Relationship between constitutive and acute gene regulation, and physiological and behavioral responses, mediated by the neuropeptide PACAP

Abstract

Since the advent of gene knockout technology in 1987, insight into the role(s) of neuropeptides in centrally- and peripherally-mediated physiological regulation has been gleaned by examining altered physiological functioning in mammals, predominantly mice, after genetic editing to produce animals deficient in neuropeptides or their cognate G-protein coupled receptors (GPCRs). These results have complemented experiments involving infusion of neuropeptide agonists or antagonists systemically or into specific brain regions. Effects of gene loss are often interpreted as indicating that the peptide and its receptor(s) are required for the physiological or behavioral responses elicited in wild-type mice at the time of experimental examination. These interpretations presume that peptide/peptide receptor gene deletion affects only the expression of the peptide/receptor itself, and therefore impacts physiological events only at the time at which the experiment is conducted. A way to support ‘real-time’ interpretations of neuropeptide gene knock-out is to demonstrate that the wild-type transcriptome, except for the deliberately deleted gene(s), in tissues of interest, is preserved in the knock-out mouse. Here, we show that there is a cohort of genes (constitutively PACAP-Regulated Genes, or cPRGs) whose basal expression is affected by constitutive knock-out of the Adcyap1 gene in C57Bl6/N mice, and additional genes whose expression in response to physiological challenge, in adults, is altered or impaired in the absence of PACAP expression (acutely PACAP-Regulated Genes, or aPRGs). Distinguishing constitutive and acute transcriptomic effects of neuropeptide deficiency on physiological function and behaviorin mice reveals alternative mechanisms of action, and changing functions of neuropeptides, throughout the lifespan.

1. Introduction

Neuropeptides were discovered, characterized chemically, and their receptors profiled pharmacologically over the course of the decades from 1910–1970 (Hökfelt, 1991). cDNAs for transcripts encoding the neuropeptides and their G-protein coupled receptors were obtained during the 1980s and 90s. The last known neuropeptide receptors to be cloned were the opiate receptors MOR, DOR and KOR, with other peptide-liganded GPCRs subsequently de-orphanized (Hökfelt et al., 2000; Reinscheid et al., 2005). The technology for gene knockout was developed by the Capecchi lab in 1987 (Thomas and Capecchi, 1987), and originally applied to investigation of homeobox genes and their influence on mammalian development, including that of the central nervous system. Knock-out technology, applied to individual peptide and peptide receptor genes throughout the 1990’s and early 00’s (e.g., (Sherwood et al., 2007)) allowed a more definitive causality to be applied to the roles and functions of neuropeptides/neuropeptide receptors than was possible pharmacologically with chemical agonists and antagonists.

However, as Hokfelt and colleagues point out, the era of molecular biology did not lead automatically to the clear identification of physiological roles for individual peptides (Hokfelt et al., 2018). This may be partly because, as peptides are released under conditions of high neuronal activity (e.g., stress), their main contribution to neuroendocrine regulation is to allow ‘efficacy in extremis’, rather than primary action in basal physiology (Hokfelt et al., 2003). A second reason is that neuropeptides may act not only immediately and directly, following peptide secretion/release via synaptic and paracrine interactions with their receptors (Brown et al., 2020), but also indirectly, by influencing the chemoanatomy of brain, peripheral nervous system, and endocrine tissues during development. Interpreting the results of loss-of-function experiments following genetic editing of neuropeptide and neuropeptide receptor genes may therefore require accounting for both immediate (direct) effects of secreted neuropeptides and downstream activation of their receptors, and for indirect effects mediated through altered expression of mRNAs affecting the expression and function of other proteins. Thus, loss of function in neuropeptide-ablated mice could be attributed either to the absence of the peptide itself, at time of experiment, or to developmental or other indirect effects of neuropeptide loss or absence. Useful interpretation of knock-out experiments in adult mice, especially in a translational context, clearly requires assessment of the time of impact of neuropeptide deficiency. Distinguishing effects of neuropeptide deficiency on the basal wild-type transcriptome, which might indirectly affect physiological and behavioral response patterns, from deficiencies in real-time gene regulation required for execution of physiological and behavioral responses, is an important step in that assessment.

An early example of perturbation of brain function secondary to the immediate loss of peptide expression and release was provided by the Young laboratory studying the effects of knock-out of oxytocin (Oxt) and vasopressin (AVP) receptor genes in the mouse (Pagani et al., 2011). Constitutive knock-out of either the AVP type 1 or Oxt receptors did not affect freezing behavior following fear conditioning. However, forebrain-specific knock-out of the oxytocin receptor after weaning resulted in reduction in Avpr1 expression in the central nucleus of the amygdala, and subsequent impairment in fear conditioning. The authors interpreted their results as indicating a compensation for loss of Oxt or AVP receptors preserving patent fear conditioning responses if occurring early enough in development, but a subsequent lack of compensation in young, post-weaning mice, during a period in which Oxt receptor is needed for maintenance of AVP receptor gene transcription presumably required for patent function of circuits involved in fear learning. Given the intimate association between Oxt and AVP function in learning and memory circuits, and in the development of such circuitry in mammals, it is perhaps unsurprising that abolishing expression of Oxt, AVP, or the receptors for either, might have unequally penetrant effects at different onsets of peptide/receptor deficiency in the mouse.

Similar concerns about compensatory effects of PACAP and VIP mediated through the actions of their shared receptors PAC1, VPAC1 and VPAC2 prompted May and colleagues to examine the expression of PACAP, VIP, PAC1, VPAC1 and VPAC2 mRNA as a function of constitutive (from inception) knock-out of either PACAP or VIP (Girard et al., 2006). In this case, deficiency in expression of either PACAP or VIP did not affect the expression of any of the four remaining components of this overlapping and potentially self-compensating system in the brain. However, these experiments beg the question of whether neuropeptide constitutive knock-out has long-term indirect effects on the transcriptome, or immediate effects on either cellular excitability or cell plasticity. To investigate the mechanism(s) of PACAP action in the mouse we examined, in wild- type (WT) and PACAP knockout (PACAP KO) mice, the transcriptomes of several brain areas in which PACAP is released from nerve terminals, and the adrenal gland, at which PACAP is the sole identified neuropeptide neurotransmitter. Our results highlight the importance of considering the dual functional roles of central nervous system neuropeptides in nervous system development and in real time neurotransmission in adult mice.

2. Materials and Methods

2.1. Transgenic and knockout mice

PACAP- and PAC1-deficient C57BL6/N mice were generated as previously described (Hamelink et al., 2002; Mustafa et al., 2014; Jiang et al., 2016). Wild type (WT) refers to corresponding C57BL6/N mice.

