Promoter IV-BDNF deficiency disturbs cholinergic gene expression of CHRNA5, CHRM2, and CHRM5: effects of drug and environmental treatments

Abstract

Brain-derived neurotrophic factor (BDNF) promotes maturation of cholinergic neurons. However, how activity-dependent BDNF expression affects specific cholinergic gene expression remains unclear. This study addressed this question by determining mRNA levels of 22 acetylcholine receptor subunits, the choline transporter (CHT), and the choline acetyltransferase (ChAT) in mice deficient in activity-dependent BDNF via promoter IV (KIV) and control wild-type mice. Quantitative RT-PCR revealed significant reductions in nicotinic acetylcholine receptor alpha 5 (CHRNA5) in the frontal cortex and hippocampus and M5 muscarinic acetylcholine receptor (CHRM5) in the hippocampus, but significant increases in M2 muscarinic acetylcholine receptor (CHRM2) in the frontal cortex of KIV mice compared to wild-type mice. Three-week treatments with fluoxetine, phenelzine, duloxetine, imipramine, or an enriched environment treatment (EET) did not affect the altered expression of these genes except that EET increased CHRNA5 levels only in KIV frontal cortex. EET also increased levels of CHRNA7, CHT, and ChAT, again only in the KIV frontal cortex. The imipramine treatment was most prominent among the four antidepressants; it upregulated hippocampal CHRM2 and frontal cortex CHRM5 in both genotypes, and frontal cortex CHRNA7 only in KIV mice. To our knowledge, this is the first evidence that BDNF deficiency disturbs expression of CHRNA5, CHRM2, and CHRM5. Our results suggest that promoter IV-BDNF deficiency – which occurs under chronic stress – causes cholinergic dysfunctions via these receptors. EET is effective on CHRNA5, while its compensatory induction of other cholinergic genes or drugs targeting CHRNA5, CHRM2, and CHRM5 may become an alternative strategy to reverse these BDNF-linked cholinergic dysfunctions.

1. Introduction

Brain-derived neurotrophic factor (BDNF) is a major neuronal growth factor in the brain (Thoenen 1995, Barde 1990). Its expression is induced by neuronal activity (Zafra et al. 1991, Timmusk et al. 1993, Tao et al. 1998, Shieh et al. 1998) and is thought to contribute to activity-dependent neuronal maturation and synaptic function (Thoenen 1995, Poo 2001, Lu 2003, Sakata 2011). The details of how activity-induced BDNF expression modulates neurotransmitter systems remain to be established (Lu 2003, Sakata 2011).

BDNF expression is controlled by at least nine promoters in both humans (Liu et al. 2005, Pruunsild et al. 2007) and rodents (Liu et al. 2006, Aid et al. 2007). Among the nine promoters, promoter IV (classified as promoter III prior to 2007) is most responsive to neuronal activity to induce BDNF expression (Timmusk et al. 1993, Tao et al. 1998, Shieh et al. 1998). However, promoter IV undergoes epigenetic inactivation by chronic stress or early-life maltreatment (Tsankova et al. 2006, Fuchikami et al. 2009, Roth et al. 2009). Reductions in BDNF levels and inactivation of promoter IV have been observed in many cognitive and emotional disorders, such as schizophrenia (Weickert et al. 2003, Hashimoto et al. 2005, Wong et al. 2010), depression (Dwivedi et al. 2003, Keller et al. 2010, Hing et al. 2012), and Alzheimer's disease (Phillips et al. 1991, Murray et al. 1994, Ferrer et al. 1999, Holsinger et al. 2000, Garzon et al. 2002). To investigate the consequent dysfunctions caused by deficiency of activity-dependent expression of BDNF, we have generated knock-in mice that lack promoter IV-driven BDNF expression (KIV) (Sakata et al. 2009). Using KIV mice, we recently showed causal evidence that a deficiency of activity-dependent promoter IV-driven BDNF expression (promoter IV-BDNF) leads to depression-like behavior and cognitive dysfunctions, including decreased exploratory activity, stress-induced despair, anhedonia, anxiety-like behavior, inflexibility in reversal learning and fear extinction, and impaired response-inhibition (Sakata et al. 2010, Sakata et al. 2013a).

In KIV mice, activity-dependent expression of BDNF protein is particularly abolished in the CA1 region of the hippocampus (Sakata et al. 2013a) and the medial prefrontal cortex (mPFC) (Sakata et al. 2009). The basal levels of BDNF protein (without any stimulation) are also reduced in the hippocampus and frontal cortex (53% and 41% reduction, respectively), which likely reflect the lack of promoter IV-BDNF in response to the basal activity of animals (moving, grooming, food taking, etc.) (Sakata et al. 2009, Sakata et al. 2013a). The reductions may also reflect reduced BDNF expression driven by other promoters (Martinowich et al. 2011) due to reduced feed-forward BDNF transcription induced by BDNF itself (Yasuda et al. 2007, Sakata 2011, Sakata 2014) and possible promoter interaction (Maynard et al. 2016), while the transcriptional functions of other BDNF promoters are intact (Sakata et al. 2009, Jha et al. 2011, Sakata et al. 2013b). KIV mice lack promoter IV-BDNF from the start of life. This condition mimics the epigenetic condition caused by maternal depression and early-life stress that reduces promoter IV-BDNF and lasts throughout the lifespan and across generations (Roth et al. 2009, Onishchenko et al. 2008). KIV mice show no obvious structural changes in both hippocampus and frontal cortex at the young adult stage (Sakata et al. 2009, Sakata et al. 2013a). However, they show impaired long-term synaptic plasticity (>3h) in the hippocampal CA1 region (Sakata et al. 2013a) and aberrant spike-timing dependent plasticity in the mPFC (Sakata et al. 2009). The promoter IV-BDNF deficiency impairs GABAergic functions (Hong et al. 2008) particularly in the mPFC (Sakata et al. 2009), while keeping basic glutamatergic functions intact (Sakata et al. 2009, Sakata et al. 2013a). This deficiency also disturbs expression of GABA-related (Guilloux et al. 2012, Tripp et al. 2012, Maynard et al. 2016) and monoamine-related genes (Sakata & Duke 2014).

