Influence of prenatal transportation stress-induced differential DNA methylation on the physiological control of behavior and stress response in suckling Brahman bull calves

Abstract

The objective of this experiment was to examine potential differential methylation of DNA as a mechanism for altered behavioral and stress responses in prenatally stressed (PNS) compared with nonprenatally stressed (Control) young bull calves. Mature Brahman cows (n = 48) were transported for 2-h periods at 60 ± 5, 80 ± 5, 100 ± 5, 120 ± 5, and 140 ± 5 d of gestation (Transported group) or maintained as nontransported Controls (n = 48). From the offspring born to Transported and Control cows, a subset of 28-d-old intact bulls (n = 7 PNS; n = 7 Control) were evaluated for methylation of DNA of behavior and stress response–associated genes. Methylation of DNA from white blood cells was assessed via reduced representation bisulfite sequencing methods. Because increased methylation of DNA within gene promoter regions has been associated with decreased transcriptional activity of the corresponding gene, differentially methylated (P ≤ 0.05) CG sites (cytosine followed by a guanine nucleotide) located within promoter regions (n = 1,205) were used to predict (using Ingenuity Pathway Analysis software) alterations to canonical pathways in PNS compared with Control bull calves. Among differentially methylated genes (P ≤ 0.05) related to behavior and the stress response were OPRK1, OPRM1, PENK, POMC, NR3C2, TH, DRD1, DRD5, COMT, HTR6, HTR5A, GABRA4, GABRQ, and GAD2. Among altered (P < 0.05) signaling pathways related to behavior and the stress response were Opioid Signaling, Corticotropin-Releasing Hormone Signaling, Dopamine Receptor Signaling, Dopamine-DARPP32 Feedback in cAMP Signaling, Serotonin Receptor Signaling, and GABA Receptor Signaling. Alterations to behavior and stress response–related genes and canonical pathways supported previously observed elevations in temperament score and serum cortisol through weaning in the larger population of PNS calves from which bulls in this study were derived. Differential methylation of DNA and predicted alterations to behavior and stress response–related pathways in PNS compared with Control bull calves suggest epigenetic programming of behavior and the stress response in utero.

Introduction

Alteration of behavior and stress responsiveness is among the most noted outcomes of prenatal stress (PNS) in mammals (Lindsay et al., 2019; Matthews and McGowan, 2019). The mechanisms controlling these alterations are still largely unclear; however, it is known that stress incurred by a gestating dam can result in suppression of the placental barrier enzyme 11β-hydroxysteroid dehydrogenase type 2 (11-βHSD2). This enzyme converts cortisol to an inactive form prior to entering fetal circulation; therefore, decreased 11-βHSD2 can result in increased fetal exposure to corticosteroids (Benediktsson et al., 1997; Stirrat et al., 2018). Such alterations to the fetal environment may result in epigenetic modifications, such as DNA methylation, which appear to be major regulators of phenotypic changes due to prenatal stress (Szyf, 2012; Lester and Marist, 2018). Deoxyribonucleic acid methylation is defined as a covalent modification in which a methyl group is added to the carbon 5 position of a cytosine nucleotide (Wyatt, 1950; Razin and Riggs, 1980). Increased methylation of DNA within the promoter region of a gene has been reported to suppress transcriptional activity and gene expression (Levine et al., 1991; Tate and Bird, 1993), whereas increased methylation of DNA in the gene body region of a gene has been suggested to result in increased gene expression (Hellman and Chess, 2007). Therefore, the presence or absence of methyl groups at specific locations within a gene can be utilized as a tool to better understand how prenatal environment shaped postnatal phenotype. Behavior and stress response traits are multifaceted in nature and controlled by many components of neural, neuroendocrine, and endocrine systems (Chan et al., 2018; Bale et al., 2019). Some of these systems include the hypothalamic–pituitary–adrenal (HPA) axis and signaling by catecholamines, monoamines, and opioids. Differential transcript abundance in the adrenal cortex (Friedrich et al., 2017) and differential methylation of DNA in different brain regions (Cantrell et al., 2019) have been associated with temperament in cattle. Furthermore, emerging literature supports the ability of glucocorticoids to alter methylation of DNA in developing neural cells; specifically, in vitro exposure of early neural progenitor cells to dexamethasone resulted in genome-wide differences in methylation of DNA (Provençal et al., 2019). Because behavior and stress response phenotypes are of economic and biologic significance to beef cattle production, this study sought to understand the mechanisms by which a common stressor for beef cattle, transportation, incurred by a pregnant dam could affect behavior and the stress response of offspring. The objective of this study was to identify differentially methylated genes from a previously reported genome-wide assessment of DNA methylation in prenatally stressed bull calves that were related to behavior and stress responses (Littlejohn et al., 2018). Because behavior and stress response phenotypes were altered in the larger population of PNS and Control calves from the same study (Littlejohn et al., 2016), we hypothesized that prenatal transportation stress would result in differential methylation of DNA in genes associated with behavior and stress response phenotypes.

Materials and Methods

All procedures were in compliance with the Guide for the Care and Use of Agricultural Animals in Research and Teaching (FASS, 2010) and approved by the Texas A&M University Animal Care and Use Committee.

Animal Procedures

Mature Brahman cows were assigned to 1 of 2 treatment groups (Transported: n = 48 and Control: n = 48), which were balanced by age, parity, and temperament (Littlejohn et al., 2016). Transported cows were hauled for 2 h at 60 ± 5, 80 ± 5, 100 ± 5, 120 ± 5, and 140 ± 5 d of gestation (Price et al., 2015). Control cows were managed the same as Transported cows with the exception of being transported. Both treatment groups were housed in the same pasture (at the Texas A&M AgriLife Research and Extension Center at Overton, TX) and fed the same diet (Littlejohn et al., 2016). From these groups, 26 male and 18 female calves (Control group) were born to Control dams, whereas 20 male and 21 female calves (Prenatally Stressed, PNS, group) were born to Transported dams. Male calves were maintained as bulls throughout the study. Each calf was manually restrained for less than 5 min for sample collection at 28 d of age. One 10-mL vacuum tube (BD, Franklin Lakes, NJ) containing EDTA was used to collect the blood sample from each calf via jugular venipuncture with a sterile 18-gauge needle. Leukocytes were isolated and stored at −80 °C until DNA was extracted. Samples from bull calves were stratified into groups based on the following criteria: 1) adequate availability of DNA, 2) calf treatment group, and 3) calf sire. From those stratified groups, DNA samples from 7 PNS and 7 Control bull calves were randomly selected for analysis in this study.