The CRISPR-Cas9 system with a donor plasmid was used to generate the floxed PACAP (Adcyap1) allele on a C57BL/6N background. CRISPR-Cas9 mediated the specific double-strand DNA cleavage on both 5’ and 3’ of the coding sequence (CDS) of the PACAP gene where two LoxP sites are located in the edited gene within ‘floxed’ mice. The donor plasmid contained a LoxP site at both a 5’ and 3’ site within the target region (CDS of Adcyap1) and the LoxP sites were flanked by Adcyap1 sequences mediating homologous recombination of the LoxP-CDS (Adcyap1)-LoxP region into the Adcyap1 gene of the transgenic mouse. The CDS of PACAP (P38) is located in exon 5 of the Adcyap1 gene, which floxes the Adcyap1 gene to permit PACAP excision as in the constitutive knockout (Hamelink et al., 2002).

sgRNAs were designed using MIT Guide RNA design software (Feng Zhang laboratory, previously available at crispr.mit.edu) and CCTop (https://cctop.cos.uni-heidelberg.de), a CRISPR/Cas9 target online predictor was used to minimize off-target recombination and optimize specificity for the flanking sequences of the CDS of PACAP. The 5’sgRNA was designed to avoid the predicted splicing signal and corresponds to the sequence segment (5’-agtcacagtattcccgccagCGG-3’) located in the intron upstream of exon 5. The 3’sgRNA was designed to target the sequences (5’-CCCactggttgcaggggcaattc-3’) in exon 5 of the Adcyap1 gene to facilitate the insertion of LoxP 567 base pairs after the STOP codon following Cas9 cleavage of the double-stranded genomic DNA. Capital letters in the targeting sequences indicate the PAM sites. A donor plasmid was generated on a TA vector backbone and contains an ~1.3 kb sequence homologous to the Adcyap1 gene upstream of the 5’LoxP and a 1.5 kb sequence homologous to the Adcyap1 gene downstream of the 3’LoxP site.

sgRNAs were purchased from IDT as Alt-R crRNA and tracrRNA oligos, which were mixed in dH2O at a ratio of 1:2 and annealed to generate sgRNA in a thermocycler (95°C for 5 min and then ramp down to 25°C at a rate of 5°C /min). gRNA and Cas9 protein (IDT) were mixed at room temperature for 15 minutes to form a ribonucleoprotein complex with subsequent addition of donor plasmid DNA to generate the injection mix. Each component had a final concentration of 20 ng/μl in 1xTAE buffer. The injection mix was then passed through a Millipore filter (UFC30VV25). The final injection solution was injected into zygotes to produce founder mice in the NIMH-IRP Transgenic Core Facility. Accurate CRISPR-Cas9 cleavage of the target site and insertion of the LoxP sites was confirmed by PCR amplification, and sequencing, of the region.

CaMK2α-CRE PACAPfl/fl mice are a cross between mice expressing CRE under the CaMK2α promoter (Tg(Camk2a-cre)T29–1Stl) and mice with LoxP sites flanking exon 4 of the Adcyap1 gene (Hamelink et al., 2002). These animals were bred as described in Jiang, 2017. VGaT-CRE PACAPfl/fl mice are a cross between PACAP floxed mice described above and a commercially available VGaT-ires-CRE line (B6J.129S6(FVB)-Slc32a1tm2(cre)Lowl/MwarJ, Jackson Labs).

Mice used for experiments described here were between eight and 32 weeks of age. Both male and female mice were used. All procedures were conducted in accordance with the National Institutes of Health Guide of the care and use of laboratory animals, 8^th edition (2010).

2.2. Dissection of brain and peripheral tissue for qRT-PCR and microarray experiments

At least three male and three female mice of each genotype, aged 10–20 weeks, were used for microarray analysis and qRT-PCR assay. Because of the well-documented role of PACAP in the hypothalamic-pituitary-adrenal axis (Stroth, 2011; Mustafa, 2013), the hypothalamic and adrenal transcriptomes were assessed. To investigate whether the transcriptomic effects of PACAP knockout extended to other brain regions, we examined the hippocampus, where PACAP regulates synaptic transmission (see Yang, 2010 for review), and the cerebellum, where PACAP plays neurotrophic and neuroprotective roles (Vaudry, 2000; Vaudry, 2002; Botia, 2007). Mice were sacrificed by cervical dislocation and whole brains were quickly removed from the skull and rinsed in cold 1× PBS buffer. Two lateral cuts were made, at the optic chiasm and above the cerebellum and brainstem respectively, to obtain a coronal tissue block. Hypothalamus was microdissected along the anterior commissure from one hemisphere to the other using a blade and fine-tipped forceps. Whole hippocampus was removed after peeling off the cerebral cortex from the same coronal tissue block. The prefrontal cortex was dissected according to The Mouse Brain in Stereotaxic Coordinates, Paxinos and Franklin, 2001 (Bregma 1.5 to 0.5 mm) and cerebellar cortex was also collected from each hemisphere. Both adrenal glands were collected from animals. Dissected tissue was either quickly frozen in Dry Ice and stored at −80 C until RNA extraction or stored briefly on ice in QIAzol Lysis Reagent (Qiagen, Germantown, MD) until RNA extraction shortly following dissection.

2.3. RNA extraction for qRT-PCR and microarray

Total RNA was extracted using an miRNeasy Mini kit (QIAGEN, Cat#: 217004, Valencia, CA, USA), according to the manufacturer’s instructions. Briefly, frozen tissue (20 mg or less) from brain or adrenal glands was disrupted using a plastic pestle in 700 μl lysis buffer per sample, then centrifuged at 12,000 × g for 15 min at 4° C. Following centrifugation, the supernatant was transferred to a new microcentrifuge tube, and 1.5 volumes of 100% ethanol was added and mixed by pipetting. The mixed sample was transferred to an RNeasy mini-spin column and centrifuged for 30 s at 8,000 × g. The RNA bound to the column membrane was washed out with RWT buffer then with RPE buffer. The RNA was then dissolved in 30–40 μl of diethylpyrocarbonate (DEPC)- treated water. RNA concentration was evaluated by absorbance at 260 nm using the NanoVue Plus spectrophotometer (Biochrom US, Holliston, MA, USA). RNA purity and integrity were further determined by Bioanalyzer, and only RNA with RNA Integrity Number (RIN) greater than 8 was used in microarray or qRT-PCR experiments. RNA was stored at −80 °C until use.

2.4. cDNA synthesis and qRT-PCR

Single-stranded cDNAs were synthesized with the Superscript III first strand cDNA synthesis kit (Cat#: 18080051, Thermo Fisher Scientific, Waltham, MA, USA), according to the kit protocol, using 1–2 mg total RNA. qRT-PCR protocols were as reported previously (Liu, 2009; Zhang, 2014). All Taqman gene expression assay kits were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Taqman probes were designed using Partek Genomic Suite software, also used for the microarray analysis, to align probe sequences with microarray results. Detailed information for each probe is listed in Table 1.

Table 1:

Taqman gene expression assays for qRT-PCR

Gene symbol	Assay ID
Adcyap1	Mm00437433_m1
Pttg1	Mm00479224_m1
Xaf1	Mm01245815_m1
Mid1	Mm04933165_m1
Pla2g4e	Mm00625711_m1
Gapdh	Mm99999915_g1

The qRT-PCR reactions were performed in a BioRad iCycler as follows: 95o C hold for 20 s followed by 40 cycles of 95o C, denaturation for 3 s, and 60o C annealing and extension for 30 s. qRT-PCR analyses for each mRNA transcript were performed using the 2−ΔΔCt method (Livak, 2001). In the present study, data are presented as the fold-change in expression of each transcript, normalized first to the internal Gapdh transcript and then relative to the reference tissue within each tissue type. The cycle threshold (Ct) was defined as the number of cycles required for the fluorescent signal to cross the threshold. ΔCt was determined as [mean of the duplicate Ct values for each gene] – [mean of the duplicate Ct values for Gapdh]. ΔΔCt represented the difference between the paired tissue samples (e.g., PACAP KO vs. WT), as calculated by the formula ΔΔCt = [ΔCt for each transcript in PKO tissue - ΔCt for each transcript in reference WT tissue]. The fold change for each transcript in specific tissue was expressed as 2−ΔΔCt. Ct values for qRT-PCR experiments are available in Tables 3, 4, and 5.

