezing took only a few seconds. Neurons were transferred into 0.2-ml PCR tubes (Molecular Bio-Products, San Diego) and were frozen immediately at -80°C.

Fig. 1.

Neurons were collected from the ventrolateral part of the ventromedial hypothalamic nucleus (VMHVL). Upper was adapted from ref. 63. Neurons were chosen by their shape and size. (Lower) One particular neuron before and after suctioning. (Inset) An enlarged photomicrograph of a neuron collected in this project.

Real-Time Quantitative PCR. When the appropriate number of neurons were collected (an experimental group), mRNA was reverse-transcribed (RT reaction) into cDNA (1 h at 37°C, Programmable Thermal Controller, MJ Research, Cambridge, MA) following the manufacturer's protocol (Applied Biosystems): 1 μl of RNase-free water, 2.2 μl of 10× TaqMan RT Buffer, 2.0 μl of dNTPs mixture, 0.5 μl of random hexamers, 0.2 μl of RNase inhibitor, 50 units/μl MultiScribe reverse transcriptase (1×, 5.5 mM, 500 μM each dNTP, 2.5 μM, 0.4 unit/μl, 1.25 unit/μl; final concentrations, respectively), and 15 μl of mRNA. In a control group of neurons, to eliminate any possibility of amplification of product other than cDNA (e.g., to ensure that our PCR amplification methods did not amplify genomic DNA), the RT reaction was performed without using MultiScribe reverse transcriptase (which was replaced with RNase/DNase-free water). RT was followed by real-time PCR using the SYBR Green detection system (Applied Biosystems, Prism 7700). Conditions for the cycles were as follows: PCR initial activation step, 15 min at 95°C; 3 (4)-step cycling, denaturation for 15 s at 94°C; annealing for 20-30 s at 50-60°C; and extension for 10-30 s at 72°C. Number of cycles: 40. Primer pairs of 100-150 bp were designed for ERα, ERβ, OTR, PKCα, PKCδ, PKCε, PKCη, and β-actin (Primer Express, Applied Biosystems). Total volume of PCR was 20 μl, and it contained 2× QuantiTech SYBR Green PCR Master Mix, 10 μl of forward primer, 2.5 μM reverse primer, and 2.5 μM and 5 μl of template DNA. Results were analyzed by using Prism 7000 sds software (Applied Biosystems).

Primer pairs were as follows: ERα forward, CCTGGTTGGAGATCCTGATGA, and reverse, TCAAAGATCTCCACCATGCCT; ERβ forward, TGCTGGATGGAGGTGCTAATG, and reverse, GAGGAGATACCACTCTTCGCAATC; OTR forward, TTCTTCTGCTGCTCTGCTCGT, and reverse, TCATGCTGAAGATGGCTGAGA; PKCα forward, CAAGCAGTGCGTGATCAATGT, and reverse, GGTGACGTGCAGCTTTTCATC; PKCδ forward, TCAAGAACCACGAGTTCATCG, and reverse, GCATTGCCTGCATTTGTAGC; PKCε forward, CGTCACTGATGTGTGCAATG, and reverse, TCGAACTGGATGGTGCAGTTG; PKCη forward, TCTTCCAGCAGATTCGGCAT, and reverse, TCAGCACCTTCACAGCGTACA; and β-actin forward, ATCGCTGACAGGATGCAGAAG, and reverse, AGAGCCACCAATCCACACAGA.

We were conservative in determining a criterion of a gene expression. Gene expression for a particular target gene was considered to have occurred if the amplification plot crossed the cycle threshold (Ct) at or before 35 cycles (Fig. 2). Another quantitative factor for expressed genes was their dissociation curves; dissociation curve analyses were made by using the SYBR Green signal detection system. The dissociation curve determines the melting temperature (T_m) of a single target nucleic acid sequence within an unknown sample and, as expected, displayed a single sharp peak.

Fig. 2.

Coexpression of mRNA population in a single neuron was determined by using the SYBR Green signal detection system with the Prism 7700, both from Applied Biosystems. The cycle number is shown on the x axis. The y axis represents the normalized reporter signal minus the baseline signal established in the first few cycles of PCR. Each plot above the threshold line represents genes amplified at a certain cycle. Genes that were not expressed in these particular neurons were under the threshold line.

Controls. To eliminate the possibility of genomic DNA contamination, the RT reactions for seven different transcripts were performed on a group of control neurons. mRNA was reverse-transcribed according to the manufacturer's protocol (Applied Biosystems): RNase-free water, 10× TaqMan RT buffer, dNTPs mixture, random hexamers, and RNase inhibitor. MultiScribe reverse transcriptase was replaced with an appropriate amount of RNase/DNase-free water. cDNA reactions were used for real-time PCR done by using the SYBR Green signal detection system with the Prism 7700 (Applied Biosystems).