Sample Analysis

Processing of blood samples

Blood samples were centrifuged at 2,671 × g for 30 min at 6 °C. Leukocytes were isolated and transferred into 2-mL nuclease-free microcentrifuge tubes. Samples were washed in red blood cell lysis buffer solution repeatedly until a clean cell pellet was produced. Leukocyte samples were stored at −80 °C until they were thawed for DNA isolation procedures.

DNA extraction

Phenol-chloroform extraction procedures were used to isolate DNA from each leukocyte pellet for DNA methylation analysis by standard procedures as previously described (Littlejohn et al., 2018). Following DNA precipitation, purified DNA was stored at −80 °C and submitted to Zymo Research Corp (Irvine, CA) for DNA methylation analysis.

Library construction for DNA methylation analysis

A method of reduced representation bisulfite sequencing, Methyl-MiniSeq (Zymo Research; Irvine, CA), was utilized to evaluate differential methylation of DNA in samples from PNS compared with Control bull calves. Libraries were prepared from 200 to 500 ng of genomic DNA digested with 60 units of TaqαI and 30 units of MspI (New England Biolabs, Inc., Ipswich, MA) sequentially and then extracted with Zymo Research DNA Clean & Concentrator-5 kit (Cat#: D4003). Fragments were ligated to preannealed adapters containing 5′-methyl-cytosine instead of cytosine according to Illumina’s specified guidelines (www.illumina.com). Adaptor-ligated fragments of 150–250 bp and 250–350 bp in size were recovered from a 2.5% NuSieve 1:1 agarose gel (ZymocleanTM Gel DNA Recovery Kit, Zymo Research, Cat#: D4001). Then, the fragments were bisulfite-treated using the EZ DNA Methylation-Lightning Kit (Zymo Research, Cat#: D5020). Preparative-scale PCR was performed and the resulting products were purified (DNA Clean & Concentrator—Zymo Research, Cat#D4005) and sequenced on an Illumina HiSeq.

Methyl-MiniSeq sequence alignments and data analysis

Sequence reads from bisulfite-treated EpiQuest libraries were identified with standard Illumina base-calling software and analyzed with a proprietary analysis pipeline (Zymo Research), which is written in Python. Bismark (http://www.bioinformatics.babraham.ac.uk/projects/bismark/) was utilized to perform the alignment to the UMD 3.1 bovine assembly (Zimin et al., 2009). Index files were constructed using the bismark_genome_preparation command and the entire reference genome. The --non_directional parameter was applied while running Bismark. All other parameters were set to default. Filled-in nucleotides were trimmed off for methylation calling. The methylation ratio was defined as the measured number of cytosines (number of reads reporting a C) divided by the total number of cytosines (total number of reads reporting a C or T) covered at that site. The methylation difference (MD) was calculated by subtracting the average Control methylation ratio at a site from the average PNS methylation ratio at that same site. Fisher’s exact test or t-test was performed for each CpG site which had at least 5 reads coverage. Promoter, gene body and CpG island annotations were added for each CpG included in the comparison. A summary of genome-wide coverage has been previously described (Littlejohn et al., 2018). No multiple test correction was performed because multiple test corrections have been suggested to be too conservative for reduced representation bisulfite sequencing (Kanzleiter et al., 2015). Specifically, in spite of decreasing false positives, multiple test corrections would also unnecessarily increase false negatives (Rothman, 1990; Kanzleiter et al., 2015). Instead, CpG site methylation differences and P-value thresholds were utilized as criteria for inclusion in the enrichment analysis to assess potential statistical and biological significance.

Prediction of pathways and functions altered by prenatal stress

An enrichment analysis was performed (analyzed on 3 April 2018) with Ingenuity Pathway Analysis software (IPA; Redwood City, CA) to assess alterations to signaling pathways in PNS compared with Control bull calves. Because methylation of DNA in mammals primarily occurs within CpG contexts (a cytosine nucleotide followed by a guanine nucleotide; Ehrlic et al., 1982), we focused on DNA methylation of cytosines within CpG contexts. Only differentially methylated CpG sites (P ≤ 0.05) that had a methylation difference that was ≥ 10% and located within a promoter region were included in the enrichment analysis. Only CpG sites located within a promoter region were utilized in the analysis because DNA methylation within the promoter region has been established to regulate gene activity. Specifically, increased methylation of DNA within the promoter region has been reported to result in suppressed transcriptional activity and gene expression (Levine et al., 1991; Tate and Bird, 1993). Based on this relationship, hypomethylation within gene promoter regions was assumed to result in increased gene expression, whereas hypermethylation within gene promoter regions was assumed to result in decreased gene expression. It is important to note that genes can be inhibitors or activators of specific cellular processes. Genes, canonical pathways, and function groups related to the physiological control of behavior and stress response were identified by IPA software.

Results and Discussion

Previously reported alterations to behavior and stress response phenotypes in the larger population from which calves in this study were derived (Littlejohn et al., 2016) suggested coinciding alterations to the physiological systems regulating those phenotypes. To evaluate potential mechanisms controlling such phenotypic differences, leukocytes were used as a generalized surrogate cell to evaluate methylation of DNA. The following sections evaluate and discuss the potential influence of PNS-induced differences in methylation of DNA on behavior and stress response phenotypes in Brahman cattle.