Table 3:

qRT-PCR analysis of cPRGs, Ct values

Gene and Tissue	Genotype
Adcyap1	Control	PACAP KO
Hypothalamus	22.25 (0.09)	28.38 (0.15)
Cerebellum	25.46 (0.12)	29.15 (0.18)
Hippocampus	26.98 (0.17)	33.79 (0.30)
Adrenal Gland	31.97 (0.71)*	32.81 (0.63)*
Pttg1	Control	PACAP KO
Hypothalamus	22.69 (0.12)	28.15 (0.10)
Cerebellum	23.17 (0.09)	29.13 (0.10)
Hippocampus	23.36 (0.11)	29.16 (0.12)
Adrenal Gland	21.90 (0.21)	26.12 (0.29)
Mid1	Control	PACAP KO
Hypothalamus	22.09 (0.24)	25.49 (0.38)
Cerebellum	24.16 (0.13)*	26.80 (0.39)*
Hippocampus	24.85 (0.18)*	22.99 (0.43)*
Xaf1	Control	PACAP KO
Hypothalamus	25.70 (0.14)	24.11 (0.19)
Cerebellum	28.83 (0.15)*	26.81 (0.11)*
Hippocampus	28.40 (0.27)*	26.82 (0.13)*
Pla2g4e	Control	PACAP KO
Hypothalamus	25.69 (0.09)	24.03 (0.30)

Data are presented as mean (standard error of mean)

^*indicates data not presented in graphs

Table 4:

qRT-PCR analysis of Adcyap1 and Pttg1 in conditional knockout mice

Gene and Tissue		Genotype
Adcyap1	PACAP fl/fl	CaMK2α-CRE PACAPfl/fl
Prefrontal Cortex	23.88 (0.22)	26.17 (0.17)
Cerebellum	26.22 (0.16)	26.59 (0.15)
Pttg1
Prefrontal Cortex	22.93 (0.17)	23.09 (0.07)
Cerebellum	23.00 (0.07)	23.00 (0.08)
Adcyap1	PACAP fl/fl	VGaT-CRE PACAPfl/fl
Prefrontal Cortex	25.88 (0.14)	26.44 (0.10)
Cerebellum	26.61 (0.12)	28.04 (0.13)
Pttg1
Prefrontal Cortex	25.21 (0.07)	25.38 (0.11)
Cerebellum	25.06 (0.04)	25.57 (0.10)

Data are presented as mean (standard error of mean)

Table 5:

qRT-PCR analysis of Stc1 and Ier3

Gene and Tissue	Time Restrained
Stc1	0 hours	1 hour	6 hours
Hypothalamus	25.62 (0.10)	25.61 (0.12)	25.75 (0.07)
Adrenal Gland	28.06 (0.13)	27.44 (0.14)	26.62 (0.13)
Ier3	0 hours	1 hour	6 hours
Hypothalamus	25.34 (0.17)	25.24 (0.23)	25.14 (0.13)
Adrenal Gland	27.93 (0.26)	23.70 (0.27)	28.16 (0.15)

Data are presented as mean (standard error of mean)

2.5. Microarray analysis

Analysis was conducted in the NHGRI-NINDS-NIMH Microarray Core (Abdel Elkahloun, Director) using its established protocols (available at https://ma.nhgri.nih.gov/) and Clariom-S mouse microarray chips (Affymetrix).

2.6. Bioinformatics

Microarray data was analyzed using Partek Genomics Suite. CEL files from five groups of microarrays (two from hypothalamus, and one each from adrenal gland, cerebellum, and hippocampus) were imported separately for each experiment. Robust multi-array average (RMA) normalization was used to correct for inter-array variability and normalize the distribution of the data, allowing for parametric statistical analyses to be performed (Irizarry, 2003), resulting in expression values in the form of log2 values. Within each experiment, differentially expressed genes were determined by ANOVA, with pairwise contrasts between groups of interest. To compare hypothalamic transcriptomes of constitutive PACAP KO and WT mice, we designated genes as differentially expressed with a fold-change of plus-or-minus 1.5 and a raw p-value of p <= 0.01. The relatively relaxed raw p-value was chosen rather than an adjusted p-value, acting as a wide filter for potentially relevant genes. Further refinement was achieved by identifying the overlapping set of genes commonly regulated in four independent microarray experiments (two in hypothalamus, one in cerebellum, and one in hippocampus). Finally, a subset of genes within the identified gene sets were validated with qRT-PCR, including appropriate controls for multiple comparisons. Other thresholds used are identified and described in text and/or figure legends. Where raw expression values are shown, data are given as Log2 expression value, normalized in Partek via RMA normalization of CEL files from all microarray analysis experiments.

2.7. RNAscope in-situ hybridization

In-situ hybridization was conducted using the RNAScope Multiplex Fluorescent V2 Assay (ACD Bio), as directed in the manual. Briefly, fresh-frozen mouse brains (one PACAP KO and one PACAP^fl/fl control, both female) were sectioned sagittally on a cryostat to 16μM. Sections were mounted on Superfrost Plus slides (Fisher Scientific) and dried on slides at −20 degrees. Slides were fixed with 4% paraformaldehyde for 15 minutes at 4 degrees, then dehydrated in an ethanol gradient. Hydrogen peroxide and protease treatments, probe hybridization and signal development proceeded as described in the manual. Probes were Mm-Pttg1-C3 (Cat No. 560291-C3), and Mm-Xaf1(Cat No. 504181). These were developed with Opal 620 and Opal 690 fluorophores (Akoya Biosciences). Imaging was performed in the NIMH Systems Neuroscience Imaging Resource (SNIR) on a Stellaris confocal microscope.

2.8. Restraint stress, corticosterone, and hypophagia

Restraint stress and measurement of hypophagia-related weight loss and corticosterone levels were carried out as described in Jiang, 2016. Briefly, mice were transported to the testing room in home cages and restrained in a Decapicone with added slits for ventilation, which was gently twisted closed over the base of the tail and secured with tape. Mice were kept in restraint or in holding cages unrestrained for 1 hour in Hypothalamus Experiment 1 and for 3 hours in Hypothalamus Experiment 2 and in the Adrenal Experiment. Mice were checked every 15 minutes for well-being or escape. To detect corticosterone, blood was taken from tail vein, and plasma collected by centrifugation. Corticosterone levels were measured with DetectX corticosterone ELISA kits (Arbor Assays, Ann Arbor, MI). Stress-induced weight loss was calculated as the weight of the mouse 24-hours after the end of restraint divided by the initial weight of the mouse, times 100. For microarray, tissue was dissected as described above from restrained mice and non-stressed controls.

2.9. Repetitive jumping

Repetitive jumping was assessed using the CleverSys Homecage Scan system (CleverSys Inc., Reston, VA), as in (Mustafa, 2015; Lehmann, 2013). Individual mice were transferred to cages containing some fresh bedding and some bedding from the home cage and placed in the monitoring boxes. Side capture video recording was carried out for approximately 24 hours and analyzed by the program. Jumping behavior was assessed statistically during the dark portion of the cycle alone, as very little jumping occurs during the day (see Figure 7E).

Figure 7. Physiological and Behavioral phenotypes: Hypophagia, Corticosterone, Repetitive jumping.