Statistical Analyses. The χ² test for k independent samples (50) was used to determine the significance of differences that occurred between females and males with regard to number of cells displaying triple coexpression of ERα⁺β/OTR mRNAs with four isozymes of PKC mRNA. Also, this test, useful for percent of cells, was performed to see differences between the ERα⁺ neurons coexpressing PKCs and PKC⁺ cells coexpressing ERα. Relative differences in gene expression were calculated by using a modification of the ΔΔCt method described in ref. 51. First, each gene of interest was normalized to β-actin gene expression per cell. Briefly, normalized gene expression , where E = efficiency and Ct = cycle threshold. The resulting normalized gene expression for each cell was then analyzed by using a Mann-Whitney U test.

Results

Results obtained from real-time PCR of a control group (where RT was performed without reverse transcriptase) showed that samples were not contaminated with genomic DNA, and the only amplified product was cDNA. Thus, we were able to study individual neurons in a hypothalamic cell group known to control a specific hormone-regulated behavior.

ERα vs. ERβ. These likely gene duplication products were not always expressed in the same neuron (Tables 1 and 2) and did not display identical patterns of coexpression with other transcripts. A comparison between the two forms of ER showed the following: in intact females, the number of ventrolateral VMN neurons expressing ERα was higher than the number of those expressing ERβ (65 vs. 23, respectively; P < 0.005). A difference also was found in males (41 expressing ERα vs. 25 expressing ERβ; P < 0.05). When intact females were compared with males with regard to ERs, females expressed higher numbers of ERα (65 vs. 41, P < 0.05), whereas ERβ expression remained almost the same (23 vs. 25) (Tables 1 and 2). Regarding coexpression of the PKC isoforms studied, percentages of ERα neurons coexpressing a given PKC were never identical to the percentages of ERβ neurons expressing that PKC isoform (Tables 1 and 2).

Table 1. Gene coexpression in neurons of intact males.

	ERα (41 cells)	ERβ (25 cells)	OTR (37 cells)	PKCα (8 cells)	PKCδ (15 cells)	PKCε (17 cells)	PKCη (16 cells)
ERα	—	76.0	67.6	62.5	80.0	70.6	62.5
ERβ	48.8	—	48.6	37.5	40.0	52.9	75.0
OTR	63.4	72.0	—	37.5	53.3	58.8	43.8
PKCα	9.8	12.0	8.1	—	26.7	23.5	25.0
PKCδ	29.3	24.0	21.6	50.0	—	23.5	37.5
PKCε	29.3	36.0	27.0	50.0	26.7	—	50.0
PKCη	29.3	48.0	18.9	50.0	33.3	47.1	—

The column headings show in parentheses the number of neurons expressing the listed genes in the 82 tested neurons. The body of the table shows percent of cells expressing the genes listed in the left column.

Table 2. Gene coexpression in neurons of intact females.

	ERα (65 cells)	ERβ (23 cells)	OTR (50 cells)	PKCα (13 cells)	PKCδ (22 cells)	PKCε (21 cells)	PKCη (22 cells)
ERα	—	65.2	90.0	69.2	72.7	61.9	81.8
ERβ	32.3	—	28.0	38.5	45.5	42.9	31.8
OTR	86.2	56.5	—	53.8	72.7	61.9	54.5
PKCα	13.8	21.7	14.0	—	27.3	23.8	22.7
PKCδ	24.6	39.1	32.0	46.2	—	47.6	40.9
PKCε	21.5	34.8	26.0	46.2	45.5	—	36.4
PKCη	29.2	30.4	24.0	38.5	45.5	42.9	—

The column headings show in parentheses the number of neurons expressing the listed genes in the 82 tested neurons. The body of the table shows percent of cells expressing the genes listed in the left column.

Massive Coexpression of ERs with OTR in Intact Females and Males. Our experiments showed that in intact females, ERα⁺ neurons coexpressed OTR in 86.2% of the cells (Table 2); in males, the corresponding number was 63.4% of the cells (Table 1). For ERβ-expressing neurons, coexpression of OTR was detected in 56.5% and 72% of cells in female and males, respectively.

When we compared ERα/OTR or ERβ/OTR coexpression between intact female and male neurons, we noticed a trend toward a higher percent of ERα⁺ cells expressed OTR in females, compared with male ERα⁺ cells (86.2% vs. 63.4%). This trend also appeared when we compared OTR/ERα-coexpressing neurons between the two groups with higher coexpression in females (90% vs. 67.6%) (Tables 1 and 2).