Influence of Prenatal Stress on Neural, Neuroendocrine, and Endocrine Function

Many function terms, canonical pathways, and specific genes related to behavior, the stress response, and neural function were altered (P < 0.05) in PNS compared with Control bull calves. A list of canonical pathways related to behavior, the stress response, and neural function that were predicted to be altered in PNS compared with Control bull calves are presented in Table 1. A list of IPA-designated genes associated with the Behavior function term that contained CpG sites within the promoter region that were hypomethylated or hypermethylated in PNS compared with Control bulls are presented in Tables 2 and 3, respectively. In general, opioid signaling, HPA axis signaling, dopamine signaling, serotonin signaling, and γ-aminobutyric acid (GABA) signaling were among the most notable pathways related to behavior, the stress response, and neural function that were predicted to be altered in PNS compared with Control bulls and will be discussed in the following paragraphs.

Table 1.

Canonical pathways related to neural function, behavior, and/or stress response that were significantly altered in PNS compared with Control bull calves¹

Ingenuity Canonical Pathways	-log (P-value)	Ratio	Z-score	Differentially methylated genes (P ≤ 0.05)
Opioid Signaling	8.48	0.119	0.392	RGS9,PLD2,CAMK1,POMC,AP2A2,SLC12A5,PRKCZ,TH,YES1,LCK,CACNA1E,PRKCE,ADCY8,CACNG6,GNAS,OPRM1,EGR4,CREBBP,CREB3L3,RAC1,ADCY6,ADCY9,CACNA1B,KCNJ9,ADCY1,PENK,OPRK1,PDE1B,CDKN1B,GNAL
Axonal Guidance Signaling	5.44	0.078	NaN	FZD10,MMP7,NTN4,BMP2,LIMK2,NFATC1,SEMA4C,PRKCZ,NTN1,LIMK1,FGFR3,PLCD3,ITGA3,RHOD,ADAM19,SMO,PRKCE,ADAM23,ERBB2,ADAMTS5,ADAMTS4,EPHA7,GNAS,EPHA1,BMP8A,WNT2B,RAC1,TUBA4A,EPHA10,WNT10A,PIK3R6,BMP7,SEMA3C,WNT1,GNAL,NTN3
GABA Receptor Signaling	5.07	0.139	NaN	GABRQ,CACNG6,GNAS,GABRA4,ADCY6,AP2A2,ADCY9,SLC6A11,GAD2,CACNA1E,CACNA1B,ADCY1,ADCY8,SLC6A12
Dopamine-DARPP32 Feedback in cAMP Signaling	4.39	0.101	2.183	PPP1R14C,GNAS,PPP2R5D,CREBBP,CREB3L3,ADCY6,CSNK1D,DRD5,PRKCZ,ADCY9,PLCD3,CACNA1E,PRKG1,DRD1,ADCY1,KCNJ9,PRKCE,ADCY8
CDK5 Signaling	4.31	0.126	1.732	PPP1R14C,GNAS,EGR1,PPP2R5D,ADCY6,DRD5,MAPK9,ADCY9,ITGA3,DRD1, ADCY1,ADCY8,GNAL
CRH Signaling	3.74	0.101	1.291	CACNG6,GNAS,CREB3L3,CREBBP,ADCY6,POMC,PRKCZ,ADCY9,CACNA1E,CACNA1B,NPR1,ADCY1,SMO,PRKCE,ADCY8
Huntington’s Disease Signaling	3.27	0.078	−0.302	NEUROD1,PSMF1,CREBBP,CREB3L3,MAPK9,HDAC10,HIP1,VTI1B,AP2A2,GPAA1,PRKCZ,ATP5F1E,FGFR3,DNAJC5,HDAC3,CACNA1B,PENK,CASQ1,PIK3R6,PRKCE
CREB Signaling in Neurons	3.23	0.082	2.138	CACNG6,GNAS,GRIK3,CREBBP,CREB3L3,ADCY6,PRKCZ,FGFR3,ADCY9,PLCD3,CACNA1E,CACNA1B,ADCY1,PIK3R6,PRKCE,ADCY8,GNAL,GRIK1
Agrin Interactions at Neuromuscular Junction	3.22	0.129	0	ITGA3,DVL1,RAC1,MAPK9,ERBB3,DAG1,ERBB2,CTTN,ACTA1
Dopamine Receptor Signaling	2.97	0.11	1.89	ADCY9,TH,PPP1R14C,DRD1,PPP2R5D,COMT,ADCY1,ADCY6,DRD5,ADCY8
Adrenomedullin Signaling	2.79	0.079	1.5	ADM,TFAP2C,GNAS,ADCY6,MAPK9,SHE,FGFR3,ADCY9,PLCD3,PRKG1,TFAP2A,NPR1,ADCY1,PIK3R6,RAMP3,ADCY8
Renin-Angiotensin Signaling	2.75	0.092	0.707	FGFR3,ADCY9,ADCY1,RAC1,REN,ADCY6,PIK3R6,MAPK9,PRKCE,ADCY8,SHE,PRKCZ
Synaptic Long Term Depression	2.38	0.077	0	CACNG6,GNAS,PPP2R5D,PLA2R1,PRKCZ,PLCD3,CACNA1E,PRKG1,LCAT,CACNA1B,NPR1,PLA2G4B,PRKCE,GNAL
Neuregulin Signaling	2.35	0.097	−1.134	HSP90B1,ITGA3,HBEGF,PRKCE,ERBB3,ERRFI1,ERBB2,CDKN1B,PRKCZ
Ephrin A Signaling	2.29	0.113	NaN	EPHA7,FGFR3,EPHA10,EPHA1,RAC1,PIK3R6,LIMK1
NRF2-mediated Oxidative Stress Response	2.18	0.073	1.134	NQO2,CREBBP,HSPB8,MAPK9,GSTT1,PRKCZ,FGFR3,DNAJC5,ERP29,SCARB1, KEAP1,PIK3R6,PRKCE,ACTA1
Serotonin Receptor Signaling	2.04	0.113	NaN	ADCY9,HTR6,HTR5A,ADCY1,ADCY6,ADCY8
Netrin Signaling	2.01	0.1	NaN	CACNG6,PRKG1,CACNA1E,CACNA1B,RAC1,NFATC1,NTN1
P2Y Purigenic Receptor Signaling Pathway	1.98	0.076	1.508	FGFR3,PLCD3,ADCY9,CREBBP,CREB3L3,ADCY1,ADCY6,PIK3R6,PRKCE,ADCY8,PRKCZ
Cardiac β-adrenergic Signaling	1.87	0.074	1.508	ADCY9,PPP1R14C,CACNA1E,GNAS,PPP2R5D,PDE1B,ADCY1,ADCY6,ADCY8, PDE11A,PKIG
FGF Signaling	1.80	0.084	0.707	FGFR3,FGF4,FGF14,CREBBP,CREB3L3,RAC1,PIK3R6,FGF5
α-Adrenergic Signaling	1.75	0.083	1	ADCY9,GNAS,ADCY1,ADCY6,PRKCE,ADCY8,ADRA1A,PRKCZ
RAR Activation	1.74	0.066	NaN	ADCY9,BMP2,CREBBP,ADCY1,NR2F2,RAC1,ADCY6,MAPK9,PRKCE,ADCY8,PRKCZ,SMARCD3,CRABP1
NGF Signaling	1.52	0.071	0.333	FGFR3,MAP3K11,CREBBP,CREB3L3,RAC1,PIK3R6,MAPK9,SMPD1,PRKCZ
Neuroinflammation Signaling	1.50	0.055	1.387	GABRQ,GDNF,GABRA4,CREBBP,CREB3L3,MAPK9,BMPR2,IDE,IRF3,NFATC1,FGFR3,SLC6A11,GAD2,PLA2G4B,KCNJ9,PIK3R6,WNT1,SLC6A12
Semaphorin Signaling in Neurons	1.47	0.094	NaN	RHOC,RHOD,RAC1,LIMK2,LIMK1
Reelin Signaling in Neurons	1.42	0.076	NaN	FGFR3,YES1,LCK,ITGA3,MAP3K11,PIK3R6,MAPK9
Ephrin Receptor Signaling	1.37	0.062	0	EPHA7,EPHA10,ITGA3,GNAS,EPHA1,CREBBP,CREB3L3,RAC1,LIMK2,GNAL, LIMK1

