Acute sleep deprivation upregulates serotonin 2A receptors in the frontal cortex of mice via the immediate early gene Egr3.

Serotonin 2A receptors (5-HT2ARs) mediate the hallucinogenic effects of psychedelic drugs and are a key target of the leading class of medications used to treat psychotic disorders. These findings suggest that dysfunction of 5-HT2ARs may contribute to the symptoms of schizophrenia, a mental illness characterized by perceptual and cognitive disturbances. Indeed, numerous studies have found that 5-HT2ARs are reduced in the brains of individuals with schizophrenia. However, the mechanisms that regulate 5-HT2AR expression remain poorly understood. Here, we show that a physiologic environmental stimulus, sleep deprivation, significantly upregulates 5-HT2AR levels in the mouse frontal cortex in as little as 6-8 h (for mRNA and protein, respectively). This induction requires the activity-dependent immediate early gene transcription factor early growth response 3 (Egr3) as it does not occur in Egr3 deficient (-/-) mice. Using chromatin immunoprecipitation, we show that EGR3 protein binds to the promoter of Htr2a, the gene that encodes the 5-HT2AR, in the frontal cortex in vivo, and drives expression of in vitro reporter constructs via two EGR3 binding sites in the Htr2a promoter. These results suggest that EGR3 directly regulates Htr2a expression, and 5-HT2AR levels, in the frontal cortex in response to physiologic stimuli. Analysis of publicly available post-mortem gene expression data revealed that both EGR3 and HTR2A mRNA are reduced in the prefrontal cortex of schizophrenia patients compared to controls. Together these findings suggest a mechanism by which environmental stimuli alter levels of a brain receptor that may mediate the symptoms, and treatment, of mental illness.

Introduction

The serotonin 2A receptor (5-HT2AR) is extensively expressed in the cerebral cortex where it is believed to play a critical role in perception, cognition, and psychosis. Much of the evidence for this comes from studies on the effects of drugs that bind to the 5-HT2AR [1-3]. This includes agonists, such as lysergic acid (LSD), psilocybin, and mescaline, which cause hallucinations [4-6], as well as antagonists and inverse agonists, including second-generation antipsychotics, which reverse the perceptual disturbances of psychiatric illnesses such as schizophrenia [6-8]. Additional findings supporting a role for the 5-HT2AR in the symptoms of psychosis include a long history of studies showing abnormal levels of 5-HT2ARs in the brains of patients diagnosed with schizophrenia. While some studies did not detect differences [9-11], and one group identified increased levels [12, 13], the vast majority of these studies, including a meta-analysis [14], have revealed reduced 5-HT2AR levels in schizophrenia subjects. These include post-mortem studies measuring 5-HT2AR ligand binding and mRNA expression, and as well as in vivo Positron Emission Tomography (PET) scan investigations [14-24]. Notably, studies in antipsychotic-naïve patients indicate that the reduction in 5-HT2AR levels is not simply a consequence of medication treatment [16, 18]. Despite the longstanding recognition of the importance of 5-HT2ARs in the response to psychedelic drugs, and the symptoms and treatment of psychotic disorders, the processes that regulate expression of this critical receptor remain unknown.

Our prior work identified that mice lacking the immediate early gene (IEG) Egr3 (Egr3−/− mice) have reduced levels of 5-HT2ARs in the frontal cortex [25]. This suggested that Egr3 is required for expression of Htr2a, the gene that encodes the 5-HT2AR. If so, we reasoned that stimuli that activate expression of Egr3 may also induce Htr2a expression. Sleep deprivation (SD) is a physiological stimulus that upregulates Egr3 in the mouse cerebral cortex [26]. Using this approach, we found that just six hours of SD increases Htr2a mRNA levels in the mouse cortex and, moreover, this required Egr3 [27].

This discovery that Htr2a expression can be rapidly induced in the rodent brain in response to a physiologic stimulus is supported by a prior study in humans. In 2012 Elmenhorst and colleagues reported that preventing healthy subjects from sleeping for a period of 24 hours resulted in a significant increase in 5-HT2AR binding in the neocortex, detected on PET scan [28].

These findings raised the question of whether EGR3, an activity dependent transcription factor, directly regulates the Htr2a gene, and thereby mediates the environmental induction of Htr2a mRNA in the cortex. In the current study we have addressed the questions of where in the cerebral cortex SD is inducing Htr2a expression, does this also result in increased 5-HT2AR protein, and is EGR3 directly regulating the Htr2a gene to effect this environmentally-induced receptor expression. We show that six hours of SD significantly upregulates Htr2a in the prefrontal cortex and adjacent sensorimotor cortex, and that 8 hours of SD upregulates 5-HT2AR levels in the same regions. In both cases this upregulation requires Egr3. Further, we show that the EGR3 protein binds to the Htr2a promoter in vivo and drives expression of an in vitro reporter construct, suggesting that the activity dependent IEG transcription factor EGR3 directly regulates expression of Htr2a in response to the acute environmental stimulus of SD.

Materials and Methods

See online supplementary methods for details and primer sequences.

Animals

Previously generated Egr3−/− mice [29], backcrossed to a C57BL/6 background for >30 generations, were housed on a 14/10 h light/dark schedule with ad libitum access to food and water. Matched pairs of littermate Egr3−/− and Wildtype (WT) mice were designated at weaning and randomly assigned to experimental groups. Male mice ages 3–5 months were used for qRT-PCR studies. Male and female mice we used for in situ hybridization (4 months old) and autoradiography (7–9 weeks old) experiments. WT male C57BL/6NTac mice (2.5–4 months old) were obtained from Taconic Biosciences for Western blot and chromatin immunoprecipitation (ChIP) experiments. Mice were acclimated to the facility for two weeks before experiments. Animal studies were performed in accordance with the University of Arizona Institutional Animal Care and Use Committee (IACUC) guidelines under an approved IACUC protocol.