A) Blunting of stress-induced hypophagia occurs in both male and female PACAP KO mice. Three-way ANOVA found no main effect of sex and no interactions between sex and either stress or genotype, so sexes were combined and analyzed with 2-way ANOVA. There were main effects of Stress (F(2,94) = 53.76, p < 0.0001), of Genotype (F(2,94) = 7.21, p = 0.0012) and a significant Stress by Genotype interaction (F(2,94) = 14.6, p < 0.0001). Šídák’s multiple comparisons test showed that while (N = 18) WT mice lost significant weight following stress (p < 0.0001), neither PACAP nor PAC1 KO mice did. Therefore, stressed WT mice lost significantly more weight than stressed PACAP KO or PAC1 KO mice (p < 0.0001 for both). N = 18 stressed and 12 non-stressed WT, 19 stressed and 10 non-stressed PACAP KO, and 20 stressed and 22 non-stressed PAC1 KO mice. B) Corticosterone increases following 3 hours of restraint stress in PACAP KO, PAC1 KO, and WT mice of both sexes (3-way ANOVA, Main effect of Stress F (1, 88) = 397.2, p < 0.0001). Tukey’s multiple comparisons test showed significant increases after stress in all groups (p < 0.0001, stars). However, the increase is blunted in both male and female PACAP KO mice (hashes; Male WT Stressed (N = 14) vs Male PACAP KO Stressed (N = 10), p < 0.0001; Male WT Stressed vs Male PAC1 KO stressed (N = 15), p < 0.0001; Female WT Stressed (N = 4) vs Female PACAP KO Stressed (N = 7), p< 0.0001; Female WT Stressed vs Female PAC1 KO stressed (N = 5), p = 0.012). C) Lack of jumping is correlated with non-depletion of the cPRG Pttg1. In prefrontal cortex of (N = 6) CaMK2α-CRE PACAP^fl/fl mice, compared with N = 6 PACAP^fl/fl control mice, Adcyap1 was significantly reduced (2-way ANOVA, Šídák’s multiple comparisons test p = 0.0003), but Pttg1 was not. In cerebellum, neither gene was altered in CaMK2α-CRE PACAP^fl/fl mice relative to controls. D) In (N = 5) VGaT-CRE PACAP^fl/fl mice, Adcyap1 was significantly reduced in both PFC (2-way ANOVA, Šídák’s multiple comparisons test p = 0.015) and cerebellum relative to control PACAP^fl/fl mice (N = 6) (2-way ANOVA, Šídák’s multiple comparisons test p <0.0001),but Pttg1 was not altered relative to controls in either tissue. E, E’) While (N=12) constitutive PACAP knockout mice perform a repetitive jumping behavior at high rates (yellow line), none of the other tested animals (N = 7 PACAP^fl/fl controls, N = 4 PAC1 KO, N = 5 CaMK2α-CRE PACAP^fl/fl, and N = 4 VGaT-CRE PACAP^fl/fl ) jump. Repetitive jumping over the course of 24 hours is shown in 7E, while average number of repetitive jumping bouts per half-hour during the dark cycle is shown in E’. Brown-Forsythe ANOVA test revealed a significanteffect of genotype (F(4,11) = 12.96), with Dunnett’s T3 multiple comparisons test showing significantly higher jumping in PACAP KO than all other groups (p = 0.036 for all comparisons).

2.10. Statistical analysis

Statistics for microarray data are described above, in Bioinformatics (section 2.6). Additional statistical analyses are described in text and/or figure legends. Statistics were performed in GraphPad Prism.

3. Results

3.1. Transcriptome of PACAP-deficient compared to wild-type mice

Because of the well-documented role of PACAP in the hypothalamic-pituitary-adrenal axis (Stroth, 2011; Mustafa, 2013), the hypothalamic transcriptome was assessed. To investigate whether the transcriptomic effects of PACAP knockout extended to other brain regions, we examined the hippocampus, where PACAP regulates synaptic transmission (see Yang, 2010 for review), and the cerebellum, where PACAP plays neurotrophic and neuroprotective roles (Vaudry, 2000; Vaudry, 2002; Botia, 2007).

First, differences in the hypothalamic transcriptomes of WT and PACAP-deficient mice were assessed. In the first experiment, designated “Hypothalamus Experiment 1”, RNA was harvested from hypothalamus of four non-stressed PACAP KO male mice and three non-stressed WT male mice. Nineteen genes differed between PACAP KO and WT males at a fold-change of +/− 2, p <= 0.01. This includes 4 upregulated and 16 downregulated genes, including Adcyap1. Given the sexually dimorphic nature of the hypothalamus, we repeated this experiment with an equal number of non-stressed male and female mice (n = 3 male and n = 3 female animals of each genotype, “Hypothalamus Experiment 2”). We identified transcripts whose expression was altered regardless of sex by performing sex by genotype ANOVA and excluding from consideration genes with significant interactions at p < 0.01. In Hypothalamus Experiment 2, 16 transcripts were altered (5 upregulated, 11 downregulated) in PACAP KO versus WT animals.

As shown in Figure 1, a subset of 13 transcripts, including Adcyap1 itself, was altered in PACAP KO versus WT mice in both hypothalamic experiments. We will refer to these as cPRGs (constitutively PACAP-Regulated Genes). Ten transcripts (Pttg1, Mid1, 4833420G17RIK, Pisd- ps3, Cd59a, Gm5148, Cyb5r3, Tmem260, and Cetn4, as well as Adcyap1 itself) showed significantly lower expression in PACAP-deficient mice and three transcripts (Pla2g4e, Mrpl20, and Xaf1) showed significantly higher expression in constitutively PACAP-deficient mice. The fact that these genes are regulated in both experiments, despite differences in experimental protocols (see Methods) suggests that these genes differ due to genotype rather than other experimental factors. Although none of the identified cPRGs are regulated differently in mice of different sexes, we did find several sex-dependently expressed transcripts in hypothalamus, in good agreement with the literature (Mozhui et al., 2012) (Figure 1 Supplemental).

Figure 1. Transcripts differentially expressed in hypothalamus of constitutively PACAP- deficient mice.

Microarray analysis reveals genes which are differentially expressed (p < 0.01, fold-change +/− 2 or greater) between non-stressed PACAP KO and non-stressed WT mice in both hypothalamic microarrays. Hypothalamus experiment 1: N = 3 male wildtype and N= 3 male PACAP KO. Hypothalamus experiment 2, N = 3 males and N = 3 females of each genotype. Bar graph shows the mean fold-change (+/− standard error of the mean) of PACAP KO hypothalamus versus WT hypothalamus across both experiments. Color bar indicates p-value of comparison. Inset Venn depicts overlap between the two experiments.

Identification of constitutively PACAP-regulated transcripts was then extended to the cerebellum and hippocampus, which express modest amounts of Adcyap1 in excitatory (hippocampus) or inhibitory (cerebellum) neurons (Zhang et al., 2020). Because of these lower baseline levels of Adcyap1, the threshold for fold-change was shifted to +/− 1.1 in these tissues (Figure 2A). The loss of power resulting from lowered threshold was compensated for by identifying genes which replicate, that is, which are regulated in multiple experiments. In hippocampus, 63 genes were differentially regulated in PACAP KO versus WT animals. In cerebellum, 58 genes met the criteria.

Figure 2. Constitutively PACAP-regulated genes across three brain regions.

A) Fold-change in Adcyap1 expression in PACAP-deficient mice in various tissues. Bar graph shows fold-change of PACAP KO mice versus WT mice within each experiment. Color bar indicates p-value of comparison. Fold-change in Adcyap1 expression varies across tissues/brain regions despite identical deletion of the Adcyap1 gene at all sites due to differences in the intrinsic expression of the Adcyap1 transcript and the threshold for transcript detection by microarray analysis (see text). B) Constitutively PACAP-regulated genes (cPRGs) identified by microarray analysis in hippocampus and cerebellum (N = 3 males and N = 3 females of each genotype in each tissue) and hypothalamus, as described previously. Genes were identified as cPRGs if they were differentially expressed between PACAP KO and WT mice (p < 0.01, fold-change +/− 1.1 or greater) in PACAP knockout versus WT cerebellum and hippocampus and differentially expressed p < 0.01, fold-change +/− 2 or greater in both hypothalamic experiments. Mean +/− standard error of the mean, color bar indicates p-value of comparison.