¹IPA software (analyzed on 3 April 2018) was utilized to determine canonical pathways significantly altered in PNS bull calves. The analysis was conducted using differentially methylated (a ≥10% difference between treatment groups) CpG sites located within promoter regions (P ≤ 0.05). Of the 113 total canonical pathways significantly altered in PNS calves (Littlejohn et al., 2018), 28 were related to neural function, behavior, and/or stress response.

Table 2.

Hypomethylated CpG sites located within promoter regions of genes associated with the Behavior function term (IPA software) in PNS compared with Control bull calves^1,2

BTA	Gene	Location, Mb	Methyl Diff	P
13	growth hormone secretagogue receptor; GHSR	95.8	−0.22	0.01832
13	glutamate ionotropic receptor kainate type subunit 1; GRIK1	5.4	−0.12	0.01293
2	G protein-coupled receptor 39; GPR39	65.5	−0.27	0.00344
2	phosphodiesterase 11A; PDE11A	18.8	−0.25	0.00409
2	microtubule-associated protein 2; MAP2	97.8	−0.19	0.04946
2	selenoprotein N, 1; SEPN1	127.9	−0.18	0.02758
23	protein tyrosine phosphatase receptor type N; PTPRN	108.0	−0.17	0.04341
23	5-hydroxytryptamine receptor 6; HTR6	133.6	−0.16	0.03218
23	T-box brain transcription factor 1; TBR1	35.1	−0.15	0.00925
23	inhibin subunit beta B; INHBB	72.5	−0.13	0.00929
4	adenylate cyclase 1; ADCY1	77.0	−0.12	0.02256
43	leptin; LEP	93.3	−0.15	0.01131
54	cyclin dependent kinase inhibitor 1B; CDKN1B	97.6	−0.24	0.02051
53	potassium voltage-gated channel subfamily H member 3; KCNH3	30.5	−0.16	0.03118
53,4	phosphodiesterase 1B; PDE1B	25.7	−0.14	0.03023
53,4	vitamin D receptor; VDR	32.5	−0.14	0.04952
63	dopamine receptor D5; DRD5	107.6	−0.15	0.00648
7	PTTG1 regulator of sister chromatid separation, securing; PTTG1	74.1	−0.13	0.00891
73	histone deacetylase 3; HDAC3	54.4	−0.11	0.03307
73,4	POU class 4 homeobox 3; POU4F3	59.5	−0.11	0.03876
84	tenascin C; TNC	106.1	−0.19	0.02080
83	adrenoceptor alpha 1A; ADRA1A	75.4	−0.20	0.04682
9	parkin RBR E3 ubiquitin protein ligase; PARK2	99.6	−0.23	0.01609
9	opioid receptor mu 1; OPRM1	92.2	−0.20	0.00125
104	dopamine receptor D1; DRD1	5.7	−0.19	0.00030
103,4	cysteine dioxygenase type 1; CDO1	4.6	−0.11	0.03595
11	fatty acid binding protein 1; FABP1	47.8	−0.29	0.03130
114	solute carrier family 1 member 4; SLC1A4	63.3	−0.21	0.01018
113	proopiomelanocortin; POMC	74.1	−0.18	0.01736
113	protein kinase C epsilon; PRKCE	27.9	−0.16	0.03505
113,4	calcium voltage-gated channel subunit alpha1 B; CACNA1B	105.4	−0.11	0.01762
123	protocadherin 8; PCDH8	10.9	−0.17	0.04747
134	DnaJ heat shock protein family (Hsp40) member C5; DNAJC5	54.4	−0.12	0.04970
133	glutamic acid decarboxylase 2; GAD2	26.9	−0.12	0.01438
133,4	GNAS complex locus; GNAS	58.0	−0.35	0.01470
133,4	adrenoceptor alpha 1D; ADRA1D	51.4	−0.11	0.03773
14	adenylate cyclase 8; ADCY8	10.8	−0.21	0.03585
144	snail family transcriptional repressor 2; SNAI2	21.6	−0.11	0.03954
143,4	preproenkephalin; PENK	25.2	−0.14	0.04169
153	apelin receptor; APLNR	81.7	−0.23	0.00044
153	adrenomedullin; ADM	42.9	−0.14	0.00546
153	paired box 6; PAX6	63.4	−0.11	0.03895
153,4	potassium voltage-gated channel subfamily C member 1; KCNC1	35.3	−0.18	0.03103
163	hes family bHLH transcription factor 5; HES5	51.6	−0.15	0.00986
163,4	ERBB receptor feedback inhibitor 1; ERRFI1	46.2	−0.13	0.03823
174	neuropeptide Y receptor Y2; NPY2R	2.1	−0.12	0.03787
173	catechol-O-methyltransferase; COMT	74.9	−0.13	0.03446
193	complement C1q like 1; C1QL1	45.3	−0.21	0.00931
193	regulator of G protein signaling 9; RGS9	62.6	−0.21	0.00880
203	T cell leukemia homeobox 3; TLX3	3.1	−0.22	0.00807
203	CART prepropeptide; CARTPT	9.8	−0.14	0.03470
203,4	glial cell derived neurotrophic factor; GDNF	36.6	−0.13	0.02902
21	creatine kinase, mitochondrial 1B; CKMT1B	55.9	−0.14	0.03775
214	cholinergic receptor nicotinic alpha 7 subunit; CHRNA7	30.2	−0.29	0.00054
213	ST8 alpha-N-acetyl-neuraminide alpha-2,8-sialyltransferase 2; ST8SIA2	14.9	−0.17	0.02089
23	protein phosphatase 2 regulatory subunit B’delta; PPP2R5D	16.6	−0.12	0.00234
25	LIM domain kinase 1; LIMK1	33.8	−0.30	0.01825
254	Rac family small GTPase 1; RAC1	38.8	−0.25	0.02451
26	prolactin releasing hormone receptor; PRLHR	39.2	−0.22	0.00009
26	fucose mutarotase; FUOM	25.9	−0.17	0.01690
263	protein kinase cGMP-dependent 1; PRKG1	8.3	−0.13	0.03617
294	potassium voltage-gated channel subfamily Q member 1; KCNQ1	49.7	−0.15	0.03606
294	tyrosine hydroxylase; TH	50.0	−0.13	0.01405