One neurotransmitter system that is likely affected by activity-dependent BDNF expression, but is still unclear, is the cholinergic system. Acetylcholine is implicated in cognitive functions and reward/motor activity (Sarter & Parikh 2005) and is a treatment target for many psychiatric and neurological disorders, such as schizophrenia, depression, attention deficit hyper activity (ADHD), and Alzheimer's disease (Levin et al. 2006, Taly et al. 2009, Parikh et al. 2016, Lewis & Picciotto 2013, Olincy et al. 2006, Court et al. 2001, Hellstrom-Lindahl et al. 2004, Srivareerat et al. 2011). Previous studies have shown reciprocal regulation between acetylcholine and BDNF. Basal forebrain cholinergic neurons that release acetylcholine project to the cortex and hippocampus (Sarter & Parikh 2005), where BDNF mRNA is expressed and regulated by acetylcholine input (Phillips et al. 1990, Lindefors et al. 1992, Lapchak et al. 1993, Mahy et al. 1996, Berchtold et al. 2002). For example, hippocampal BDNF mRNA expression is induced by activation of muscarinic acetylcholine receptors (mAChRs) (Lapchak et al. 1993, Knipper et al. 1994) and by subchronic activation of nicotinic acetylcholine receptors (nAChRs) (Kenny et al. 2000, Hellstrom-Lindahl et al. 2004, Srivareerat et al. 2011), whereas BDNF expression is decreased by acute activation of nAChRs (Kenny et al. 2000). By contrast, BDNF promotes survival and differentiation of cholinergic neurons (Knusel et al. 1991, Klein et al. 1999) and the release of acetylcholine (Auld et al. 2001, Knipper et al. 1994). Cooperation between cholinergic and glutamatergic receptors has been found to be essential for the induction of BDNF-dependent long-lasting synaptic plasticity (Navakkode & Korte 2012).

The importance of BDNF in the cholinergic system has been reported for more than two decades. Yet, BDNF regulation of the specific cholinergic gene expression remains unclear. The cholinergic system is divergent, having 5 muscarinic and 17 nicotinic receptors that each has distinct functions and distributions (Boulter et al. 1986, Caulfield 1993, Gotti et al. 2006, Taly et al. 2009). Only limited reports have shown effects of BDNF on expression of individual cholinergic genes: BDNF null deficiency reduces choline acetyltransferase (ChAT) expression in the medial septum (Ward & Hagg 2000) and BDNF reduces (Fernandes et al. 2008) or increases alpha7 nAChRs expression in hippocampal interneurons (Massey et al. 2006). The effects of deficiency of activity-dependent BDNF expression on the diverse cholinergic molecules remain unknown. In the present study, we addressed this issue by determining changes in mRNA levels of 22 acetylcholine receptors, the choline transporter (CHT), and the acetylcholine synthesizing enzyme (ChAT) in KIV mice, compared to control wild-type (WT) mice using qRT-PCR. We focused on the frontal cortex and hippocampus, because these are the target regions of basal forebrain cholinergic neurons (Sarter & Parikh 2005, Phillips et al. 1990) and the regions where defective promoter IV largely abolish activity-driven BDNF expression (Sakata 2011, Sakata et al. 2009, Sakata et al. 2013a). We also explored the effects of chronic drug and environmental treatments on the affected expression of acetylcholine-related genes. We used four different types of antidepressants – fluoxetine (a selective serotonin reuptake inhibitor: SSRI), phenelzine (a monoamine oxidase inhibitor), duloxetine (a serotonin–norepinephrine reuptake inhibitor: SNRI), and imipramine (a tricyclic antidepressant: TCA) – or an enriched environment treatment (EET). This was because we have previously shown that these treatments partially or completely normalize depression-like behavior of KIV mice, while EET, but not antidepressant treatment, normalizes hippocampal BDNF levels (Jha et al. 2011, Sakata et al. 2013b, Jha et al. 2016). We asked whether changes in acetylcholine gene expression were involved in the treatment effects.

2. Materials and Methods

2.1. Animals

The generation of BDNF promoter IV knock-in (KIV) mice has been described previously (Sakata et al. 2009). Briefly, the BDNF promoter/exon IV DNA region was cloned and a green fluorescent protein (GFP) gene with a stop codon was inserted into exon IV to disrupt promoter IV-driven expression of BDNF protein. The construct was used for homologous recombination in 129/sv ES cells (from NIH). The ES cells and C57BL/6J blastocytes were used to generate chimeric mice, which were crossed to C57BL/6J females for >12 generations. Adult (2–3 months old) gender-matched pairs of KIV and control WT mice were used. Five pairs (2 males and 3 females for each genotype) were used for initial screening. Ten pairs of another cohort (5 males and 5 females for each genotype) were used to confirm effects of promoter IV-BDNF deficiency. Six pairs of another cohort (3 males and 3 females for each genotype) were randomly assigned to each treatment using randomization table. The sample size was determined based on the previous studies (Jha et al. 2011, Sakata et al. 2013b, Sakata & Duke 2014). All animals were housed in a group in a normal 12:12 h dark-light cycle and had ad libitum access to food and water. All animal experiments were approved by the University of Tennessee Laboratory Animal Care and Use Committee and were conducted in accordance with NIH guidelines.

2.2. Drug treatment

Mice received 21 days of four different types of antidepressants as described previously (Sakata et al. 2013b). Briefly, mice received a single intraperitoneal (i.p.) injection per day with saline or with one of the four different classes of antidepressant: fluoxetine (10 mg/kg, Mallinckrodt, Inc., Hazelwood, MO), phenelzine (10 mg/kg, Sigma, St. Louis, MO), duloxetine (20 mg/kg, Eli Lilly, Indianapolis, IN), and imipramine (10 mg/kg, Sigma). Animals were sacrificed 3 h after the last injection. The method of drug administration (i.p.), the dosage, and the sacrifice timing were decided based on previous studies (Dwivedi et al. 2006, Khundakar & Zetterstrom 2006, Molteni et al. 2009, Mannari et al. 2008, Tsankova et al. 2006, Nibuya et al. 1995).