Estrogen Treatment Increases Number of ERα, ERβ, OTR, and PKCs in Female Rats. Comparisons between female rats treated with estrogen vs. oil for 2 days showed a significant difference in the number of cells expressing all genes of interest, except PKCα. In estrogen-treated animals, these numbers were higher then in the oil group, ERα (66% vs. 20%), ERβ (39% vs. 14%), OTR (56% vs. 15%), PKCδ (28% vs. 8%), PKCε (30% vs. 8%), and PKCη (44% vs. 12%) (P < 0.05 in all cases) (Tables 3 and 4).

Table 3. Gene coexpression in neurons of estrogen-treated OVX females.

	ERα (66 cells)	ERβ (39 cells)	OTR (56 cells)	PKCα (12 cells)	PKCδ (28 cells)	PKCε (30 cells)	PKCη (44 cells)
ERα	—	100.0	98.2	100.0	100.0	70.0	97.7
ERβ	59.1	—	64.3	58.3	67.9	43.3	70.5
OTR	81.8	92.3	—	83.3	96.4	66.7	97.7
PKCα	18.2	17.9	17.9	—	17.9	13.3	13.6
PKCδ	42.4	48.7	48.2	41.7	—	53.3	47.7
PKCε	31.8	33.3	33.9	33.3	57.1	—	34.1
PKCη	66.7	79.5	76.8	58.3	78.6	63.3	—

OVX females were treated with 17β-estradiol (10 μg/0.1 ml) for 2 consecutive days before collection of neurons. The column headings show in parentheses the number of neurons expressing the listed genes in the 82 tested neurons. The body of the table shows percent of cells expressing the genes listed in the left column. Oil-treated controls are shown in Table 4.

Table 4. Gene coexpression in neurons of oil-treated OVX females.

	ERα (20 cells)	ERβ (14 cells)	OTR (15 cells)	PKCα (5 cells)	PKCδ (8 cells)	PKCε (8 cells)	PKCη (12 cells)
ERα	—	100.0	100.0	100.0	100.0	100.0	100.0
ERβ	70.0	—	86.7	80.0	100.0	100.0	100.0
OTR	75.0	92.9	—	80.0	87.5	100.0	91.7
PKCα	25.0	28.6	26.7	—	25.0	37.5	25.0
PKCδ	40.0	57.1	46.7	40.0	—	87.5	58.3
PKCε	40.0	57.1	53.3	80.0	87.5	—	66.7
PKCη	60.0	85.7	73.3	60.0	100.0	100.0	—

OVX females were treated with 0.1 ml of sesame oil for 2 days before collection of neurons. The column headings show in parentheses the number of neurons expressing the listed genes in the 82 tested neurons. The body of the table shows percent of cells expressing the genes listed in the left column. Compare results from estrogen-treated animals in Table 3.

ER/PKC, PKC/ER Coexpression in Female and Male Rats. In all of the experimental groups (Tables 1, 2, 3, 4), the neurons that expressed ERα displayed the entire variety of different PKC isoforms distributed across the ERα population (Fig. 3). The reverse was not true. In all groups, given that a neuron expressed any of the PKC isoforms, virtually all expressed ERα (Fig. 3). This fact was not true for ERβ (Fig. 4).

Fig. 3.

Neurons expressing ERα also coexpress certain numbers of different PKC isozymes and vice versa. The comparison was made between ERα⁺ neurons coexpressing PKCs and PKC⁺ cells coexpressing ERα in intact female rats (difference significant at P < 0.0001).

Fig. 4.

The same kind of comparison as in Fig. 3 involving ERβ-expressing neurons coexpressing PKCs and vice versa did not show a significant difference in the same group of intact female rats.

Triple Coexpression ER/OTR with PKCs. Interestingly, triple identification for ERα/OTR/PKCs and ERβ/OTR/PKCs showed that the patterns of coexpression with PKCα, PKCδ, PKCε, and PKCη were significantly different (P < 0.0001) (Fig. 5) when female and male neurons were compared.

Fig. 5.

Triple comparisons of ERα/ERβ/OTR mRNAs with mRNAs coding four isozymes of PKCs. Comparing male and female neurons, differences in PKC coexpression were significant (P < 0.0001).