A subset of 9 of the differentially expressed transcripts in both hippocampus and cerebellum overlapped with the hypothalamically identified cPRGs, without being differentially expressed in male and female tissue (Figure 2B: 4833420G17Rik, Cetn4, Cyb5r3, Gm5148,Mrpl20,Mid1, Pisd-ps3, Pttg1, and Xaf1). A further fourteen genes (1700048O20Rik, Abdh1, Adat2, Gvin1, Irgm2, Itgb3bp, Kbtbd11, Lacc1, Nmrk1, Pyurf, Slc25a37, Slc39a2, Stxbp2, Zfp125) were commonly regulated between hippocampus and cerebellum but were not hypothalamic cPRGs. The cPRG transcripts most dramatically down-regulated (2-fold or more) are Pttg1, a transcriptional activator associated with development (Manyes et al., 2018); Mid1, implicated in structural brain defects, Huntington’s and Alzheimer’s pathology (Winter et al., 2016), 4833420G17Rik, a predicted protein (orf) of unknown function, and Pisd-ps3, phosphatidylserine decarboxylase pseudogene 3, regulated inaging (Shema et al., 2015). Mrpl20 and Xaf1 are genes encoding known proteins of unknown function in the nervous system.

Adcyap1 itself does not register as significantly downregulated in hippocampus (Figure 2A). This is an artefact of the threshold for detection of rare transcripts by microarray. As Adcyap1 is expressed at already very low levels in hippocampus and cerebellum, compared to hypothalamus (Figure 3A), its apparent fold-change is constrained by the dynamic range of microarray in these tissues (Figure 3A, B). Indeed, qRT-PCR for expression of Adcyap1 in hypothalamus, cerebellum, and hippocampus in WT vs. PACAP KO mice shows a main effect of genotype (F(1,53), p < 0.0001, with significant differences between PACAP KO and WT mice in hypothalamus (p = .0026), cerebellum (p = 0.014), and hippocampus (p = .0081) (Figure 3B), but not in adrenal gland (not shown), in which Adcyap1 is not detectable in our assay (average Ct value of 31.9 in WT and 32.8 in KO mice; see Table 3).

Figure 3. Comparison of expression values for transcripts encoding the cPRGs Pttg1, Mid1, Xaf1, Pla2g4e, and Adcyap1, and independent verification of regulation assessed by qRT-PCR and in-situ hybridization.

A) Log2 expression values (mean expression value of all animals, +/− standard error of the mean) of Adcyap1 in non-stressed PACAP knockout versus wildtype mice of both sexes, in hypothalamus (combined experiments 1 and 2), cerebellum, and hippocampus. Expression of Adcyap1 is higher in wildtype hypothalamus than in other neural tissues. Comparison of wildtype and PACAP KO mice within each tissue shows that tissues with lower baseline Adcyap1 expression have smaller reductions in Adcyap1 expression in PACAP KO versus wildtype mice (Stars indicate significance of ANOVA comparing WT expression values to KO expression values. In hypothalamus, p = 3.39e-13, fold- change = −6.55; in cerebellum p = 3.17e-4, fold-change = −1.14; but not in hippocampus p = 1.22e-2, fold-change −1.09). B) qRT-PCR, with larger dynamic range, reveals significant decreases in Adcyap1 expression in all PACAP producing tissues. Two-way ANOVA revealed a significant main effect of genotype F(1,33) = 476.9, p < 0.0001. Post-hoc Šídák’s comparisons shows differences between PACAP KO and WT animals in hippocampus, hypothalamus, and cerebellum, p < 0.0001 in all tissues. Bar graph depicts mean and standard error of fold-change calculated using the 2^−ΔΔCt technique with Gapdh as control. N = 6 mice per genotype in all tissues but hypothalamus, where N = 7 WT and 8 PACAP KO mice. C) Log2 expression values of Pttg1 in non-stressed PACAP knockout and wildtype mice of both sexes, in all microarray experiments. Putative cPRG Pttg1 is highly expressed in all tissues studied and is significantly reduced (ANOVA, p < 0.0001) in PACAP KO versus wildtype animals in all tissues. D) qRT-PCR confirms significant loss of Pttg1 in hypothalamus, cerebellum, hippocampus, and andadrenal gland (F(1,43) = 967.5, p < 0.0001, Post-hoc Šídák’s show p < 0.0001 for all comparisons). Bar graph depicts mean and standard error of fold-change calculated using the 2^−ΔΔCt technique with Gapdh as control. N = 6 mice per genotype in all tissues but hypothalamus, where N = 7 WT and 8 PACAP KO mice. Animals used are identical to those used in B. E) Log2 expression values, in hypothalamus, of Mid1, Xaf1, and Pla2g4e, identified as cPRGs, in hypothalamus. ANOVA indicates significant differences (p < 0.0001) in each gene in hypothalamus. F) qRT-PCR confirms that transcriptomic alterations are significant in the direction found in microarrays. Multiple t-tests with Bonferroni correction shows significant differences between WT and KO for Mid1 (t(13) = 10.97, adjusted p < 0.0001), Xaf1 (t(13) = 8.89, adjusted p = 0.000002) and Pla2g4e (t(13) = 3.156, adjusted p = 0.022). Bar graph depicts mean and standard error of fold-change calculated using the 2^−ΔΔCt technique with Gapdh as control. N = 7 WT and N = 8 PACAP KO mice. G, H) RNAscope in-situ hybridization (Max z-projection of stacks) visualizing mRNA for Pttg1 (red) and Xaf1 (green) are depicted in lateral hypothalamus of one WT (G) and one PACAP KO (H) mouse. The decrease in Pttg1 expression and increase in Xaf1 expression are visually evident in the PACAP KO mouse.

The most consistently and dramatically downregulated cPRG, from the microarray experiments described above, was Pttg1. Its microarray expression value is significantly depressed in PACAP KO versus WT animals in all tissues studied (Figure 3C). We confirmed its decreased expression in hippocampus, hypothalamus, cerebellum, and adrenal gland of PACAP KO mice using qRT-PCR. Two-way tissue × genotype ANOVA revealed a significant effect of genotype (F(1,43) = 967.5, p < 0.0001), with Šídák’s multiple comparisons showing differences at p < 0.0001 in all tissues (Figure 3D).

To confirm additional cPRGs (microarray expression in hypothalamus in Figure 3E), we used qRT-PCR to assess expression of Mid1, Xaf1, and Pla2g4e in hypothalamus. Unpaired t-tests with Benjamini, Krieger, and Yekutieli False Discovery Rate compensation for multiple testing were performed for each gene in each tissue, comparing PACAP KO and WT mice. All comparisons were significant (hypothalamic results shown in Figure 3F, remaining results in Tables 2 and 3).

Table 2:

qRT-PCR analysis of cPRGs, statistics

Gene	Tissue	P value	t ratio	df	q value
Mid1	Hypothalamus	<0.000001	10.97	13	<0.000001
Xaf1	Hypothalamus	<0.000001	8.892	13	0.000002
Pla2g4e	Hypothalamus	0.003913	3.561	12	0.004517
Xaf1	Hippocampus	<0.000001	8.892	13	0.000002
Mid1	Hippocampus	0.0001	6.212	10	0.000118
Xaf1	Cerebellum	0.000021	7.502	10	0.000036
Mid1	Cerebellum	0.000082	6.367	10	0.000116

3.2. Effects of restraint stress on the hypothalamic transcriptome in wild-type and PACAP-deficient mice

Examination of transcripts up-regulated by restraint stress, versus the non-stressed condition in WT and in PACAP KO mice allows detection of transcripts whose stress-associated induction is PACAP-dependent. In both PACAP KO and WT mice, a number of transcripts, including the immediate-early genes (IEGs) Nr4a1, Nr4a3, Sgk1, Cyr61, and Nfkbia were upregulated significantly (p < 0.01, fold-change +/− 1.5) by one hour of restraint stress (Figure 4A), demonstrating that there is an intact stress transcriptomic response in PACAP-deficient mice.

Figure 4. Effects of stress on hypothalamic transcriptome of PACAP KO versus WT mice.