Sleep deprivation (SD)

SD was initiated at the onset of “lights on” (8:00 a.m.) and conducted for 6h (for mRNA studies) or 8h (for protein), using the “Gentle handling” method as previously described [27]. Investigators were blinded to genotype during SD experiments. The number and type of stimulation required to keep each animal awake during the SD did not differ between WT and Egr3−/− mice (Supplementary Figure 1).

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Tissue was dissected on ice, transferred to RNAlater (Thermo Fisher Scientific), frozen on dry ice and stored at −80°C. Tissue homogenization and RNA isolation was performed in TRI reagent (Life Technologies) using ceramic beads and a MagMAX express processor (Applied Biosystems). mRNA was reverse transcribed using M-MLV reverse transcriptase kit (Life Technologies) and qRT-PCR performed using FastStart SYBR green master mix (Roche). Each sample was amplified in triplicate for the gene of interest and the housekeeping gene phosphoglycerate kinase 1 (Pgk1). Fold changes in gene expression were calculated and data were plotted using the 2-ΔΔCt method [30]. Relative gene expression of all samples was first normalized to Pgk1 (ΔCt). Then, average ΔCt data were normalized to WT SD controls for data points representing WT animals that underwent SD, or Egr3−/− SD controls for Egr3−/− animals that underwent SD.

In situ hybridization

In situ hybridization was performed on 12μm sections using the RNAscope Multiplex Fluorescent Reagent Kit V2 (Catalogue # 323100). The Htr2a probe was purchased from Advanced Cell Diagnostics, Inc (Probe-Mm-Htr2a-C3, 401291-C3) and developed with the fluorescent Opal dye 570 (Akoya Biosciences, Cat. # FP1488001KT) at a concentration of 1:750. Sections were counterstained with DAPI (Vector Laboratories). Qualitative analyses performed via thorough review of all tissue sections and in reference to anatomical regions described in The Allen Mouse Brain Atlas (Reference Atlas, Interactive P56, Coronal) [31, 32].

Radioligand Binding Analysis – Autoradiography

Six 12μm sections of anterior frontal cortex (Bregma 2.8 – Bregma 2.34) and posterior frontal cortex (Bregma 1.70 – Bregma 0.86), and two sections of mid-cortex (Bregma 0.02 – 0.82), and posterior cortex (Bregma −1.34 – 2.30) were thaw mounted on the same slide. A total of ten slides/brain were serially collected and stored at −80°C. Tissue was treated with selective 5HT2AR antagonist 3-H-MDL100907 [R(+)-α-(2,3-dimethoxyphenyl)-1-[2-(4-fluorphenyl)-ethyl]-4-piperidin-methanol] (specific activity; 59.22 Ci/mmol) and detected by autoradiography. Non-drug exposed control tissue was used to assess non-specific binding.

Western blot

Six hours after SD, the frontal cortex was dissected. Right and left hemispheres were used for Western blot and ChIP experiments respectively. Tissue was Dounce homogenized in RIPA buffer and centrifuged at 12,000 rpm, 4°C x 20 min. Supernatant protein was denatured at 95 °C x 5 min and 20ug of protein was loaded per lane on an Any kD TGX polyacrylamide gel (Bio-Rad Laboratories). Samples were transferred to nitrocellulose membrane and blocked overnight at 4°C in 5% non-fat dry milk in Tris-buffered saline with 0.05% Tween 20 (TBS-T). Blocked membranes were incubated for 2h at room temperature (RT) with rabbit anti-EGR3 antibody (1:500, kindly donated by Dr. Mary Kay Lobo, University of Maryland School of Medicine, original source: Baraban Lab, Johns Hopkins University School of Medicine [33]) and mouse anti-β actin antibody (1:5000, Sigma-Aldrich, Cat. # A5441) diluted in 1% non-fat dry milk/TBS-T. Following washes, membranes were incubated with secondary antibodies IRDye800CW (1:10,000 for EGR3, Li-COR Biosciences, Cat. # 926–32211) and IRDye680RD (1:10,000 for β-actin, Li-COR Biosciences, Cat. # 926–68072) for 1h at RT. Blots were washed in TBS with 0.1% Tween 20, and imaged. Protein levels were determined as ratios of EGR3 to β-actin internal control and reported as percentage of the control group. For Western blot experiments, recombinant EGR3 protein was synthesized using the TnT Coupled Reticulocyte Lysate System (Promega, Cat. # L4611) using template pcDNA 3.1(−) mouse Egr3, a gift from Peter Johnson (Addgene plasmid # 107998; http://n2t.net/addgene:107998; RRID:Addgene_107998) [34].

Promoter Analysis

The genomic region, including 4 kb upstream of the Htr2a transcription start site (NM_172812, chr14:74636840–74640839), was obtained from the UCSC genome browser (https://genome.ucsc.edu/). Consensus binding sites for EGR3 were identified using the software ‘Find Individual Motif Occurrences’ (FIMO, http://meme-uite.org/tools/fimo) [35]. The motif occurrences with a p value less than 0.0001 were selected.