2.3 Enriched environment treatment (EET)

Mice received 21 days of standard condition treatment (SCT) or EET, as described previously (Jha et al. 2011). Briefly, SCT consisted of a regular cage (27×16×12 cm) containing 2–5 mice (group-housed to avoid isolation stress). EET consisted of a larger cage (44×22×16 cm) containing one plastic running wheel per 5 mice to increase physical exercise, an assortment of toys (igloo, dome, balls, tunnels, etc.) weekly changed, Foraging Crumble (Bio-Serv, Frenchtown, NJ) to increase perception/mental exercise, and 5–10 company mice with nesting material to increase social interaction.

2.4. Tissue collection, RNA extraction, and qRT-PCR

Each animal was euthanized by i.p. injections of 90 mg/kg ketamine and 10 mg/kg of xylazine. The hippocampus and frontal cortex were removed and immediately frozen, within 6 min after the euthanasia, and stored at −80°C until RNA extraction. Total RNA was extracted from the tissues using QIAshredder (Qiagen, Valencia, CA) and RNeasy Kit (Qiagen) with on-column DNase (Qiagen), and quantified using the NanoDrop spectrophotometer (Agilent Technologies, Santa Clara, CA), as described previously (Sakata & Duke 2014). One microgram of total RNA was reverse transcribed into single-strand cDNA using the Transcriptor First Strand cDNA Synthesis Kit (Roche Applied Science, Indianapolis, IN). qRT-PCR was performed using BioMark, as described previously (Reiner et al. 2012, Sakata & Duke 2014). Briefly, a 0.5 µl sample of the cDNA (from 25 ng of total RNA) was preamplified for 14 cycles, then the diluted (1:5) reactions were used for qPCR using 96.96 dynamic array chips (Fluidigm, South San Francisco, CA). The cycling program was 2 min at 50 °C and 10 min at 95 °C, followed by 40 cycles of 95 °C for 15 s and 1 min at 60 °C. The cycle threshold (Ct) values were obtained using the BioMark Gene Expression Data Analysis after automatic inspection for quality. Relative gene expression values were determined by using the 2^−ΔΔCt method of Livak and Schmittgen (Livak & Schmittgen 2001). The average Ct values of cyclophilin D and HGPRT were used as a reference; these housekeeping genes showed the least deviation in Ct values among the samples when six housekeeping genes (cyclophilin D, HGPRT, TBP, β-actin, β-tubulin, and S19) were tested in a pre-assay. Genotype effects were measured by normalizing each gene expression value of KIV mice to the average of the gene expression values of WT mice (% expression of WT). Treatment effects were determined by normalizing each gene expression value to the average of the gene expression values of their respective controls (% expression of saline controls for antidepressant treatment and of SCT for EET). The information on primers and probes is described in Supplementary Table 1.

2.5. Statistical analysis

The data were analyzed using Prism (GraphPad Software, Inc., La Jolla, CA). Student’s t-tests were performed on the two data groups comparing KIV vs. WT. To avoid type I errors (false positive) for measuring expression of many (24) genes, we confirmed the results by repeating the experiment with another cohort of mice (N=10 per groups). The D'Agostino-Pearson omnibus normality tests were performed to analyze the data distribution for each tested gene in each genotype. All the data were normally distributed (P>0.05). Two-way analyses of variance (ANOVA) were performed for detecting treatment and genotype effects, followed by post hoc Bonferroni multiple comparisons. Bonferroni correction would cause type II errors (false negative) when using a large number of testing groups (i.e., treatments); detection p-values are divided by the group number. To avoid type II errors, Student’s t-tests were also performed on the two data groups comparing control vs. treatment. Our specific hypotheses were tested when significant main effects were seen by two-way ANOVA, but post hoc Bonferroni analyses showed no significance. No discrepancies were found between Bonferroni's and Student’s t-tests except for imipramine effects on hippocampal CHRNA7 and frontal cortex CHT in WT mice (see results). These inconclusive data were further analyzed by the adjusted p-values (i.e., q-values) with the false discovery rate (FDR) (Storey and Tibshirani, 2003) using the classical one-stage method (Benjamini and Hockberg, 1995). Using pre-established criteria, we excluded data that showed greater than two standard deviations away from the mean to avoid data from technical errors. Statistical differences were considered significant at P<0.05. Data are presented as mean ± standard error of mean (SE).

3. Results

3.1. Detecting expression of 24 cholinergic genes

We first examined the mRNA levels of 24 cholinergic genes in the frontal cortex and hippocampus of KIV and control WT mice. These genes were muscarinic acetylcholine receptors 1–5 (CHRM1-5), nicotinic acetylcholine receptors α1–10 (CHRNA1-10), nicotinic acetylcholine receptors β1–4 (CHRNB1-4), nicotinic acetylcholine receptors δ, ε, and γ (CHRND, CHRNE, and CHRNG), the choline transporter (CHT), and the choline acetyltransferase (ChAT). When compared to WT mice, KIV mice showed a significant increase in CHRM2 expression in the frontal cortex (P<0.05), a significant decrease in CHRM5 expression in the hippocampus (P<0.05), and significant decreases in CHRNA5 in both frontal cortex (P<0.01) and hippocampus (P<0.01) (Supplementary Table 2, N=5 per group). No significant differences were observed between KIV and WT mice for the expression of the other 17 cholinergic genes in either the frontal cortex or the hippocampus, and no mRNA expression of nicotinic acetylcholine receptors δ, ε, and γ 2b was detected (Supplementary Table 2, N=5 per group).