¹Methylation difference (Methyl Diff): defined as the average PNS methylation ratio minus the average Control methylation ratio at a site.

²Average read coverage for the listed hypomethylated sites was 13.0 and 11.3 for Control and PNS bull calves, respectively.

³DNA methylation was located within a CpG island.

⁴DNA methylation was exclusively located within the promoter region.

Table 3.

Hypermethylated CpG sites located within promoter regions of genes associated with the Behavior function term (IPA software) in PNS compared with Control bull calves^1,2

BTA	Gene	Location, Mb	Methyl Diff	P
1	somatostatin; SST	80.3	0.13	0.03745
2	lin-28 homolog A; LIN28A	127.3	0.12	0.02204
2	neurogenic differentiation 1; NEUROD1	15.0	0.15	0.01520
33	prostaglandin E receptor 3; PTGER3	74.5	0.19	0.04315
4	SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily d, member 3SMARCD3	114.6	0.12	0.02467
4	5-hydroxytryptamine receptor 5A; HTR5A	117.7	0.20	0.03440
74	mannosidase alpha class 2B member 1; MAN2B1	14.0	0.25	0.00233
73	early growth response 1; EGR1	51.4	0.11	0.02447
93	neuromedin B receptor; NMBR	80.6	0.12	0.03940
93	protein phosphatase 1 regulatory inhibitor subunit 14C; PPP1R14C	88.4	0.12	0.03313
11	receptor accessory protein 1; REEP1	48.4	0.12	0.03986
11	RIC8 guanine nucleotide exchange factor A; RIC8A	107.2	0.14	0.00631
113	leucine rich repeat transmembrane neuronal 1; LRRTM1	55.1	0.14	0.04496
113	RAB11 family interacting protein 5; RAB11FIP5	11.3	0.15	0.03192
123	protocadherin 17; PCDH17	5.2	0.34	0.00044
133,4	solute carrier family 12, member 5; SLC12A5	75.5	0.13	0.04118
133,42	transcription factor 15; TCF15	61.0	0.24	0.00551
143,4	opioid receptor kappa 1; OPRK1	23.4	0.30	0.01213
15	sphingomyelin phosphodiesterase 1; SMPD1	47.3	0.12	0.01855
154	midkine; MDK	77.2	0.20	0.02052
153	cholinergic receptor, muscarinic 4; CHRM4	77.3	0.13	0.00031
153	progesterone receptor; PGR	8.1	0.13	0.02459
16	calcium voltage-gated channel subunit alpha1 E; CACNA1E	64.1	0.11	0.03893
16	renin; REN	1.8	0.11	0.03689
164	DNA fragmentation factor subunit alpha; DFFA	44.0	0.24	0.02651
163	protein kinase C zeta; PRKCZ	51.9	0.11	0.03226
163	disheveled segment polarity protein 1; DVL1	52.4	0.13	0.00297
163	natriuretic peptide B; NPPB	42.7	0.14	0.04007
173	chemerin chemokine-like receptor 1; CMKLR1	66.8	0.10	0.02338
173	nuclear receptor subfamily 3 group C member 2; NR3C2	9.7	0.16	0.02930
18	gigaxonin; GAN	8.0	0.11	0.01103
184	junctophilin 3; JPH3	13.2	0.10	0.03613
19	integrin subunit alpha 3; ITGA3	37.2	0.12	0.00009
193	peptide YY; PYY	44.4	0.17	0.02333
213,4	creatine kinase B; CKB	69.8	0.15	0.00779
223,4	GATA binding protein 2; GATA2	60.0	0.11	0.04886
233	N-ribosyldihydronicotinamide:quinone reductase 2; NQO2	50.5	0.42	0.02622
233,4	transcription factor AP-2 alpha; TFAP2A	45.5	0.17	0.00496
243	G protein subunit alpha L; GNAL	43.1	0.11	0.02352
243	adenylate cyclase activating polypeptide 1; ADCYAP1	36.1	0.14	0.03890
243	neuropilin and tolloid like 1; NETO1	5.4	0.21	0.02225
253,4	CREB binding protein; CREBBP	31.7	0.13	0.03924
294	insulin like growth factor 2; IGF2	50.0	0.17	0.00315
294	cholinergic receptor muscarinic 1; CHRM1	41.9	0.20	0.04241
293	protein tyrosine phosphatase non-receptor type 5; PTPN5	26.2	0.12	0.00280