Chromatin immunoprecipitation

Following 6h of SD, frontal cortex was dissected, minced, incubated in 1% formaldehyde to crosslink DNA-protein, and quenched with 125 mM glycine. Right and left hemispheres were used for Western blot and ChIP experiments respectively. Samples were homogenized, and chromatin sheared to 200 bp-1000 bp at 4°C for 25 cycles. Fragment size was verified via agarose gel electrophoresis. Samples were incubated with 70μl of anti-EGR3 antibody [33] complexed to magnetic beads (Invitrogen) x 16h at 4°C. Control IgG IP was performed in parallel. Beads were sequentially washed with low salt, high salt, and LiCl wash buffers. Samples were reverse crosslinked at 65°C overnight and treated with 2μl RNase (Roche) at 37°C x 1h and 2μl proteinase K (Invitrogen) at 55°C x 2h. DNA was isolated via QIAGEN kit and qPCR was performed. Relative occupancy (aka. fold enrichment) of the EGR3 proteins at each Htr2a promoter site was estimated using the following equation 2^(ΔCt MOCK- ΔCt SPECIFIC), where ΔCT MOCK and ΔCT SPECIFIC are mean normalized threshold cycles of PCR performed in triplicate on DNA samples from MOCK (anti-IgG antibody) and transcription factor EGR3 immunoprecipitations to the input IPs.

Promoter reporter vector design

The Htr2a promoter luciferase reporters were generated by Genecopeia Inc. using the dual-reporter system. The “Htr2a proximal promoter” reporter includes genomic DNA from 1061 bp upstream to 200 bp downstream of the Htr2a gene transcription start site (TSS) and contains the putative EGR3 binding sequence GCGCGGGGGAGGGG. The “Htr2a distal promoter” reporter includes genomic DNA from −2727 bp to −2841 bp upstream of the Htr2a TSS and contains the putative EGR3 binding site AGGAGGGGGAGTCT. The control Arc promoter luciferase reporter includes genomic DNA ranging from 1049 bp upstream to 200 bp downstream of the TSS of the Arc gene, which contains a previously validated EGR3 binding site [36]. The “non-promoter” luciferase reporter, containing non-promoter sequence (TGCAGATATCCTCGCCC), was used as a negative control.

Cell culture and transfection

Neuro2a cells (ATCC) were seeded into 6-well plates and grown to 70–90% confluence. 3h prior to transfection, culture medium was replaced with 3ml of fresh medium. For each promoter construct, 1.5μg of CMV-Egr3 vector (or CMV vector control) was co-transfected with 1μg of promoter reporter, using 3.75μl of Lipofectamine 3000 reagent, 5μl of P3000 reagent and 250μl of Opti-MEM per well. Cells were incubated at 37°C in 5% CO2: 95% air. Transfections were conducted in triplicate and replicated twice.

Luciferase signal measurement

Twenty-four hours after transfection, 0.2ml of medium from each cell culture was collected and luciferase activities were measured using the secrete-pair dual luminescence assay kit (Genecopoeia). Each sample was run in duplicate. For all measurements, the GLuc value was first normalized to the internal control SEAP luciferase value (GLuc/SEAP ratio) and then to the non-promoter luciferase reporter. Following collection of medium, cells were processed for protein isolation and Western blot analysis to determine EGR3 protein levels.

NCBI GEO Data Analysis

Published data were obtained from NCBI GEO, Gene Expression Omnibus [37] dataset GSE53987 [38]. The microarray platform file GPL570 was used to ascertain gene-specific probe IDs for EGR3 and HTR2A, which were used to download gene expression data from prefrontal cortex (Brodmann Area 46).

Statistical Analyses

All statistical analyses were carried out using GraphPad prism with significance set at p < 0.05. For statistical analyses, unpaired, two-tailed, Student’s t-tests, two-way analysis of variance (ANOVA) with Bonferroni post-hoc tests, or two-tailed Mann-Whitney U tests were performed to determine significance for conditions in which there were more than two groups or two factors. Outliers were identified and excluded using the ROUT method [39] in GraphPad Prism with a False Discovery Rate (FDR) of 1%. Data were examined graphically within each group, and no strong deviation from normality was observed. Power analyses were performed based on previous data using G*Power (Heinrich-Heine Universität Düsseldorf). Data were plotted as means ± standard error of the mean (SEM).

Results

We had previously reported that SD rapidly upregulates Htr2a expression in the mouse cortex [27]. However, since this study was performed in cortical homogenates, it was unclear in what cortical regions this upregulation was taking place. This is important since levels of both Htr2a mRNA and 5-HT2AR protein display a distinctive anterior-posterior gradient in the rodent cortex [40-42]. This raised the question of whether SD was increasing Htr2a expression in regions where the receptor is already expressed or inducing it de novo in regions where it is normally not expressed. We therefore examined the cortical region in which SD induces Htr2a expression.

We first tested whether we could replicate published in situ hybridization findings showing that SD upregulates Egr3 [26] using quantitative reverse transcription (qRT) PCR. Figure 1A shows the SD protocol and coordinates for regional brain dissection. In WT mice, we found that 6h of SD did not increase Egr3 expression in the most anterior region of frontal cortex (AFC) but did significantly upregulate Egr3 mRNA in the posterior part of the frontal cortex (PFC), as well as in more posterior regions of cortex (labeled “mid-posterior cortex (MPC)) (Fig.s 1B – 1D).

Figure 1.

Sleep deprivation upregulates Egr3 in a region-dependent manner in the frontal cortex.

(A) SD protocol. In WT mice quantitative RT-PCR shows that 6h of SD (B) does not increase Egr3 expression in AFC regions (t27 = 1.956; p = 0.0609; SDc, n = 14; SD, n = 15 ), but significantly upregulates Egr3 mRNA in (C) PFC (t26 = 3.979, p = 0.0005; SDc, n = 14; SD, n = 14) and (D) MPC (t27 = 7.307; p < 0.0001; SDc, n = 15; SD, n = 14) regions, compared to SDc. Unpaired student’s t-test, *** p < 0.001, **** p < 0.0001. Values represent means ± SEM. (Abbreviations: AFC- anterior frontal cortex; PFC- posterior frontal cortex; MPC- mid to posterior cortex; SD- sleep deprivation; SDc- SD control; WT- wildtype; h: hours).