3.2. Confirmation of our findings on CHRM2, CHRM5, and CHRNA5

To avoid type I errors (false positives) by measuring expression levels of a large number (24) of genes, we confirmed our results of expression changes of CHRM2, CHRM5, and CHRNA5 using another cohort of mice. We also re-measured the expression of ChAT, CHT, and CHRNA7 using a larger number of samples (N=10 per group: 5 males and 5 females) to avoid type II errors (false negatives) and to confirm our results of no expression changes in comparison to previous studies that had shown BDNF effects on ChAT (Ward & Hagg 2000, Klein et al. 1999) and CHRNA7 (Fernandes et al. 2008, Massey et al. 2006). The secondary BioMark qRT-PCR experiment reproduced our results of significant increases in CHRM2 expression in the frontal cortex (27% increase, P<0.05), significant decreases in CHRM5 in the hippocampus (30% decrease, P<0.01), and significant decreases in CHRNA5 in both frontal cortex (46% decrease, P<0.001) and hippocampus (21% decrease, P<0.01) in KIV mice when compared to WT mice (Figure 1, N=10). These results indicated that deficiency of activity-dependent promoter IV-driven BDNF expression disturbs expression of these three cholinergic receptors in these target regions of the basal forebrain cholinergic neurons. No significant differences between genotypes were observed again in mRNA expression levels for CHRNA7, CHT, and ChAT (P>0.05, Figure 1). No gender-specific effects were detected in any genes tested (males vs. females: P>0.05 for CHRM2, CHRM5, CHRNA5, CHRNA7, CHT, and ChAT by Students's t-test; Supplementary Table 3).

Figure 1.

Effects of deficiency of activity-dependent promoter IV-BDNF expression on selected cholinergic genes. KIV mice showed significantly increased mRNA levels of CHRM2 and significantly reduced mRNA levels of CHRNA5 in the frontal cortex (left) and significantly reduced levels of CHRM5 and CHRNA5 in the hippocampus, when compared with WT mice (N=10 pairs). Results are expressed as mean ± S.E. *P<0.05; **P<0.01; ***P<0.005 by Student's t-test.

3.3. Treatment effects

We next asked whether expression levels of the affected cholinergic genes might be changed by chronic (21 days) antidepressant treatment (fluoxetine, phenelzine, duloxetine, and imipramine) or EET, because these treatments can partially or fully normalize depression-like behavior of KIV mice (Sakata et al. 2013b, Jha et al. 2011). Our hypothesis was that these treatments might reverse the alteration in mRNA levels that serve as underlying mechanisms of the behavioral changes.

First, we examined whether the treatments reduced CHRM2 expression, particularly in the frontal cortex of KIV mice. Two-way ANOVA showed significant effects of drug treatment on CHRM2 levels in the frontal cortex [F_(4,42)=4.11, P<0.01] and the hippocampus [F_(4,46)=12.9, P<0.001]. However, post hoc tests revealed that no treatment significantly affected CHRM2 expression in the frontal cortex of either genotype (P>0.05 for each, Figure 2a left, see Supplementary Table 4 for statistics). In the hippocampus, imipramine treatment increased CHRM2 expression in both genotypes (at least P<0.01 for each, Figure 2a right).

Figure 2.

Effects of chronic drug and environmental treatments on mRNA expression of (a) CHRM2, (b) CHRM5, and (c) CHRNA5. Both WT and KIV mice received 21 days of four different types of antidepressants – fluoxetine (Flx), phenelzine (Phe), duloxetine (Dul), or imipramine (Imi) – or enriched environment treatment (EET). Expression changes due to the treatments are showed as % control for each genotype. Controls were saline treatment for the antidepressant treatments and a standard environmental condition treatment for EET. No treatment effects were observed except for significant inductions of CHRM2 and CHRM5 by imipramine in the hippocampus and frontal cortex, respectively, in both WT and KIV mice, and a significant induction of CHRAN5 by EET in the frontal cortex of KIV mice (N=4–6 pairs). Bonferroni-corrected significance: *P<0.05; **P<0.01; ***P<0.005 by Two-way ANOVA with Bonferroni post hoc test.

Second, we examined whether the treatments increased CHRM5 expression, particularly in the hippocampus of KIV mice. No significant treatment effects on CHRM5 expression were observed in the hippocampus of either genotype (P>0.05), while duloxetine showed a genotype differences (P<0.05, WT vs. KIV, Figure 2b right). On the other hand, a significant treatment effect [F_(4,40)=8.9, P<0.001] was found in the frontal cortex, where imipramine significantly increased CHRM5 expression in both WT and KIV mice (at least P<0.05 for each, Figure 2b left).

Third, we examined whether the treatments increased CHRNA5 expression in the frontal cortex and hippocampus of KIV mice. Two-way ANOVA showed significant effects of environmental treatment [F_(1,12)=5.22, P<0.05], genotype [F_(1,12)=7.12, P<0.05], and a treatment × genotype interaction [F_(1,12)=5.47, P<0.05] on frontal cortex CHRNA5. Post hoc tests revealed that EET significantly increased CHRNA5 expression in the frontal cortex of only KIV mice (P<0.05, Figure 2c left), resulting in a significant genotype difference in the effect (P<0.0.5). No significant treatment effects were observed other than the EET effects.

Fourth, we examined treatment effects on mRNA expression of CHRNA7, CHT, and ChAT, although the basal expression levels of these genes did not differ between genotypes. Interestingly, significant effects of environmental treatment were found on CHRNA7 [F_(1,12)=17.3, P<0.01], CHT [F_(1,14)=7.6, P<0.05], and ChAT [F_(1,14)=11.8, P<0.01] in the frontal cortex. EET significantly increased expression of these genes only in the KIV frontal cortex (at least P<0.05 for each, Figure 3a, b, and c, left), where a genotype-specific difference was observed in the EET effects on CHRNA7 (P<0.001). Imipramine also significantly increased CHRNA7 levels in the KIV frontal cortex (P<0.05, Figure 3a, left). Two-way ANOVA and post hoc Bonferroni tests showed no other significant effects, although Student t-test showed that imipramine significantly decreased levels of hippocampal CHRNA7 (P<0.05) and frontal cortex CHT (P<0.05) in WT mice. These inconclusive results were further assessed by FDR-adjusted p-values (q-values). q-values showed 0.31 for hippocampal CHRNA7 and 0.07 for frontal cortex CHT, which indicated a false positive and a trend of effects, respectively.

Figure 3.

Effects of 21 days fluoxetine (Flx), phenelzine (Phe), duloxetine (Dul), or imipramine (Imi) drug treatment or enriched environmental treatment (EET) on mRNA expression of (a) CHRNA7, (b) CHT, and (c) ChAT. Expression changes are shown as % control of a saline treatment for the antidepressant treatments and % control of a standard environmental condition treatment for EET. Note that EET significantly induced CHRNA7, CHT, and ChAT in the frontal cortex, but only in KIV mice, and that imipramine significantly induced CHRNA7 in the frontal cortex of KIV mice, but significantly decreased CHRNA7 in the hippocampus and CHT in the frontal cortex of WT mice (N=4–6 pairs). Bonferroni-corrected significance: *P<0.05; **P<0.01; ***P<0.005 by Two-way ANOVA with Bonferroni post hoc test.^#P<0.05 by Student's t-test (q=0.07).