A) Conserved stress response genes are regulated by stress in WT animals (comparison of WT stressed to WT non-stressed (purple) p < 0.01) and in PACAP KO mice (comparison of PACAPKO stressed to wildtype non-stressed (pink), p < 0.01). These genes do not differ between WT and PACAP KO mice without stress (yellow). B) A PACAP-dependent stress response is indicated by genes which are upregulated in response to stress among wildtype mice but fail to be recruited in PACAP KO mice. These genes do not differ between WT and PACAP KO mice without stress (yellow). Fold-changes are absolute values. All significant changes were in the positive direction, but in B, some comparisons between non-stressed animals were non-significantly negative.

Several other IEGs, however, were PACAP-dependent, as they were upregulated by stress in wildtype animals and not in PACAP knockout mice (Figure 4B). These were Rrad, Fos, Gem, Egr1, Fosb, Fosl2, Arc, Npas4, Gadd45g, and Plk2. Fos and Egr1 have previously been shown to be PACAP-dependent upon one hour of restraint stress in hypothalamus and adrenal gland (Stroth and Eiden, 2010). Because these genes are not altered in the baseline transcriptome of the PACAP KO mouse (Figure 1), lack of stress-induced regulation is due not to regulation of these target genes during development, but to acute effects of loss of PACAP transmitter function during stress in hypothalamus, making these genes acutely PACAP-Regulated Genes (aPRGs).

3.3. Effects of PACAP knockout on adrenal gland transcriptome

The adrenal gland is a neuroendocrine organ in which Adcyap1 mRNA is expressed at very low levels, but which expresses both specific (PAC1) as well as relatively non-specific (BAM22P) receptors for PACAP. It is copiously innervated by the splanchnic nerve, which allows PACAP release onto adrenomedullary chromaffin cells throughout the lifespan, beginning in early postnatal life, and may therefore be altered by the loss of PACAP signaling throughout development (Guo and Wakade, 1994; Holgert et al., 1994; Kesse et al., 1988). We identified cPRGs in adrenal gland, in which PACAP KO and WT mice differed significantly (p < 0.01, fold-change +/− 1.5). Of the 93 differentially regulated genes (70 downregulated and 23 upregulated), 7 genes had been identified as cPRGs in our hypothalamic analysis. These genes are 4833420G17Rik, Cd59a, Cyb5r3, Mrpl20, Pisd-ps3, Pttg1, and Tmem260. Three genes were members of the catecholamine biosynthesis pathway (Th, Pmnt, and Hand1). These final three may not be cPRGs, strictly speaking, as developmental and adult functions at the adrenomedullary synapse may significantly overlap.

The adrenal glands are the major peripheral effectors of hypothalamic stress signaling (Goldstein and Kopin, 2007; Goldstein and Kopin, 2008), and PACAP is a known regulator of these circuits (Agarwal et al., 2005; Kuri et al., 2009; Stroth and Eiden, 2010). We therefore examined the stress-regulated genes in adrenal gland that were either PACAP-dependent (expression increased significantly, p < 0.01, at a fold-change of 1.5 or more in WT but not PACAP KO mice) or PACAP-independent (increased in both PACAP KO and WT animals). Figure 5A shows a subset of the 192 genes (those most upregulated by 3 hours of restraint stress, fold change > 3, p <= 0.01) in both PACAP-deficient and WT adrenal gland. Figure 5B depicts a subset of the 303 PACAP-dependent adrenal transcripts, for which WT animals showed increased transcription (fold-change > 3, p <= 0.01 depicted) but PACAP KO animals did not, and for which expression did not differ between non-stressed PACAP KO and WT mice. As in hypothalamus, genes regulated acutely by PACAP (aPRGs) can be distinguished from those whose expression is altered by constitutive loss of PACAP (cPRGs): Only two genes were altered in the non-stressed comparison and in the stressed comparison (Scgn and Pcsk1).

Figure 5. Effects of PACAP knockout and stress on adrenal transcriptome.

A) Conserved stress response genes are regulated by 3 hours of restraint stress in WT animals (comparison of WT stressed to WT non-stressed (purple), p < 0.01, fold-change +/− 3 or greater depicted here) and in PACAP KO mice (comparison of PACAP KO stressed to wildtype non-stressed (yellow), p <= 0.01, fold-change +/−3 or greater depicted here). Red line indicates fold-change threshold of 3 used for graphing, but note that in text (section 3.3) a threshold of 1.5-fold reveals many more conserved genes. B) A PACAP-dependent stress response is indicated by genes which are upregulated in response 3 hours of restraint stress among WT mice (comparison of WT stressed to WT non-stressed, p <= 0.01, fold-change +/− 3 or greater depicted here) but fail to be recruited in PACAP KO mice. Red line indicates fold-change threshold of 1.5. Genes regulated by more than 1.5-fold in PACAP KO stressed vs WT non-stressed mice (yellow bars extending above red line) fail to meet the p-value significance cutoff of p <= 0.01. Comparison of PACAPKO stressed to WT non-stressed not depicted in this figure, but no genes included differ between genotypes in the non-stressed condition.

One gene regulated by stress in WT but not PACAP KO mice is immediate early gene 3 (Ier3). Ier3 has also been referred to as PACAP-regulated gene 1, or Prg1, because it was identified as a gene up-regulated by PACAP in rodent pancreatic tumor cells (Schafer et al. et al., 1996). Another regulated gene is Stc1: PACAP regulates expression of both Ier3 and Stc1 in bovine chromaffin cells in culture (Ait-Ali et al., 2010; Eiden et al., 2008; Samal et al., 2007) and in cultured rodent cortical neurons . We confirmed these aPRGs with qRT-PCR (Figure 6A’ and A”). The alteration in Stc1 is adrenal-specific and does not occur in hypothalamus in either microarray (Figure 6B) or qRT-PCR analysis (Figure 6B’, B”).

Figure 6. Adrenal gland, stress, and regulation of the aPRGs Stc1 and Ier3.

A) Log2 expression values of IEGs Ier3 and Stc1 in adrenal gland microarray show increased Ier3 and Stc1 following 3 hours of restraint in wildtype but not PACAP knockout mice, relative to non-stressed wildtype mice. A’) Two-way ANOVA analysis of qRT-PCR for adrenal Stc1 shows a significant interaction between time restrained and genotype F(2,25) = 6.99, p = 0.0039: While there was no difference between non-stressed WT (N = 4) and PACAP KO mice (N = 5), Šídák’s test showed significant increases in WT (p < 0.0001), but not in PACAP KO (p = 0.419) adrenal gland following 1 hour restraint stress (N = 6 stressed mice of each genotype). Wildtype animals, however, had significantly higher Stc1 expression (p = 0.0035) than PACAP KO mice at 1-hour stress. At 6 hours of stress (N = 6 mice of each genotype), Stc1 levels had dropped relative to 1-hour stress significantly in wildtype (p = 0.0015) mice, returning to baseline levels. Asterisks indicate significant difference between genotypes within timepoints, while pound signs indicate significant difference between untreated and 1 hour stressed animals within each genotype. A”) qRT-PCR shows that immediate early gene Ier3 is increased after 1-hour restraint stress in wildtype and PACAP KO adrenal gland. Two-way ANOVA shows a significant interaction between time restrained and genotype F(2,25) = 15.60, p < 0.0001: While there was no difference between non-stressed wildtype and PACAP KO mice, Šídák’s test showed significant increases in both WT, p < 0.0001, and PACAP KO p = 0.0002 adrenal gland after 1 hour stress, compared to non-stressed animals. Wildtype animals, however, had significantly higher Ier3 expression after stress (p < 0.0001) than PACAP KO mice. By 6 hours of restraint, Ier3 levels had dropped relative to 1-hour stress significantly in wildtype (p < 0.001) and PACAP KO (p = 0.0004) animals, returning to baseline levels. Asterisks indicate significant difference between genotypes within timepoints, while pound signs indicate significant difference between untreated and 1 hour stressed animals within each genotype. Animals are the same as in A’. B) Log2 expression values of IEGs Ier3 and Stc1 in hypothalamus show no significant difference in wildtype and PACAPKO regulation of Ier3 or Stc1 following 3 hours of restraintrelative to non-stressed wildtype mice. B’, B”) qRT PCR confirms no alteration of Stc1 or Ier3 in either wildtype or PACAP KO mice after 1 or 6 hours of restraint stress.