We next examined whether 6h of SD can upregulate Htr2a expression in the same cortical regions, and whether this requires Egr3 (Fig.s 2A – 2C). In the AFC, SD increased Htr2a expression when both genotypes were analyzed together (two-way ANOVA), but not in either WT or Egr3−/− groups independently (post-hoc analyses were not significant) (Fig. 2A). However, in the PFC of WT mice, SD significantly increased Htr2a mRNA compared to SDc, a result that was absent in Egr3−/− mice (Fig. 2B). In the MPC, SD also increased Htr2a mRNA in WT mice, though the relative amount of the increase was less than in the PFC. Again, this effect did not occur in Egr3−/− mice (Fig. 2C).

Figure 2.

Sleep deprivation upregulates Htr2a in an Egr3-dependent, and region-specific, manner.

(A – C) In WT and Egr3−/− mice qRT-PCR shows that 6h of SD (A) increases Htr2a overall in the AFC when both genotypes were analyzed (two-way ANOVA, sig. main effect of SD (F1, 50 = 8.279, p = 0.0059; WT: SDc, n = 15; SD, n = 14; Egr3−/−: SDc, n = 12; SD, n = 13)) but not in either genotype alone (post-hoc analyses showed no sig. differences between genotypes or SD conditions). (B) However, SD significantly upregulates Htr2a expression in the PFC of WT, but not Egr3−/−, mice, compared to SDc (two-way ANOVA, sig. main effect of SD (F 1, 53 = 12.08, p = 0.0010; WT: SDc, n = 15; SD, n = 15; Egr3−/−: SDc, n = 12; SD, n = 15); post-hoc analyses showed a sig. increase of Htr2a mRNA after SD (vs. SDc) in WT mice (p = 0.0088), but not in the Egr3 −/− mice (p = 0.6777)). (C) In the MPC, SD increased Htr2a in WT, but not Egr3−/−, mice, compared to SDc (two-way ANOVA, sig. main effect of SD (F 1,47 = 13.14, p = 0.0007; WT: SDc, n = 15; SD, n = 14; Egr3−/−: SDc, n = 10; SD, n = 12)); post-hoc analyses showed sig. increase of Htr2a mRNA after SD (vs. SDc) in WT mice (p = 0.0205) but not in the Egr3 −/− mice (p = 0.2343). (D) Representative images from n = 3 per group of RNAscope in situ hybridization demonstrating Htr2a expression in SDc and SD WT and Egr3−/− mice. Bonferroni-corrected comparisons: * p < 0.05, ** p < 0.01. Values represent means ± SEM. (Abbreviations: p - post).

To further characterize the regional distribution of Htr2a expression in response to SD, and in the absence of Egr3, we conducted RNAscope in situ hybridization. Figure 2D shows that 6h of SD increases Htr2a expression in WT mice in a region-specific manner. Compared to WT mice, Egr3−/− mice have lower levels of Htr2a mRNA throughout most of the cortex, sparing selected brain regions. SD may produce small increases in some regions of Egr3−/− mice, though this is less consistent than in WT mice.

In WT mice that did not undergo SD (SDc), RNAscope in situ hybridization showed a distribution pattern of Htr2a mRNA similar to prior studies in rats using radioactive in situ hybridization [40, 41]. This includes high levels of expression in the anterior frontal cortex, particularly in a thin layer (L) of cells at the L1/L2 border and in the deep L2/3 – L5 (L4 not present in anterior frontal cortex), with relatively lower signal in L6a. Also consistent with prior studies, Htr2a mRNA levels decrease with progression toward more posterior cortical regions, with notable reductions in L2/3, less drastic reductions in expression in L5, overall minimal L6a expression, and strong expression in L6b bordering the corpus callosum. In the posterior frontal cortex (Bregma +2.1 – +0.14) expression of Htr2a is present in L2/3 in the somatomotor areas (MOs, MOp) and medial somatosensory cortex (SSp), but notably absent from the lateral SSp, and SSs, with posterior progression. This absence of Htr2a in L2/3 in the SSp/s is evident in prior studies in rat as well [40, 41].

In areas outside of the cortex, Htr2a expression in WT SDc animals is seen in the piriform cortex (PIR), olfactory tubercle (OT), taenia tecta dorsal and ventral (TTd/TTv), anterior olfactory nucleus (AON), and substantia innomata (SI) with strong expression in the claustrum (CL) and endopiriform cortex (dorsal) (EPd). Low level Htr2a expression is also seen in patches in the caudate/putamen (CP) as well as in the lateral and medial septum (LS, MS). Low level expression is also seen in the hypothalamus.

Following sleep deprivation in WT mice, the overall level of Htr2a in the anterior frontal cortex (anterior to Bregma +2.1) does not appear to be increased compared to SDc, however the laminar distribution that is apparent in SDc animals becomes less clearly delineated, suggesting a potential increase in expression in some regions and decrease in expression in others. In contrast to the cortex, in this anterior part of the brain SD produces a marked increase in expression in the olfactory regions, including the AON, OT, as well as the PIR (particularly the pyramidal layer, PIR2), EPd, and TTd & TTv, compared to WT SDc mice.