4. Discussion

4. 1. Effects of promoter IV-BDNF deficiency

Nicotinic acetylcholine receptor (nAChR) α5: The most significant finding in the present study was that deficiency of promoter IV-driven BDNF expression lead to significant reductions in mRNA levels of CHRAN5 in both the frontal cortex and hippocampus, suggesting its receptor dysfunctions in the regions. CHRNA5 encodes the nAChR α5 subunit, which forms a fast permeable ion channel in pentamer combinations of α_2/3β_2/4α5 (Conroy & Berg 1995, Ramirez-Latorre et al. 1996, Gotti et al. 2006, Berrettini & Doyle 2012). The presence of α5 subunit increases acetylcholine sensitivity, calcium permeability, and fast desensitization (Ramirez-Latorre et al. 1996, Wang et al. 1996, Gerzanich et al. 1998). Its presence also increases behavioral accuracy on attention tasks under highly demanding conditions (Bailey et al. 2010). α5 is highly expressed in the frontal cortex and hippocampus, the critical regions for attention and learning (Wada et al. 1990, Salas et al. 2003, Winzer-Serhan & Leslie 2005). The reduction in nAChR α5 levels in these regions of KIV mice, particularly in the frontal cortex (a 46% reduction), suggests that cognitive enhancement by acetylcholine/nicotine via nAChR α5 is compromised by the deficiency in activity-dependent BDNF expression via promoter IV. This has a clinical implication because deficiency of promoter IV-BDNF expression can occur under chronic social and physical stress (Tsankova et al. 2006, Fuchikami et al. 2009), early-life maltreatment (Roth et al. 2009), and lack of neuronal activity (Tao et al. 1998, Shieh et al. 1998), and is observed in many psychiatric conditions associated with cognitive impairments, including schizophrenia (Weickert et al. 2003, Hashimoto et al. 2005, Wong et al. 2010), depression (Dwivedi et al. 2003, Keller et al. 2010, Hing et al. 2012), and Alzheimer's disease (Phillips et al. 1991, Murray et al. 1994, Ferrer et al. 1999, Holsinger et al. 2000, Garzon et al. 2002).

Interestingly, the CHRNA5 gene is well known for its polymorphisms associated with reduced cognitive performance (Winterer et al. 2010) and nicotine dependence (Saccone et al. 2007, Bierut et al. 2008, Thorgeirsson et al. 2008, Consortium 2010, Falvella et al. 2010, Ware et al. 2011, Hamidovic et al. 2011, Minica et al. 2017, Breetvelt et al. 2012). It is also associated with psychiatric conditions, such as schizophrenia (Hong et al. 2011), bipolar disorders (Jackson et al. 2013) and PTSD (Boscarino et al. 2011). One single nucleotide polymorphism (SNP) most studied is rs16969968 (Asp398Asn) (Saccone et al. 2007, Bierut et al. 2008, Thorgeirsson et al. 2008, Consortium 2010), which reduces functions of α5 nAChR (Bierut et al. 2008, Kuryatov et al. 2011) and co-occurs with another SNP that lowers CHRNA5 mRNA expression (Wang et al. 2009). Other polymorphisms in the CHRNA5 promoter regions are also associated with reduced mRNA expression of CHRNA5 (Smith et al. 2011, Wang et al. 2013). Reductions in expression and function of α5 nAChR are associated with reduced firing of GABA neurons and impaired social interactions in rodents (Koukouli et al. 2017). Our results suggest that decreased levels of α5 nAChR due to the BDNF deficiency may, in part, account for the defective GABAergic functions observed in the prefrontal cortex (Sakata et al. 2009) and abnormal cognitive and emotional functions observed in KIV mice; i.e., inflexible learning and depression/anxiety-like behavior (Sakata et al. 2013a, Sakata et al. 2010). Another plausible possibility is that the gene variations and expression regulation of α5 nAChR work together to alter the motivational and cognitive effects of nicotine/acetylcholine. In contrast to the well-studied polymorphism associations, the gene expression regulations between BDNF and CHRNA5 remain unstudied. To our knowledge, this study is the first to present causal evidence that BDNF deficiency reduces transcription levels of CHRNA5. This finding offers a novel insight into the gene-environment interplay, i.e. deficiency of promoter IV-BDNF expression, which can occur under environmental stress (Tsankova et al. 2006, Fuchikami et al. 2009, Roth et al. 2009) and lack of neuronal activity (Tao et al. 1998, Shieh et al. 1998), leads to α5 nAChR-mediated cholinergic dysfunctions. Negative environments may therefore worsen nicotine dependence and cognitive dysfunction, particularly in subjects with the high-risk gene variations of functionally-reduced CHRNA5/BDNF. Future studies can verify this gene expression–variation interplay.

Muscarinic acetylcholine receptors (mAChRs)

Additional novel findings included the small but significant increase in CHRM2 mRNA levels in the frontal cortex and the significant decrease in CHRMA5 mRNA in the hippocampus levels caused by the lack of promoter IV-driven BDNF expression. These findings suggest an additional disturbance in cholinergic functions via these muscarinic receptors in the subjects with promoter IV-BDNF deficiency.