Quantification of Single-Cell Gene Expression. When quantitative real-time PCR was analyzed, there was a significant difference observed in normalized gene expression of ERα (P < 0.05). Specifically, estrogen-treated animals had increased ERα gene expression, compared with vehicle-treated animals. In addition, there was a significant difference in normalized gene expression of PKCα (P < 0.05). Estrogen-treated animals had increased PKCα gene expression, compared with vehicle-treated animals. There was a trend toward increased expression of the OTR gene in estrogen-treated animals, compared with vehicle-treated animals, that approached statistical significance (P = 0.058).

Finally, there were no significant differences observed in normalized gene expression of ERβ, PKCδ, PKCε, and PKCη (P > 0.05) (Fig. 6).

Fig. 6.

Relative differences in gene expression were calculated by using a modification of the ΔΔCt. First, each gene of interest was normalized to β-actin gene expression per cell. Briefly, normalized gene expression , where E = efficiency and Ct = cycle threshold. There was a significant difference observed in normalized gene expression of ERα (P < 0.05): estrogen-treated animals had increased ERα gene expression, compared with vehicle-treated animals. Also, estrogen-treated animals had increased PKCα gene expression, compared with vehicle-treated animals. There was a noticeable increase of OTR gene expression in estrogen-treated animals, compared with vehicle-treated animals, approaching statistical significance (P = 0.058). There were no significant differences observed in normalized gene expression of ERβ, PKCδ, PKCε, and PKCη (P > 0.05).

Discussion

Two likely gene duplication products, ERα and ERβ, were not always expressed by the same nerve cell and, in particular, had different patterns of coexpression with other mRNAs coding for signal-transduction moieties. This finding is interesting because these two proteins are very similar, differing only in a 54-aa sequence at the N terminus, yet function differently in the CNS. In some assays, ERα and ERβ have diametrically opposite effects (52). We note that the ability of estrogens to increase ERα expression after this 2-day treatment differs from a briefer estrogen treatment (53) in which estrogen down-regulated its own receptors, thus indicating a biphasic curve as a function of duration of treatment. Within VMN, the importance of these two genes derives from their strong expression (9, 54, 55) in a cell group with demonstrable sexual dimorphism (9, 42, 56-58). This cell group controls the entire neural circuit for estrogen-dependent sexual behavior (59).

The high degree of coexpression between ERs and the mRNA for OTR was striking. OTRs respond to a 9-aa peptide, which plays a major role with respect to estrogenic effects on reproductive behaviors (2). Estrogens induce OTR expression in both females and males in the VMN, specifically, but not, for example, in the amygdala (16, 29, 60-62). Although receptors for other transmitters and neuropeptides also are influenced by estrogen administration (13), the present data point to the gene for OTR as representing a nodal point for the hormonal control of neuronal activity in this part of hypothalamus.

Once OTR is induced, what happens? Estrogenic signaling, whether strictly genomic or membrane-initiated, clearly involves several signal-transduction pathways (28, 34, 36-38). Prominent among virtually all of the data in experiments reported to date, particularly with respect to membrane-initiated hormone effects, is the participation of PKCα. Besides its well recognized kinase cascades, PKCα may have roles important for neuroendocrine functions in having the capacity to phosphorylate nuclear receptors, ligand-dependent transcription factors, and subunits of neurotransmitter receptors. Among the several genes coding for PKCs, the reasons that PKCα apparently figures so prominently are not well understood and require further investigation.

Piecing Together Transcriptional Networks. Some of our more interesting results regarding OTR and PKCs came not from simple coexpression analyses but from triple-identification of hypothalamic neurons. These results comprise a modest beginning of the discovery of potential molecular pathways activated by estrogen treatment. That is, as techniques for mRNA amplification become more sensitive, extending this approach to ever-larger, well chosen sets of transcripts, this type of work will amount to a rational attempt to piece together transcriptional networks in individual neurons. Quite distinct from purely computational approaches to the deduction of transcriptional networks based on microarrays, which usually mix together different cell types, an extension of our methodology will comprise a pathway toward network identification within the transcriptome of single, functionally identified neurons.

Implications. The present results allow us to go beyond the “one-gene/one-behavior” formulations that typify some approaches to genomic influences on mammalian behavior. Specifically, we have asked how ER signaling in ventromedial hypothalamic neurons can be revealed through mRNA coexpression, to help explain sociosexual behaviors. Three themes emerge from the data. First, the data imply different roles for ERα, compared with ERβ. Second, although the OTR results clearly point to oxytocin signaling as being of primary importance, the success of this approach encourages a search for mRNAs coding for other neuropeptides' receptors such as the receptor for gonadotropin-releasing hormone. Third, the apparent distribution of ER signaling across different PKCs provides for a multiplicity and flexibility of signal-transduction routes in ER-expressing VMN neurons.