In the PFC (Bregma +2.1 – +0.14) SD increases overall Htr2a expression in WT mice, while also resulting in a more diffuse pattern, with loss of clear laminar signal, compared to SDc. Cortical areas in which SD produces the greatest increases in Htr2a levels include the anterior cingulate (ACAv, ACAd), & somatomotor areas (MOs, MOp), as well as more lateral regions, including the agranular insular area (AI). Notably, Htr2a expression in L6b is unchanged or only slightly increased by SD in WT mice. SD induces diffuse Htr2a expression in the CP, and potentially small increases in the MS and hypothalamus (this was not consistent across different animals in the same group).

Compared with WT mice, in Egr3−/− mice that did not undergo SD (SDc) Htr2a expression is reduced throughout the cortex, an effect that becomes more pronounced with progression in the posterior direction. The exceptions to this Egr3-depencence of Htr2a expression are the cells in L6b, as well as in the CL and EP and PIR, in which expression is only mildly reduced in Egr3−/− mice. Despite the reduced expression, a clear laminar distribution remains in the SDc mice, with most of the residual cortical expression in L5 & L4, and almost no expression evident in L2/3.

Despite the reduced expression of Htr2a in Egr3−/− mice, SD does appear to increase Htr2a mRNA in some brain regions of these animals. This induction is not consistently seen in the anterior frontal cortex but is clear in more posterior cortical regions (most evident in the ACA and MO, the same regions where SD increases expression in WT mice). L6b expression is only mildly increased, if at all, by SD in Egr3−/− mice. In regions outside of the cortex, SD increases Htr2a expression in the EPd, PIR, TT, AON, and OT, in Egr3−/− mice, though not to the levels seen in WT mice. Expression in the CL is not obviously affected by SD in Egr3−/− mice.

To determine if SD also increases 5-HT2AR protein levels, we performed receptor autoradiography with the selective 5-HT2AR antagonist 3H-M100907 on brain sections from WT and Egr3−/− mice at baseline and following 8h SD (to allow time for translation of mRNA) (Fig. 3A). We found that, in the AFC, SD results in an overall increase in 5-HT2AR binding that differs between genotypes, but this increase is not significant in either WT or Egr3−/− genotypes alone (Fig. 3B). However, SD increases WT expression sufficiently to produce a significant difference in 5-HT2AR levels between WT and Egr3−/− mice that is not present in SDc animals (Fig. 3B, comparison between WT and Egr3−/− in the SD groups). In the PFC, 8h of SD significantly increases 5-HT2AR levels in WT mice but not in Egr3−/− mice (Fig. 3C). In addition, 5-HT2AR levels are significantly greater in WT than Egr3−/− mice in this region both at baseline (replicating our prior radioligand binding assay findings [25]), and following SD. In the MPC, where endogenous Htr2a expression is lower than in more anterior cortical regions, SD does not increase 5-HT2AR levels in WT or Egr3−/− mice (Fig. 3D), and levels of 5-HT2AR are lower in Egr3−/− mice than WT mice under at baseline and following SD. Figures 3E–G show representative autoradiographic images from WT and Egr3−/− mice under SDc and SD conditions.

Figure 3.

SD increases 5-HT2AR levels in the PFC of WT mice in an Egr3-dependent manner.

(A) 8h SD protocol. Quantification of 3H-M100907 binding autoradiography shows that SD, compared with SDc: (B) in AFC results in significantly greater 5-HT2AR levels in WT mice than Egr3−/− mice after SD (two-way ANOVA, sig. main effects of SD (F 1, 62 = 4.61, p = 0.0358) and genotype (F 1, 62 = 14.78, p = 0.0003); (C) in the PFC SD significantly upregulates 5-HT2AR levels in WT, but not Egr3−/−, mice (two-way ANOVA, sig. interaction between SD and genotype (F 1, 62 = 4.18, p = 0.0451)). (D) In the MPC, SD did not significantly increase 5-HT2AR levels; notably, 5-HT2ARs were lower in Egr3−/− mice than WT under both basal (SDc) and SD conditions (two-way ANOVA, sig. main effect of genotype (F 1, 62 = 38.79, p < 0.0001)) but not of SD (F 1, 62 = 1.371, p = 0.2461). Representative 3H-M100907 autoradiography images of brain tissue sections from (E) AFC, (F) PFC, and (G) MPC. For experiments in (B-D) WT: SDc, n = 17; SD, n = 16; Egr3−/−: SDc, n = 16; SD, n = 17. Bonferroni-corrected comparisons: * p < 0.05, ** p < 0.01, *** p < 0.001. Values represent means ± SEM.

These data reveal the novel finding that 5-HT2ARs can be upregulated in the PFC in a matter of hours in response to an environmental stimulus, and that this requires Egr3. The results suggest that EGR3, an activity dependent IEG transcription factor, may directly regulate expression of Htr2a in response to environmental events. For this to be happening, EGR3 would have to be expressed in the same cells as Htr2a and EGR3 consensus binding sequences would have to be present in the Htr2a promoter. Our prior study demonstrated that the former is true [27]. To determine the other requirement, we used the FIMO program [35] which revealed two high probability EGR3 binding sites in the Htr2a promoter, a distal binding site at −2777 bp (site A, AGGAGGGGGAGTCT) and a proximal site at −61 bp (site B, GCGCGGGGGAGGGG) upstream of the start ATG (Fig. 4A).

Figure 4.

EGR3 binds to the Htr2a promoter in frontal cortex.