CHRM2 encodes M2 mAChR, a G_i protein-coupled receptor that inhibits adenylate cyclase activity (Caulfield 1993, Felder 1995). The M2 receptor is an autoreceptor that inhibits further release of acetylcholine from the presynaptic neurons into the synaptic cleft (Chau et al. 2001, Zhang et al. 2002). Our result suggests that increased CHRM2 expression due to BDNF deficiency may limit acetylcholine release in the frontal cortex. Conversely, the increased CHRM2 expression could be a compensatory upregulation aimed at reducing a large amount of acetylcholine released under the BDNF deficiency. Further studies are needed to establish the acetylcholine levels and cholinergic functions in the frontal cortex. Interestingly, CHRM 2 has long been hypothesized to be involved in depression and cognitive inflexibility (Jeon et al. 2015), which are the behavioral phenotypes observed in KIV mice (Sakata et al. 2010). Human neuroimaging studies have shown a decrease (Cannon et al. 2006, Gibbons et al. 2009) or no change (Zavitsanou et al. 2005) in CHRM2 binding in the frontal cortex of subjects with mood disorders. Rodent studies have shown that blocking M2 mAChRs by an antagonist or deletion (knock-out) increases extracellular acetylcholine levels in the hippocampus and nucleus accumbens, and leads to depressive behavior (Chau et al. 2001) and impaired performance in the passive avoidance test (Tzavara et al. 2003, Chau et al. 2001). Generally, hyper-cholinergic activation is hypothesized to cause depression-like behavior (Janowsky et al. 1972, Jeon et al. 2015). Gene variations in CHRM2 have been also associated with a risk for major depression, major depressive syndrome and alcohol dependence (Luo et al. 2005, Wang et al. 2004, Hill et al. 2013), alcohol dependence in subjects with schizophrenia (Kendler et al. 2011), nicotine addiction (Mobascher et al. 2010), and efficacy of cognitive therapy (Bakker et al. 2014). Although the roles of M2 mAChRs in depression seem controversial (Cohen-Woods et al. 2009) and need further clarification, targeting the CHRM2 with antagonists or allosteric modulators may be efficacious as a treatment for mood disorders and possibly for cognitive flexibility.

CHRM5 encodes M5 mAChR, a G_q-protein-coupled receptor that activates the phospholipase C-phosphatidylinositol-calcium signaling pathway (Caulfield 1993). M5 mAChR mRNA is expressed in relatively selected regions/cells (Weiner et al. 1990), including the hippocampal CA1 pyramidal neurons (Vilaro et al. 1990) and the nigrostriatal dopaminergic neurons (Weiner et al. 1990, Reever et al. 1997), where M5 mAChR controls prolonged dopamine release (Yeomans et al. 2001, Forster et al. 2002). M5 mAChR is involved in reward-taking behavior; M5 knockout mice display reduced reward and withdrawal responses following morphine (Basile et al. 2002) and cocaine administration (Fink-Jensen et al. 2003). One possibility is that reduced expression of mAChR M5 in KIV mice may mediate their reduced responses for reward taking, i.e., depression-like behavior (Sakata et al. 2013a). Elucidation of how reduced gene expression of BDNF and CHRM5 affects the acetylcholine-induced dopaminergic function, and how this relates to affective and cognitive behavior, remains to be established in future studies.

Unchanged expression of other cholinergic genes

Our results showed no significant changes in expression of cholinergic genes other than CHRM2, CHRM5, and CHRNA5, either in the frontal cortex or the hippocampus, in the activity-dependent BDNF deficient mice. This lack of changes in expression of CHRNA7 and ChAT were somewhat surprising, when compared to previous reports.

CHRNA7 encodes α7 subunit, which forms a homomeric α7 nAChR (Gotti et al. 2006) with a relatively high permeability to calcium (Seguela et al. 1993, Bertrand et al. 1993). The α7 subunit is widely expressed in the brain, including in the hippocampus and cortex (Seguela et al. 1993). It is perhaps the most studied acetylcholine receptor related to BDNF. Previous studies have shown either an increase or a decrease in CHRNA7 expression, depending on the treatment conditions. Prolonged BDNF application (several hours to days) to cultured hippocampal neurons increases the clustering and the surface and intracellular pools of α7 nAChRs, particularly on GABAergic neurons (Kawai et al. 2002, Massey et al. 2006); however, acute BDNF application rapidly inhibits α7 nAChR-mediated currents in CA1 hippocampal interneurons (Fernandes et al. 2008). Decreased binding or expression of both CHRNA7 and BDNF have been also reported in patients with cognitive dysfunction, such as schizophrenia (Freedman et al. 1997, Freedman et al. 1995, Parikh et al. 2016, Weickert et al. 2003, Hashimoto et al. 2005, Wong et al. 2010), while both CHRNA7- and BDNF-mutant mice show cognitive inflexibility (Parikh et al. 2016, Sakata et al. 2013a). Therefore, we predicted that a positive correlation would exist between BDNF and α7 nAChRs levels. However, our results showed no change in CHRNA7 expression in KIV mice, indicating that the BDNF deficiency does not affect the basal levels of α7 nAChR expression.

Similarly, KIV mice showed no significant reduction in ChAT. This finding contrasted with findings of previous studies that showed positive corrections between BDNF and ChAT levels. BDNF has been shown to increase ChAT expression and the release of constitutive acetylcholine when applied to cultured septal or basal forebrain neurons (Alderson et al. 1990, Auld et al. 2001). Conversely, BDNF null (−/−) knockout mice have been shown to reduce ChAT staining levels at postnatal day 15 (Ward & Hagg 2000). The reason for this difference in findings is unclear, but may be attributed to the extent of the BDNF deletion: ChAT levels may be affected by constitutive BDNF levels in null knockouts (entire deletion), but not by activity-dependent expression in KIV mice that retain constitutive BDNF levels driven by the other eight BDNF promoters. Another possibility is that age differences (during development vs. adult) may alter the effect of BDNF deficiency on ChAT expression.

CHT is a high-affinity choline transporter that mediates the uptake of choline for acetylcholine synthesis in cholinergic neurons (Sarter & Parikh 2005). Activation of CHT functions is associated with attentional performance, whereas its dysregulation is associated with a decline in cholinergic transmission in Alzheimer's disease (Sarter & Parikh 2005). We did not observe any changes in expression of CHT due to promoter IV-BDNF deficiency.

The results of the effects of BDNF deficiency and treatments on cholinergic genes are presented in Figure 4.

Figure 4.

Hypothetical mechanisms to explain cholinergic gene expression changes by a reduction in activity-dependent BDNF expression and treatments. Black arrows show the expression changes in KIV mice compared to WT mice. Red arrows show the expression changes by treatments compared to saline or a standard environmental condition. Genotypes shown in parenthesis indicate a treatment effect only in that genotype. Fluoxetine (Flx); phenelzine (Phe); duloxetine (Dul); imipramine (Imi); enriched environment treatment (EET).