¹Methylation difference (Methyl Diff): defined as the average PNS methylation ratio minus the average Control methylation ratio at a site.

²Average read coverage for the listed hypermethylated sites was 11.4 and 10.1 for Control and PNS bull calves, respectively.

³DNA methylation was located within a CpG island.

⁴DNA methylation was exclusively located within the promoter region.

Opioid Signaling

The Opioid Signaling pathway was the most significant canonical pathway altered in PNS compared with Control bull calves (Table 1). Two genes encoding for opiate agonists (endogenous opioids) contained a hypomethylated CpG site within the promoter region, including the proenkephalin (PENK; MD = −0.14) gene encoding for Meta-enkephalin and Leu-enkephalin and the pro-opiomelanocortin (POMC; MD = −0.18) gene encoding for β-endorphin (Figure 1; Table 2). Hypomethylation of these sites suggests potential increased production of these opiate agonists in PNS compared with Control bulls. Stress has been reported to increase production of opiate agonists (Bruchas et al., 2010). Because PNS calves from the larger population from which calves in this study were derived were found to exhibit increased circulating concentrations of cortisol (Littlejohn et al., 2016), an indicator of stress, increased production of opiate agonists might be expected. Opiate agonists, however, are also known to play a role in overriding the physiological response to a stressor, thereby, acting as a mechanism to aid in recovery from the stress response (Bali et al., 2015; Valentino and Van Bockstaele, 2015). For example, in times of stress POMC production increases, giving rise to both adrenocorticotropic hormone (ACTH) and β-endorphin (Mains et al., 1977; Cawley et al., 2016). Adrenocorticotropic hormone stimulates adrenocortical secretion of cortisol, the product of an activated HPA axis, and β-endorphin binds to mu-opioid receptors in the intercalated nucleus of the amygdala to prevent over-activation of the HPA axis (Nakagawasai et al., 1999). Furthermore, enkephalins have long been known to play a role in stress adaptation (Van Loon et al., 1990), and enkephalin knock out models result in increased anxiety-like behavior in rodents (Bérubé et al., 2014). Additionally, the Opiod Receptor Mu 1 (OPRM1) gene contained a CpG site within the promoter region that was hypomethylated (MD = −0.20; Figure 1; Table 2) in PNS compared with Control bulls, suggesting increased gene expression. Mu-opioid receptor (MOR) pathways have been associated with the stress response and the HPA axis (Chong et al., 2006). It is also well established that MOR pathways are involved in pain modulation. For example, MOR can be found on GABA interneurons that activate pain-inhibitory neurons (Al-Hasani and Bruchas, 2011). Also, MOR signaling activates dopamine signaling pathways, thereby acting as mood enhancers (Al-Hasani and Bruchas, 2011; Saigusa et al., 2017). This may play a role in the predicted activation of both the Dopamine Receptor Signaling and Dopamine-DARPP32 Feedback in cAMP Signaling pathways (Table 1). The Opioid Receptor Kappa 1 (OPRK1) gene contained a CpG site within the promoter region that was hypermethylated (MD = 0.30) in PNS compared with Control bulls, suggesting decreased gene expression (Table 3). In humans, the kappa opioid receptor (KOR) pathway is involved in pro-addictive tendencies in times of stress (Bruchas et al., 2010). Disruption of the KOR pathway with KOR antagonists has been reported to suppress stress responsiveness and even serve as treatment for psychiatric disorders (Van’t Veer et al., 2012; Van’t Veer and Carlezon, 2013). Furthermore, activation of KOR by opiate agonists, such as dynorphin, has been reported to have an inhibitory influence on dopamine pathways (Spanagel et al., 1994; Chefer et al., 2005). Therefore, the predicted repression of the KOR-mediated opioid pathway coincides with activation of both the Dopamine Receptor Signaling and Dopamine-DARPP32 Feedback in cAMP Signaling pathways (Table 1). With regard to the opioid signaling pathway, IPA software predicted increased Membrane Hyperpolarization and Receptor Internalization, but decreased Neurotransmission, Membrane Potential, Synaptic Transmission, and Analgesic Tolerance in PNS compared with Control bulls (Figure 1). Alterations to various components of the opioid signaling pathway have been reported to be influenced by prenatal stress (Insel et al., 1990; Poltyrev and Weinstock, 1997; Vey et al., 2016). For example, rats whose dams were exposed to random instances of noise and light stress during gestation exhibited decreased locomotion after naloxone administration compared with Controls (Poltyrev and Weinstock, 1997). Furthermore, rats whose dams were exposed to various chronic stressors during gestation exhibited increased glucocorticoid concentrations, anxiety phenotypes, and preference for morphine after withdrawal (Vey et al., 2016). Rats whose dams experienced a combination of heat and restraint stress during gestation exhibited differential alterations to binding of opiate agonists with MORs in various regions of the brain (Insel et al., 1990). It is clear that the opioid signaling pathway is sensitive to programming during prenatal development, and that complex interactions with multiple signaling pathways existed in PNS bulls regulating behavior and stress response.

Figure 1.