(A) Schematic showing high probability EGR3 consensus binding sites in the Htr2a promoter. (B) Representative Western blot image (n = 3 shown of n = 11 per group, Chameleon Duo Pre-Stained Protein Ladder (Li-COR Biosciences) and (C) Western blot average protein levels, show significant upregulation of activity dependent EGR3 protein following 6h of SD in WT frontal cortex (t20 = 4.778; p = 0.0001; n = 11 per group). (D) ChIP-qPCR shows SD increases binding of EGR3 to Arc promotor (positive control; SDc, n = 8; SD, n = 10) and Htr2a distal promoter (SDc, n = 9; SD, n = 9), but not Htr2a proximal promoter (SDc, n = 9; SD, n = 11), in frontal cortex tissue (Respectively, t16 = 2.802; p = 0.0128, t16 = 2.297; p = 0.0354, and t18 = 0.8463; p = 0.4085). Unpaired student’s t-test, * p < 0.05, *** p < 0.001. Values represent means ± SEM. (Abbreviations: rEGR3 - recombinant EGR3 protein).

To determine whether EGR3 protein binds to these sites in mouse cortex, we conducted chromatin immunoprecipitation (ChIP). As an IEG, Egr3 is expressed in a stimulus dependent manner. Under basal, unstimulated conditions, levels of EGR3 protein in the brain are low, and little binding to the promoters of target genes would be expected. SD, which increases Egr3 mRNA ([26, 27] and Fig. 1), should increase EGR3 protein levels and, we expected, binding to the Htr2a promoter. Figures 4B – 4C show that 6h of SD significantly increases EGR3 protein levels in the frontal cortex of WT mice, measured by Western blot. Next, we conducted ChIP on samples of chromatin from WT mice that underwent 6h of SD compared to SDc mice. We used the promoter region of activity-regulated cytoskeleton associated protein (Arc) that includes a validated EGR3 binding domain as a positive control [36]. Figure 4D shows that SD significantly increased EGR3 binding to the Arc promoter region, as well as to the distal Htr2a promoter, compared to SDc conditions. EGR3 binding to the proximal promoter was not significantly changed following SD.

To confirm that the binding of EGR3 to the Htr2a promoter results in a change in gene expression, we conducted in vitro luciferase-reporter assays. We co-transfected neuro2a cells with luciferase/SEAP constructs driven by either the positive control Arc promoter [36], the distal Htr2a promoter, or the proximal Htr2a promoter, with either a CMV vector overexpressing EGR3, or a control empty CMV vector (Fig. 5A–C). Figure 5D shows Western blot results validating EGR3 overexpression in the cell culture assays. We found that both regions of the Htr2a promoter containing high-probability EGR3 binding sites (Fig. 4A) drive expression of luciferase in response to EGR3 expression. EGR3 expression induces a 4.9-fold increase in positive control Arc promoter-driven luciferase (Fig. 5A) and 3.9-fold increase in the Htr2a distal promoter-driven luciferase (Fig. 5B). In addition, although the proximal Htr2a promoter did not show a statistically significant increase in EGR3 binding in the in vivo ChIP assay (Fig. 4D), in vitro expression of EGR3 induced a significant 4.2-fold increase in Htr2a proximal promoter-driven luciferase signal, compared to CMV vector alone (Fig. 5C). These results suggest that the physiologic stimulus of SD upregulates EGR3, which directly binds to the Htr2a promoter and activates Htr2a expression, results in increased levels of cortical 5-HT2ARs in the mouse brain.

Figure 5.

EGR3 drives gene expression via binding sites in the Htr2a promoter and EGR3 and HTR2A expression is reduced in schizophrenia brains.

(A-C) Schematics of dual luciferase/SEAP reporter constructs containing EGR consensus binding sites and results of in vitro assays in neuro2a cells. CMV-driven EGR3 overexpression significantly upregulates expression of luciferase reporters driven by (A) Arc promoter (t4 = 42.28; p < 0.0001), (B) Htr2aD distal promoter (t4 = 21.17; p < 0.0001), and (C) Htr2aP proximal promoter (t4 = 8.977, p = 0.0009) regions. (D) Western blot validation of EGR3 expression following transfection with CMV-EGR3 versus CVM empty vector, from cultures expressing reporter constructs driven by promoters from Arc, Htr2aD, Htr2aP, or negative promoter control vector. Unpaired student’s t-test, (A-D) n = 3 per group. (E, F) EGR3 and HTR2A mRNA levels are significantly decreased in human brain tissue samples from the prefrontal cortex of schizophrenia patients compared to controls. Microarray (Robust Multi-Array Average) gene expression data derived from NCBI Geo database GSE53987 showing significant decrease in (E) EGR3 (U = 81, p = 0.0331) expression and (F) HTR2A (U = 63, p = 0.0050) expression, in control (n = 19) vs. schizophrenia (n = 15) patients. Mann-Whitney U test, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Values represent means ± SEM. (Abbreviations: CMV - cytomegalovirus, GLuc - Gaussia luciferase, SEAP - secreted alkaline phosphatase).

As discussed above, numerous studies have reported deficient levels of 5-HT2ARs in schizophrenia patients, as well as decreased levels of HTR2A in postmortem patient brain tissue. If our findings showing that EGR3 regulates Htr2a in mice are also true in humans, then the deficiency in HTR2A expression in the brains of schizophrenia patients could be a consequence of reduced expression of EGR3, a gene that has also been found to be reduced in schizophrenia patient brains. To further explore this possibility, we analyzed the results of a published gene expression dataset from postmortem prefrontal cortex tissue samples from schizophrenia patient and control brains in the NCBI GEO Database [38]. Figure 5E and 5F show that both EGR3 and HTR2A expression are reduced in the schizophrenia subject samples compared to controls.

Discussion

The results we report here reveal several novel mechanisms of how 5-HT2AR levels in the brain are regulated. Our findings demonstrate that a physiological stimulus, SD, rapidly upregulates expression of both Htr2a expression and levels of membrane bound 5-HT2AR in the mouse brain in a matter of several hours. They also demonstrate that this environmentally induced upregulation requires the activity dependent IEG transcription factor EGR3, which binds to the Htr2a promoter in the frontal cortex in vivo and is able to activate expression of Htr2a promoter-driven reporter constructs in vitro. Thus, these results identify a direct transcriptional regulator of the 5-HT2AR in the frontal cortex and reveal a previously unrecognized characteristic of 5-HT2AR regulation; that it is environmental-stimulus responsive.