4.2. Treatment effects

Overall, our results suggested that the antidepressant treatments and EET were unlikely to affect the disturbed expression levels of CHRM2, CHRM5, and CHRNA5 caused by promoter IV-BDNF deficiency. This lack of effect suggests that these changes in cholinergic gene expression are involved in the previously observed improvement in antidepressive behavior due to the treatments (Sakata et al. 2013b). The exception was that EET significantly increased CHRNA5 mRNA levels in the frontal cortex only in KIV mice (Figure 2c), suggesting that EET reverses the nAChR α5 dysfunctions in that brain region. The imipramine results were somewhat unexpected; imipramine treatment did not affect the altered expression of cholinergic genes caused by promoter IV-BDNF deficiency, but it increased expression of CHRM2 in the hippocampus and of CHRM5 in the frontal cortex in both WT and KIV mice. These results indicated that the induction of these genes by imipramine was not specific to promoter IV-BDNF deficiency or to normalize the altered expression caused by the deficiency. Imipramine is actually an antagonist of muscarinic receptors, particularly M2 (Baldessarini 1996, Raisman et al. 1979). The expression changes of M2 and M5 by imipramine could be from the homeostatic changes in the cholinergic system after blocking/acting on the muscarinic receptors.

Interestingly, our results showed that EET and imipramine increased CHRNA7 expression in the frontal cortex, but only in KIV mice, whereas KIV mice showed no significant decreases in the basal expression levels of CHRNA7. The distribution of α7 mRNA expression strikingly overlaps with that of α5 in the hippocampal and cortical areas (Winzer-Serhan & Leslie 2005). The expression overlap raises the possibility that the induction of frontal cortex CHRNA7 by EET and imipramine may represent an additional compensation to normalize the dysfunctions caused by reduced CHRNA5 expression due to BDNF deficiency. α7 nAChR agonists or positive allosteric modulators has reported therapeutic potential as neuroprotective agents and cognitive enhancers (Levin et al. 2006), and for treatments of psychiatric and neurological disorders (e.g., schizophrenia, ADHD, depression, and Alzheimer's disease) (Taly et al. 2009). These α7 nAChR-modulating compounds may compensate for α5 nAChR dysfunctions due to promoter IV-BDNF deficiency in those subjects.

Many antidepressant drugs, including fluoxetine, are potent noncompetitive functional inhibitors of nAChR subtypes (Fryer & Lukas 1999), and they also increase the extracellular levels of serotonin, another potent noncompetitive antagonist of α7 nAChR (Palma et al. 1996). Therefore, we predicted that SSRI, SNRI, and TCA antidepressants would also inhibit α7 nAChRs and increase its adaptive mRNA induction. However, only the imipramine treatment results supported this hypothesis.

Our results also showed that EET increased CHT and ChAT expression in the frontal cortex of KIV mice, whereas the basal expression of these genes was not altered in KIV mice compared to WT mice. The increases in acetylcholine synthesis and transport by EET may represent a compensation for the reduced nicotinic cholinergic functions due to promoter IV-BDNF deficiency. Our observed lack of changes in ChAT expression by EET in WT mice agrees with a previous study that showed no changes in hippocampal ChAT activity in response to EET (Dhanushkodi et al. 2007), but disagrees with another study showing its increase (Wang et al. 2016). We also found the trend that imipramine reduced the expression of CHT in the WT frontal cortex, which needs to be further confirmed. The reason is unknown, but this reduction may reflect a homeostatic feedback operation due to the anti-cholinergic functions of imipramine.

Overall, EET effects were the most prominent of all the treatments, inducing CHRNA5, CHRNA7, CHT, and ChAT in the frontal cortex of KIV mice only, and thereby suggesting that EET may be particularly efficacious in subjects with BDNF deficiency (Figure 4). The specific and strong effects of EET are in accordance with the findings of our previous study — EET reversed depression-like behavior of KIV mice (Jha et al. 2011) more effectively than antidepressant treatments (Sakata et al. 2013b). One possibility is that enhancement of the cholinergic system underlies the behavioral effects of EET, particularly via increasing motivation and cognitive flexibility, even though increases in cholinergic activity is historically hypothesized to cause depression (Janowsky et al. 1972). However, the cholinergic hypothesis of depression may need revision, based on the detailed receptor-specific functional changes and interpretations. Patients with mood disorders can respond supersensitively to a cholinergic challenge, where the cholinergic supersensitivity is related to stress intolerance rather than the depressive state itself (Fritze 1993). Hypersensitivity may arise from an imbalanced expression of acetylcholine receptor subtypes. For example, increased expression of autoreceptor M2 AChRs in the frontal cortex of subjects with promoter IV-BDNF deficiency may cause a decrease in basal extracellular levels of acetylcholine. These reduced basal levels may in turn increase the cell-surface expression of AChRs as a way to increase sensitivity to acetylcholine. However, upon release of acetylcholine, the reduction in α5 nAChR expression may prohibit α5 nAChR-specific neuronal functions in subjects with BDNF deficiency. EET may reverse this dysfunction by upregulation of α5 nAChR as well as α7 nAChR, ChAT, and CHT.

The cholinergic system also interplays with other systems, namely the monoaminergic and GABAergic systems (Janowsky et al. 1983, Koukouli et al. 2017, Kawai et al. 2002, Massey et al. 2006). These systems are also involved in depression (Janowsky et al. 1983, Luscher et al. 2011) and are also disturbed by the BDNF deficiency in KIV mice (Sakata & Duke 2014, Tripp et al. 2012). A complex interplay likely occurs involving the altered cholinergic expression by the BDNF deficiency and the treatment effects. Further studies are needed to uncover the functional changes in the cholinergic system under the BDNF deficiency and their relationships to other neurotransmitter systems.