3.4. Behavioral and physiological responses in PACAP-deficient mice

Finally, we examined the possible links between the transcriptomic impact of PACAP deficiency on cPRGs and aPRGs, and mouse spontaneous and stress-elicited behaviors. Figures 7A and 7B depict the acute effects of PACAP deficiency on corticosterone levels, and on weight loss secondary to decreased food intake (hypophagia), as measured 24 hours after three hours of restraint stress. As reported previously for chronic stress effects on feeding (Jiang and Eiden, 2016; Mustafa et al., 2015), PACAP deficiency abrogates weight loss in both male and female PACAP-deficient mice and attenuates corticosterone elevation after three hours of restraint. In contrast, repetitive jumping appears to be a constitutively established behavior not triggered by a specific physiological event such as restraint stress: With no discrete triggers, PACAP KO mice perform stereotyped repetitive jumping (Hattori et al., 2012) throughout the dark cycle, up to 147 times per half hour period in the current study, while control mice (PACAP^fl/fl) repetitively jumped significantly less (Figure 7C, 7C’).

3.5. Behavioral phenocopying in conditional PACAP- and in constitutive PAC1-deficient mice

Because acute effects of PACAP transmission occur primarily via the PAC1 receptor, we examined the phenocopying of the above features of PACAP KO mice by constitutive PAC1- deficient mice (PAC1 KO). The abrogation of stress-induced weight loss is phenocopied in PAC1-deficient mice, suggesting that the effect depends on intact PACAP-PAC1 signaling. Attenuation of corticosterone increase was also evident in the PAC1 KO mouse. On the other hand, repetitive jumping, does not occur in PAC1-deficient mice (Figure 7E, F), suggesting that repetitive jumping stems from long-term effects of PACAP deficiency arising from altered expression of cPRGs.

We wished to further correlate the potential role(s) of cPRGs and aPRGs in PACAP control of physiological responses and behavioral phenotypes in the mouse. Anticipating that cPRGs may have an indirect effect on function due to changes in the basal transcriptome, we sought to understand how early in development PACAP control of the transcription of genes such as Pttg1, and of behavioral outputs like repetitive jumping might ensue. We created two conditional knockout mice to investigate, by crossing a newly-developed conditional PACAP knock-out mouse (PACAP^fl/fl), to two CRE-driver mice; the CaMK2α PACAP^fl/fl mouse in which CRE expression, and therefore PACAP knock-out, occurs under the control of the CaMK2α promoter, which becomes active in forebrain during early postnatal life (Burgin et al., 1990), and the VGaT-CRE PACAP^fl/fl mouse, in which PACAP-knockout ensues postnatally in cerebellum (Takayama and Inoue, 2005). This allows us to examine the behavioral and transcriptomic effects of postnatal PACAP loss in neurons expressing either CaMK2α-promoter-driven CRE or VGaT-promoter-driven CRE

We measured, using qRT-PCR, levels of Adcyap1 and Pttg1 in prefrontal cortex (PFC) and cerebellum of each line. In VGaT-CRE PACAP^fl/fl animals, Adcyap1 was reduced in both tissues, although more dramatically in cerebellum where more VGaT expression is expected. In CaMK2α-CRE PACAP^fl/fl mice, Adcyap1 transcript levels are reduced by more than eighty percent in prefrontal cortex, and not at all in cerebellum, where PACAP expression is largely limited to expression in inhibitory neurons (Purkinje cells) (Allen Cell Types, 2015; Zhang et al., 2020). Pttg1 expression was unaffected by conditional PACAP-deficiency in either tissue in either knockout line (Figure 7C, D). It may be that control of Pttg1 (and other cPRG) transcription by PACAP is supported by PACAP release from cell types in which these promoters are not active. Nevertheless, the complete lack of effect on Pttg1 expression in brain regions in which post-natal loss of PACAP expression is profound, does suggest that PACAP plays an earlier-than-postnatal role in regulation of aPRGs such as Pttg1.

Behaviorally, while constitutive PACAP KO mice perform hundreds of repetitive jumping bouts per half-hour period, neither control mice (PACAP floxed, CRE negative) nor either conditional knockout have the repetitive jumping phenotype (Figure 7E, E’). This suggests that Pttg1, or other aPRGs, may be functionally important in this phenotype, although reverse genetic analysis will be required to demonstrate this.

4. Discussion

The neuropeptide PACAP, like several other neuropeptides, is expressed during embryonic development (Sheward et al., 1996; Skoglösa et al., 1999; Waschek et al., 1998), and throughout adult life, in both peripheral and central nervous systems. PACAP functions as a neurotransmitter upon co-release with acetylcholine from preganglionic sympathetic and parasympathetic neurons, with glutamate and other neuropeptides from sensory neurons, and with either glutamate or GABA from both projection and interneurons of the CNS (Beaulé et al., 2009; Hamelink et al., 2002; Hannibal, 2002; Moller and Sundler, 1996; Zhang et al., 2020). How PACAP and other neuropeptides act as instructive or trophic factors, at various stages of development, is less well understood (Botia et al.; Maduna and Lelievre, 2016; Shioda et al., 2006; Waschek et al., 1998). Here, we have compared the brain and adrenal transcriptomes of wild-type mice and constitutively PACAP-deficient mice (i.e. mice lacking PACAP from conception) in the quiescent (basal) state, and after restraint stress. Transcriptomic analysis revealed that PACAP is required to support constitutive expression of a small cohort of developmentally regulated genes potentially related to tonic motor activity, and again in adulthood to support gene induction linked to stress responding. A core group of transcripts were identified that are up- or down-regulated in PACAP KO compared to WT mice in hypothalamus, hippocampus and cerebellum, under basal conditions, i.e. in quiescent mice. Human variants in PACAP have been linked to major depressive disorder (Hashimoto, 2009) and schizophrenia (Hashimoto, 2007), treatment efficacy in generalized anxiety disorder (Cooper, 2013), and increased alcohol consumption (Kovanen, 2010), suggesting translationally relevant long-term effects of PACAP dysregulation.

Two prominent transcripts are the transcription factor Pttg1 (Manyes et al., 2018; Tong and Eigler, 2009), which is down-regulated in constitutively PACAP-deficient mice, and the caspase-binding protein Xaf1 (Liston et al., 2001; Xia et al., 2006), which is up-regulated. We refer to these genes as constitutive PACAP-regulated genes, or cPRGs, indicating their likely regulation by PACAP during development, when PACAP acts as a neurotrophic factor, rather than as a neurotransmitter released at fully formed synapses in the CNS. Abundance of these transcripts in brain is not affected if PACAP knockout in either excitatory or inhibitory neurons occurs late in development. Regulation of cPRGs appears to be correlated with a spontaneous motor activity (repetitive jumping) in PACAP-deficient mice, and, like cPRG regulation, this effect is not reproduced in mice in which PACAP knock-out occurs selectively in neurons expressing either CaMK2α-promoter-driven CRE or VGaT-promoter-driven CRE. The mechanism of these alterations in transcription are unknown and may be a result of these genes being directly targeted by PACAP, or by compensatory mechanisms or other side effects of PACAP loss.

A second important feature of cPRG regulation and repetitive jumping is that they are not phenocopied in constitutively PAC1-deficient mice (Figure 7E, E’). Other receptors for PACAP exist, such as VPAC1 and VPAC2 and have been implicated in PACAP action in vivo (Rasbach et al., 2018). MRG3B, the BAM22P enkephalin receptor, which acts as a low-affinity receptor for multiple neuropeptides including PACAP (Foster et al., 2019), is less well characterized with respect to PACAP signaling, but is a plausible candidate as a receptor through which PACAP might exert its developmental actions. This receptor is restricted to primary sensory neurons in the adult (Lembo et al., 2002), but may be more widespread earlier in development.