Our finding that 5-HT2ARs can be rapidly upregulated in the mouse frontal cortex in response to SD is supported by a study in humans that employed a similar stimulus. Elmenhorst and colleagues conducted a PET study using the 5-HT2AR selective radioligand [18F]altanserin to determine the effects on 24h of SD on 5-HT2AR levels in the frontal cortex of healthy subjects. Comparing PET scans conducted following a night of normal sleep to scans conducted the following day, after 24h of total sleep deprivation, they found a 9.6% increase of [18F]altanserin binding in neocortical regions, including the medial inferior frontal gyrus, insula, and anterior cingulate, parietal, sensomotoric, and ventrolateral prefrontal cortices [28]. Their study was the first investigation of the effects of SD on cerebral cortex 5-HT2AR levels in either humans or animals [28]. In discussing the possible mechanisms that may underlie their finding, the authors point to the fact that SD increases brain levels of serotonin (5-HT) in rodents [43]. However, the effects of serotonin, and other agonists, on 5-HT2AR levels are complicated, with reports showing both upregulation and downregulation in different systems and in a manner that may be influenced by concentration or dosage [44]. Our findings suggest, instead, that this increase in 5-HT2AR levels may result from direct upregulation of Htr2a by the IEG EGR3, which has been shown to be activated in the cortex of mice by 6h of SD by the Allen Institute for Brain Research group [26] and replicated by our laboratory here and in our prior study [27].

We chose to use SD as a stimulus in the current study because it is an environmental intervention that we and others have shown increases Egr3 expression [26, 27] and thus allowed us to test our hypothesis that induction of EGR3 should upregulate expression of Htr2a. As a class, IEGs are activated in the brain in response to a wide range of stimuli, including many types of stress [45]. However, the expression of Egr3 in response to specific stressors has been less well investigated than that of many other IEGs. So, regardless of whether SD causes mice to experience “stress”, it is an effective intervention that demonstrates how rapidly an environmental stimulus can upregulate 5-HT2AR mRNA and protein levels in the brain; in just 6–8 hours.

Roles of 5-HT2ARs and Egr3 in stress response and memory

A significant body of research has suggested that stress influences 5-HT2AR expression. Numerous types of stress, including physical stressors of immobilization, chronic forced swimming and toe pinch, social stress, including maternal separation, as well as in utero exposure to lipopolysaccharide (to mimic bacterial infection), have been found to elevate levels of 5-HT2AR mRNA and protein, as well as to increase the frequency of 5-HT2AR-mediated behaviors (head-twitch response to 5-HT2AR agonists) (reviewed in Table 1 of [46]). Though not all types of stress produce this effect [46]. These findings suggest that increased expression of 5-HT2ARs may play a role in the brain’s response to environmental events, particularly stressful ones.

One of the critical responses of the brain to environmental events is the formation of memories. In fact, numerous studies suggest a role for 5-HT2ARs in memory formation. 5-HT2ARs facilitate associative learning (reviewed in Table 2 of [46]), play a role in the retrieval of recognition memory [47], and modulate reconsolidation of contextual recognition memory [48]. If EGR3 is regulating the increase in 5-HT2ARs expression in response to environmental stimuli, this suggests that the Egr3 gene should also be required for memory formation. In fact, our prior studies showed that Egr3 deficient (−/−) mice fail to habituate to a startling stimulus (demonstrating deficits in associative memory), display social interaction responses consistent with an inability to remember familiar mice (including persistent, non-habituating social investigation and increased aggressive behavior), and show deficits in Y-maze navigation indicative of defects in spatial memory formation [49].

We also found that Egr3−/− mice show a heightened response to stress [49]. While at first this may appear inconsistent with the literature indicating that 5-HT2ARs are upregulated in response to stress, the findings are actually aligned. Without Egr3, mice fail to remember having experienced events, such as being handled by an investigator or having encountered mice they have lived with their whole lives, and they respond with the same high level of reactivity, and cortisol release, as mice being exposed for the first time [49]. Thus, the findings that increased 5-HT2AR levels are important for associative and contextual learning, and the fact that Egr3−/− mice have deficient levels of 5-HT2AR that fail to normally upregulate in response to environmental stimuli ([25, 27] and current results), provide a potential explanation for at least part of the heightened stress response of Egr3−/− mice [49]. More research will be necessary to determine if Egr3 mediates the change in 5-HT2AR levels in response to other types of stress.

Regulation of Htr2a expression

Studies examining the mechanisms by which 5-HT2AR levels are regulated have largely focused on effects of agonist and antagonist actions. These studies demonstrate the complex nature of 5-HT2AR regulation. The endogenous ligand 5-HT, as well as both agonists and antagonists, can trigger rapid internalization of 5-HT2ARs [44, 50, 51]. Though some studies have found that specific agonists can induce processes such as receptor desensitization without altering surface density [52], and other studies have shown a paradoxical increase in receptor density following antagonist exposure in specific cell types [53].

Yet, little is known about transcriptional regulation of Htr2a, or that environmental stimuli rapidly alter 5-HT2AR levels. One study employed bioinformatic resources and structural equation modeling to analyze the putative effect of a polymorphism in the human HTR2A promoter on transcription factor binding in subjects with chronic fatigue syndrome [54]. Their findings suggested a diagnosis-specific interaction between methylation of an E47 transcription factor binding site and HTR2A expression levels in peripheral blood mononuclear cells in a manner influenced by cortisol level. However, the functionality of these transcription factor binding sites was not examined in that study [54]. Thus, our findings are novel in identifying direct regulation of Htr2a by an activity dependent IEG.