Canonical pathway (Modified from IPA) for Opioid Signaling [−log(P-value) = 8.48; Z-score = 0.392].

Glucocorticoid Signaling

The Corticotropin Releasing Hormone (CRH) Signaling pathway was predicted to be activated in PNS compared with Control bull calves (Table 1; Figure 2). The IPA software predicted increased corticosteroid synthesis, as mediated primarily by POMC (MD = −0.18) in the CRH pathway (Figure 2A; Table 2). As described in the previous section, POMC gives rise to the prohormone, ACTH, which acts on the adrenal cortex to cause the production of cortisol (Charmandarie et al., 2005). Therefore, hypomethylation of a CpG site within the promoter region of POMC is predictive of increased cortisol production (Figure 2A). Cortisol is essential in mediating the stress response (Butcher and Lord, 2004) by increasing lipolysis, amino acid availability, and circulating glucose concentrations to be utilized for energy expenditure during the stress response (Gerrard and Grant, 2003). Increased cortisol has also been associated with an elevated temperament score in cattle (Curley et al., 2006). Furthermore, single-nucleotide polymorphisms (SNPs) within the POMC gene have been associated with temperament in cattle (Garza-Brenner et al., 2017). Predicted increased glucocorticoid production concurs with increased circulating concentrations of cortisol and more excitable temperaments observed in the larger population of calves from which bull calves in this study were derived (Littlejohn et al., 2016). In addition to POMC, differential methylation of CpG sites within the promoter regions of the following genes (Figure 2B) directly or indirectly predicted increased glucocorticoid concentrations: adrenomedullin (ADM; MD = −0.14), CART prepropeptide (CARTPT; MD = −0.14), potassium two pore domain channel subfamily K member 9 (KCNK9; MD = 0.10), lecithin-cholesterol acyltransferase (LCAT; MD = −0.14), protein kinase C epsilon (PRKCE; MD = −0.16), scavenger receptor class B member 1 (SCARB1; MD = −0.11), secretogranin V (SCG5; MD = 0.12), somatostatin (SST; MD = 0.13), and tyrosine hydroxylase (TH; MD = −0.13). In addition to these genes, it is important to discuss receptors with which glucocorticoids interact. Perroud et al. (2014) reported differential methylation of the glucocorticoid receptor (GR) gene (NR3C1) and mineralocorticoid receptor (MR) gene (NR3C2), along with decreased cortisol concentrations, decreased GR, and increased MR in children whose mothers were affected by the Tutsi genocide during gestation. Although PNS bulls did not exhibit differences in NR3C1 methylation, they did exhibit hypermethylation (MD = 0.16) of a CpG site within the promoter region of NR3C2 compared with Control bulls, suggesting decreased gene expression (Table 3). Cortisol is known to bind both the GR and the MR but generally has a higher affinity for the MR (Rupprecht et al., 1993). It is important to note that GR and MR are differentially expressed in different neuron types (Koning et al., 2019). In general, GR can be found throughout the brain, but MR is primarily found in limbic system tissues (Seckl et al., 1991; Patel et al., 2000). Emerging evidence suggests independent roles of the MR in regulating the stress response (Heegde et al., 2015), such as a direct influence on neurotransmission in the hippocampus and amygdala in times of stress (Karst et al., 2005). Perhaps increased cortisol played a role in the potential suppression of NR3C2 to alter cellular sensitivity to elevations in circulating cortisol concentrations. It is important to acknowledge that associations of various prenatal and early life stressors with alterations to components of the HPA axis are arguably among the most noted in the scientific literature and found across species (Lay et al., 1997a,b; Haussmann et al., 2000; Meaney et al., 2001; Roussel et al., 2004; Kertes et al., 2016; Spencer and Deak, 2017). Cord blood, placental tissue, and maternal blood from pregnancies in which mothers experienced chronic stressors or war trauma during gestation exhibited differential methylation of various CpG sites within the CRH, corticotropin releasing hormone binding protein (CRHBP), NR3C1, and FKBP prolyl isomerase 5 (FKBP5) genes (Paquette et al., 2014; Kertes et al., 2016). Guinea pigs whose dams were exposed to strobe light stressors during gestation exhibited increased circulating concentrations of cortisol, reduced GR mRNA in the hippocampus, and increased POMC mRNA in the pituitary (Kapoor et al., 2008). Pigs whose dams were restrained by snout snare daily for 5 min between 84 and 112 d of gestation exhibited elevated circulating concentrations of cortisol relative to Control pigs (Collier et al., 2011). Blue foxes whose dams were exposed to 1 min of daily handling stress during the last trimester of pregnancy exhibited heightened signs of distress, such as crawling, fighting, yelping, and aggression, when handled by a human evaluator (Braastad et al., 1998). Calves whose dams were transported for 2 h at 60, 80, 100, 120, and 140 d of gestation exhibited increased circulating concentrations of cortisol for a longer duration of time in response to restraint stress and cleared a bolus dose of cortisol at a slower rate compared with Control calves (Lay et al., 1997b). It is apparent that the CRH signaling pathway is sensitive to programming during prenatal development, and it is important to note that observed differences in methylation of DNA within genes associated with the HPA axis directly relate to the most prominent phenotypic differences observed in the larger population from which bulls in this study were derived, including increased temperament scores and circulating concentrations of cortisol (Littlejohn et al., 2016).

Figure 2.

(A) Canonical pathway (Modified from IPA) for Corticotropin Releasing Hormone Signaling [−log(P-value) = 3.74; Z-score = 1.29]. (B) Function group network (Modified from IPA) for Concentration of Corticosterone (P < 0.0001; Z-score = 2.51).