In contrast to cPRGS, acute PACAP-regulated genes, or aPRGs, are identified here as genes whose transcripts are equivalently expressed in both WT and PACAP KO mice at baseline but require PACAP expression for their induction in response to a physiological stimulation. We and others have previously shown that chromaffin cells of the adrenal medulla are targeted by cholinergic/PACAPergic innervation, and PACAP has a critical role in mediating stress transduction by the chromaffin cell (Hamelink et al., 2002; Rasbach et al., 2018; Stroth et al., 2013; Stroth et al., 2011b). We have previously examined the PACAP-regulated transcriptome in both primary chromaffin cells, and in pheochromocytoma cells induced to differentiate by this neuropeptide (Botia et al., 2008; Samal et al., 2007; Vaudry et al., 2002). Genes regulated immediately upon exposure to PACAP include Ier3 and the neuroprotective calcium storage modulator stanniocalcin (Stc1) (Ait-Ali et al., 2010). Ier3, also known as PRG-1, was the first reported PACAP-regulated gene (Schafer et al. et al., 1996). These genes were PACAP-dependent aPRGs in adrenal gland following restraint stress.

The roles of PACAP in endocrine and behavioral stress responding, as revealed in PACAP-deficient mice, have generally been interpreted as reflecting loss of PACAP neurotransmitter function, rather than as indirect sequelae of PACAP-dependent developmental events (Jiang and Eiden, 2016; Mustafa et al., 2015; Sherwood et al., 2003; Stroth and Eiden, 2010; Stroth et al., 2013; Tsukiyama et al., 2011). This interpretation is consistent with PACAP-dependent stress responding in rats, in which responses are abrogated by PACAP(6–38) infusion, or mimicked by PACAP infusion, in stress-responsive brain regions at the time of experiment (Agarwal et al., 2005; Grinevich et al., 1997; Lezak et al., 2014; Miles et al., 2018; Norrholm et al., 2005; Roman et al., 2014; Seiglie et al., 2019). Restraint stress causes up-regulation of multiple genes in the hypothalamus of wildtype animals, as in the adrenal medulla. Unlike cPRGs, aPRGS appear to be highly tissue-specific upon comparison of the hypothalamic and adrenal transcriptomes after restraint stress. In hypothalamus, several genes are prominently upregulated by restraint stress in WT but not in PACAP KO mice, including the immediate early genes Rrad, Fos, Gem, Egr1, Fosb, and Fosl2.

In addition to its effects in promoting HPA axis activation after stress, PACAP actions in CNS extend to modulation of behavioral responses to stress, including anxiety and depression. These effects are mediated through both the hypothalamus and the extended amygdala (Agarwal et al., 2005; Lezak et al., 2014; Miles et al., 2018; Norrholm et al., 2005; Roman et al., 2014; Seiglie et al., 2019). HPA axis activation after three hours of restraint (CORT elevation) is partially dependent upon PACAP, and this effect is phenocopied in PAC1 KO mice. Hypophagia induced by restraint stress is fully abolished in both PACAP KO mice, and this effect is likewise phenocopied by PAC1 KO. PACAP modulates the HPA axis (endocrine stress response) indirectly, by activating CRH transcription (Agarwal et al., 2005; Grinevich et al., 1997; Stroth and Eiden, 2010), regulating the supply of CRH available for PACAP-independent release into the portal circulation (Jiang and Eiden, 2016).

aPRGs associated with stress-induced hypophagia are more likely to be identified in extra-hypothalamic areas associated with PACAP-dependent stress responding. Interestingly, in rat, it has been reported that PACAP-dependent activation of CRH transcription occurs in BNST and central amygdala. This transcriptional regulation is associated with anxiety behaviors, but not hypophagia, following stress (Dore et al., 2013). Similarly, a role for PACAP projections from frontal cortex to diencephalon is postulated to mediate behavioral coping styles through modulation of cortical drive, based upon in vivo infusion of PACAP in infralimbic cortex at time of assessment of stress-induced behavior (Martelle et al., 2021). Although further investigation is required, it seems likely that aPRGs, rather than cPRGs, participate in these PACAP-dependent physiological responses.

Our data suggest that two discrete roles exist for the neuropeptide PACAP in modulating transcriptomic regulation of the nervous system. One of these is developmental and is associated with constitutively PACAP-regulated genes (cPRGs). At least one effect of PACAP during early development appears to be in control of motor output, as evidenced by repetitive jumping behavior following early (constitutive) but not later developmental PACAP deficiency. PACAP also has a neurotransmitter/neuromodulator role in the adult nervous system, which is manifested in the control of the stress response. This is associated with transcriptomic regulation of acute PACAP-regulated genes (aPRGs), and occurs both centrally and peripherally, as reflected in the PACAP-dependent transcriptomic response to stress in both hypothalamus and adrenal gland (Stroth and Eiden, 2010; Stroth et al., 2013).

The whole-mouse knockout of the PACAP gene has yielded significant information about the roles of this ‘master regulator’ of the stress response (see (Stroth et al., 2011a) and references therein). Complementary experiments in conditional knock-out mice in which PACAP is eliminated in specific brain regions and at different stages of development will reveal more about the dual role of this neuropeptide in trophic/developmental specification of neuronal and endocrine properties on the one hand, and real-time neurotransmitter effects in specific physiological responses on the other. These further investigations may also illuminate mechanisms of dysregulation of stress responding throughout the lifespan (Chen and Baram, 2016; Kosten et al., 2000; Rice et al., 2008; Veenema et al., 2008; Zhang et al., 2012)

The distinction between two modes of PACAP signaling to the genome, reported here, are relevant to both developmental and adult actions of PACAP. It remains to be determined if the two discrete types of neuropeptide function, constitutive and acute, exist for other neuropeptides expressed in the mammalian brain during both development and in the adult, and, if so, whether diverse signaling mechanisms, and even different sets of neuropeptide-responsive receptors, are required for each, as appears to be the case for PACAP control of cPRG and aPRG transcripts across the lifespan and in different regions of the central and peripheral nervous systems.

Supplementary Material

1

Supplemental Figure 1. Identification of sex-dependent PACAP-regulated genes in the hypothalamus. Comparison of expression of male and female hypothalamic transcriptomes derived in this study to literature-reported sex-dependent gene expression in mouse brain. Following a two-way sex by genotype ANOVA, genes in our hypothalamic experiment 2, including male and female animals, were selected if females and males differed significantly (p < 0.01) in expression value. Data from Mozhui, 2012 (Mozhui et al., 2012), comparing hypothalamic transcripts of 39 pairs of adult BDX-strain mice, was retrieved and genes with a significant difference in expression between males and females (p < 0.01) were extracted. Fifteen genes were differentially regulated by sex in both experiments. These genes matched well in both direction and magnitude between the studies.

2

Supplemental Figure 2. Dissection region used for hypothalamus.

A) Turning the brain to the ventral aspect, cuts were made laterally at the optic chiasm and above the brainstem.

B) Hypothalamic regions indicated (yellow) were then scooped out using forceps, resulting in a sample of tissue approximately 3 mm × 2 mm × 2 mm in size.

Highlights.

• Constitutive knockout of PACAP results in transcriptomic and behavioral alterations that are distinct from the acute effects of PACAP deficiency during stress responding.

• Some developmentally mediated effects of constitutive PACAP knockout are not phenocopied by PAC1 knockout.

• Comparison of neuropeptide-mediated developmental and adult transcriptomic and behavioral effects reveals the fundamentally distinct roles of neuropeptides as instructive/trophic factors, and as neurotransmitters, respectively.