5-HT2ARs in schizophrenia

5-HT2ARs are abundantly expressed in the neocortex and play important roles in cognition, mood and sleep, processes that are disrupted across numerous psychiatric disorders. Investigations of drugs that act as agonists at 5-HT2ARs have recently experienced a resurgence in the search for treatments for severe psychiatric symptoms ranging from depression [55] to anxiety disorders such as PTSD [56-58]. Yet decades of clinical practice and research have demonstrated a particular importance of 5-HT2ARs in schizophrenia.

Schizophrenia is characterized by abnormalities in perception, thinking and memory (exemplified by hallucinations, delusions and cognitive deficits). The fact that 5-HT2ARs mediate the hallucinogenic effects of numerous drugs including LSD, psilocybin, and mescaline [5, 6] suggest the possibility that this receptor may also influence the hallucinations and perceptual disturbances of schizophrenia. One of the most important discoveries that demonstrated a critical role for 5-HT2ARs in schizophrenia symptomatology was the discovery that clozapine, one of the most effective antipsychotic medications to date, binds with high affinity to this receptor [59]. In fact, 5-HT2AR binding is an essential feature of second-generation antipsychotics, which were modeled after clozapine and are the first-line of treatment for psychotic disorders [8, 59]. Furthermore, the 5-HT2AR inverse agonist pimavanserin is effective for treatment of psychosis [7]. Finally, as far back as 1976 numerous post-mortem and in vivo studies have revealed that 5-HT2AR (or 5-HT2R) levels are reduced in schizophrenia patients’ brains, and specifically in the prefrontal cortex [14, 15, 19, 22, 60, 61].

The prefrontal cortex has long been hypothesized to be a central brain region involved in schizophrenia pathogenesis [62, 63]. Activity in this region is essential for spatial working memory, dysfunction in which may contribute to the cognitive deficits that characterize schizophrenia (reviewed in [64]). Our results show that the environmental stimulus of SD increases levels of Egr3 and 5-HT2AR mRNA and protein in the antero-posterior domains of mouse cortex corresponding to the human prefrontal cortex, particularly the anterior cingulate cortex (ACAv, ACAd) [65]. These are also regions where we found that SD-induced 5-HT2AR expression is Egr3-dependent.

The fact that this process is mediated by an activity-dependent IEG transcription factor suggests the intriguing possibility that the reduced 5-HT2AR levels reported in schizophrenia patients may be a consequence of disrupted neural activity, resulting in insufficient activation of IEGs including EGR3. Such a hypothesis is intriguing in the context of recent findings that neural circuit connectivity, and the synchrony of neural oscillations, between the prefrontal cortex and other brain regions is disrupted in schizophrenia patients [66-68]. It is also supported by findings of abnormal IEG expression in patients’ brains ([69] and Fig. 5E).

In further support of this hypothesis, numerous EGR family genes are associated with risk for schizophrenia [70, 71]. Although genome wide association studies (GWAS) have recently identified numerous loci believed to increase genetic risk for illnesses like schizophrenia, the mechanism by which environment may interact with these regions remains elusive. It is notable that, EGR1, EGR4, and NAB2 (a transcriptional co-regulator that alters gene expression via binding to the EGRs) each map to one of the 145 GWAS loci for schizophrenia [71]. Although EGR3 itself is not within a GWAS locus, it interacts in co-regulatory feedback loops with the EGRs and NAB2 [70, 72], and EGR3 expression is reduced in the brains of schizophrenia patients [73, 74], including in our analyses of data from the NCBI GEO database, which shows reduced levels of both EGR3 and HTR2A mRNA in the prefrontal cortex of schizophrenia patients compared with controls (Fig. 5E and 5F) [38].

Another feature of schizophrenia, although less commonly discussed, is sleep disruption, which includes fragmentation and overall decreased duration of sleep [75, 76]. Our finding that SD increases expression of Egr3 and 5-HT2AR mRNA and protein may initially appear inconsistent with the numerous reports that expression of these genes is reduced in schizophrenia patients, since sleep disruption and loss is a common characteristic of this mental illness [75, 76]. There could be several explanations for this. First, as an IEG, expression of Egr3 is rapidly increased in response to a stimulus and returns to baseline within hours of the stimulus [77]. So, the timing of the study, or time of death (in the case of post-mortem studies), may influence whether the gene expression consequences of sleep disruption will still be detectable. However, an explanation we feel is more likely is that the altered neural activity in the prefrontal cortex of schizophrenia patients, which may be related to documented deficiencies in synaptic density [78] and abnormal circuit function [64, 66–68], could result in deficient induction of EGR3 in response to environmental stimuli, including SD. Our data suggest this would consequently result in deficient upregulation of 5-HT2ARs. Finally, genetic variations that affect expression of EGR family genes may also reduce the stimulus-dependent upregulation of 5-HT2ARs. The literature suggesting that this upregulation plays an important role in contextual and associative memory formation [46] suggests that such insufficient 5-HT2AR expression might contribute to the cognitive deficits of schizophrenia.

Together, these findings suggest that dysfunction in activity dependent EGR family IEGs, which include, and result in, decreased activity of EGR3, may contribute to the reported deficits in 5-HT2AR expression in schizophrenia patient brains. These findings thereby shed light on a potential mechanism whereby environment interacts with genetic variations to influence neurobiology that may contribute to the symptoms, and treatment, of neuropsychiatric illness.