4.3. Cholinergic system as treatment target

The cholinergic system, particularly nAChRs, is gaining interest in the development of treatments for various nervous-system disorders, such as Alzheimer's disease, age-associated memory impairment, schizophrenia, depression, ADHD, and tobacco addiction (Arneric et al. 2007, Levin et al. 2006, Romanelli et al. 2007, Taly et al. 2009). Positive therapeutic effects of nicotinic agonists have been observed in initial studies on a variety of cognitive dysfunctions (Levin et al. 2006). Nevertheless, only compounds showing selectivity between α4β2 and α7 receptors have been obtained (Romanelli et al. 2007). To our knowledge, no compounds modulating α5 are yet available or have been tested to treat nervous system disorders. Our results showing a reduction in CHRNA5 in KIV mice suggest that the use of drugs targeting the function and expression of α5 nAChR (e.g., modulators increasing Ca²⁺ permeability and blockers of its desensitization and internalization) could be a beneficial new strategy for treating cognitive dysfunctions and mood disorders caused by BDNF deficiency. These α5 nAChR compounds may become add-on drugs to supplement the already existing therapeutic nicotinic agonists targeting α4β2 and α7 (Levin et al. 2006). In addition, biological tools to study CHRNA5 protein levels/function, such as α5-specific agonists/antagonists and α5-specific antibodies, are also currently unavailable. This is likely the reason why the protein levels and functions of α5 remain unstudied in comparison to other nAChR subunits, such as α7 and α4β2 (Parikh et al. 2016). Developing these tools may greatly advance our understanding of the mechanisms by which BDNF deficiency leads to cholinergic dysfunctions in the frontal cortex and hippocampus.

The potent antimuscarinic effects of tricyclic antidepressants, including imipramine have been thought primarily to produce adverse side effects during antidepressant drug treatment (Pacher & Kecskemeti 2004, Baldessarini 1996). However, a more recent study has shown that blocking M1 and M2 muscarinic receptor subtypes produces antidepressant-like effects of the rapidly acting antidepressant scopolamine (Witkin et al. 2014). Our results suggest that novel drugs that selectively block M2, but activate M5 mAChRs, and that have negligible effects on the peripheral nervous system, might produce more rapid and robust clinical improvement in patients with mood disorders.

4. 4. Study limitations

The relatively limited effects of the antidepressant treatments may be due to the drug administering method, i.p. injection. We used this method based on the previous studies (see Methods). However, daily i.p. injection could be stressful to the animal, and the stress could counteract the treatment effects. Future studies using other (e.g., oral) methods may confirm the drug effects on the cholinergic system, although either i.p. injection and oral administration of the four drugs has been shown not to rescue the reduced BDNF levels in the KIV hippocampus despite our initial hypothesis (Sakata et al. 2013b).

Another caveat of this study is that we only measured mRNA levels. The levels of mRNA do not necessarily reflect protein levels, but could reflect a compensational upregulation of the dysfunction of the cholinergic molecules. For example, increased M2 mAChRs in KIV mice may reflect reduced protein levels of the M2 autoreceptor and the excess levels of released acetylcholine under BDNF deficiency, which compensate for the dysfunctions in nicotinic acetylcholine receptor α5. Future studies are expected to clarify the levels of acetylcholine and these cholinergic protein levels/locations in the frontal cortex and hippocampus, as well as the functional changes, through the use of specific agonists/antagonists. This clarification will further advance our understanding of the effects of BDNF deficiency and treatments on the cholinergic system.

4. 5. Conclusions

Our goal in this study was to determine the effects of promoter IV-BDNF deficiency and treatments on cholinergic gene expression. Our results provided causal evidence that promoter IV-driven BDNF reduced CHRNA5 gene expression in both frontal cortex and hippocampus, increased CHRM2 in the frontal cortex, and decreased CHRM5 in the hippocampus. This disturbance of cholinergic gene expression by BDNF deficiency may be one important mechanism that links stress and lack of activity to the defective behavioral phenotypes observed in KIV mice; namely, inflexible learning and depression-like behavior caused by impaired attention and response inhibition. Our findings also indicate that chronic treatments with fluoxetine, phenelzine, duloxetine, imipramine, and EET did not affect the disturbed expression of these genes, except that EET increased CHRNA5 mRNA levels only in the frontal cortex of KIV mice. EET also upregulated mRNA levels of CHRNA7, CHT, and ChAT in the frontal cortex of KIV mice. These results suggested a potential cholinergic recovery in the frontal cortex in the BDNF-deficient condition by EET. The imipramine effect was most prominent among the four different types of antidepressants; it upregulated hippocampal CHRM2 and frontal cortex CHRM5 in both genotypes, and frontal cortex CHRNA7 only in KIV mice, likely reflecting its anticholinergic function. Future studies are needed to address the detailed localization and the consequent dysfunction of the disturbed cholinergic expression caused by the deficiency of activity-dependent BDNF and its relationship to cognitive functions and affective disorders. It would be also interesting to examine gene variation-expression relationships of CHRNA5, CHRM2, and CHRM5 in patients who show reduced BDNF levels (e.g., schizophrenia, depression, Alzheimer's disease), correlating with cognitive function.

Abbreviations

BDNF

brain-derived neurotrophic factor

WT

wild-type

promoter IV-BDNF

promoter IV-driven BDNF expression

CHT

choline transporter

KIV

knockin-promoter IV

ChAT

choline acetyltransferase

CHRNA5

nicotinic acetylcholine receptor alpha 5

CHRM2

M2 muscarinic acetylcholine receptor

CHRM5

M5 muscarinic acetylcholine receptor

nAChR

nicotinic acetylcholine receptor

mAChR

muscarinic acetylcholine receptor

EET

enriched environment treatment

SCT

standard condition treatment

GABA

gamma aminobutyric acid

ADHD

attntion deficit hyper activity

PTSD

posttraumatic stress disorder

GFP

green fluorescent protein

ES

embryonic stem

NH

National Institute of Health

qRT-PCR

quantitative reverse transcription polymerase chain reaction

HGPRT

hypoxanthine-guanine phosphoribosyltransferase

TBP

TATA-box binding protein

Ct

cycle threshold

ANOVA

analyses o variance

SE

standard error of mean

N

number

FDR

false discovery rate

SSRI

selective serotonin transporter reuptake inhibitor

SNRI

serotonin-norepinephrine reuptake inhibitor

TCA

tricyclic antidepressant