Dopamine Signaling

The Dopamine Receptor Signaling and Dopamine-DARPP32 Feedback in cAMP Signaling pathways were activated in PNS bull calves (Table 1). Specifically, a CpG site within the promoter region of the tyrosine hydroxylase (TH) gene was hypomethylated (MD = −0.13; Table 2; Figure 3A), suggesting increased gene expression. Because the TH enzyme is responsible for the conversion of a precursor hormone in the dopamine production cascade, the predicted increased expression of TH would result in the prediction of increased dopamine production and release from the presynaptic neuron (Figure 3). At the level of the postsynaptic neuron, CpG sites within the promoter region of the dopamine receptor D1 (DRD1; MD = −0.19) and dopamine receptor D5 (DRD5; MD = −0.15) genes were hypomethylated, suggesting increased gene expression (Table 2; Figure 3A). The D1-like dopamine receptors, DRD1 and DRD5, activate adenylate cyclase; alternatively, the D2-like dopamine receptors, dopamine receptor D2 (DRD2), dopamine receptor D3 (DRD3), and dopamine receptor D4 (DRD4), inhibit adenylate cyclase (Neve et al., 2004; Xue et al., 2018). Predicted increased dopamine release from the presynaptic neuron in conjunction with predicted increased DRD1 and DRD5 on the postsynaptic neuron suggests increased activation of the postsynaptic neuron as mediated by the adenylate cyclase cascade and ultimately phosphorylation of DARPP-32, resulting in inhibition of protein phosphatase-1 (PP1) and activation of transcription in the nucleus (Scheggi et al., 2018; Figure 3A). Furthermore, SNPs within the DRD1 gene have been associated with behavioral disorders in humans (Andreou et al., 2016), and SNPs within the DRD2 and DRD3 genes have been associated with temperament in cattle (Garza-Brenner et al., 2017). In addition to the aforementioned genes, differential methylation of CpG sites within the promoter regions of the following genes (Figure 3B) directly or indirectly predicted increased dopamine concentrations: adrenoceptor alpha 1D (ADRA1D; MD = −0.11), CARTPT (MD = −0.14), cholinergic receptor muscarinic 1 (CHRM1; MD = 0.20), cholinergic receptor muscarinic 4 (CHRM4; MD = 0.13), glial cell line derived neurotrophic factor (GDNF; MD = −0.13), leptin (LEP; MD = −0.15), OPRK1 (MD = 0.30), protein tyrosine phosphatase receptor type N (PTPRN; MD = −0.17), and TH (MD = −0.13). Additionally, a CpG site within the catechol-O-methyltransferase (COMT) gene was hypomethylated (MD = −0.13; Table 2; Figure 3A) in PNS compared with Control bulls, suggesting increased gene expression. The COMT enzyme is responsible for degrading catecholamines (Wilkinson and Brown, 2015); therefore, increased expression of COMT would potentially result in increased degradation of dopamine, resulting in increased circulating metabolites of dopamine and other catecholamines. For example, dopamine metabolites 3-methyltyramine and homovanillic acid were predicted to be increased by IPA software (Figure 3A). Such metabolites have been associated with psychiatric and behavioral disorders and have long been utilized as clinical indicators for disorders such as schizophrenia and depression (Bowers et al., 1969; Ogawa and Kunugi, 2019). Although various components of the dopamine signaling pathway have been associated with behavior in rodents, humans, and cattle (Ray et al., 2006; Keltikangas-Järvinen and Salo, 2009; Garza-Brenner et al., 2017), dopaminergic regulation of behavior is complex and multifaceted due to the number and diversity of brain regions innervated by dopaminergic neurons (Cools, 2008). An increasing amount of literature suggests a complex sensitivity of the dopamine signaling pathway to programming in utero. Rats whose dams were exposed to noise and light stress during gestation spent less time in open areas of a maze and exhibited altered dopamine activity in left compared with right regions of the prefrontal cortex, caudate nucleus, and nucleus accumbens (Fride and Weinstock, 1988). Contrary to predictions in PNS bulls, young rats that were separated from their mothers daily for 6-h periods during the first 2 wk of life had decreased DRD1 gene expression in the nucleus accumbens (Zhu et al., 2010). Furthermore, young rats that were separated from their mothers for 3-h periods had decreased mid-brain tyrosine hydroxylase-immunoreactive dopaminergic neurons as juveniles (15 d of age) but increased numbers as adolescents (35 d of age) and adults (70 d of age; Chocyk et al., 2011). Adult rats whose dams were exposed to dexamethasone on days 18 and 19 of gestation had decreased dopamine and increased DRD2 gene expression in the nucleus accumbens (Rodrigues et al., 2012). Additionally, an SNP within the COMT gene was associated with emotional and behavioral issues in children whose mothers experienced substantial stress during gestion (Thompson et al., 2012). The present study reaffirms various reports of altered components of the dopamine signaling pathway in prenatally stressed mammals and suggests activation of multiple components of the pre- and post-synaptic dopamine signaling pathway.

Figure 3.

(A) Canonical pathway (Modified from IPA) for Dopamine Receptor Signaling [−log(P-value) = 2.97; Z-score = 1.89]. (B) Function group network (Modified from IPA) for Concentration of Dopamine (P < 0.0001; Z-score = 0.79).

Serotonin Signaling

The Serotonin Receptor Signaling pathway was predicted by IPA software to be altered in PNS compared with Control bull calves (Table 1; Figure 4). Serotonin (5-hydroxytryptamine, 5-HT) is a monoamine, specifically an indoleamine, neurotransmitter that is derived from dietary tryptophan and is involved in many physiological processes, including neurobehavioral control (Wilkinson and Brown, 2015; Sahu et al., 2018). Two serotonin receptor genes contained differentially methylated CpG sites. The 5-hydroxytryptamine receptor 5A (HTR5A) gene contained a CpG site within the promoter region that was hypermethylated (MD = 0.20) in PNS compared with Control bulls, suggesting decreased gene expression (Table 3). When serotonin binds to HTR5A in the postsynaptic neuron, adenylate cyclase is inhibited (Francken et al., 1998). This relationship suggests less inhibition of adenylate cyclase by HTR5A (Figure 4). Not only was adenylate cyclase predicted to be indirectly increased by suppressed HTR5A, but it was also predicted to be directly increased by differential methylation of the 5-hydroxytryptamine receptor 6 (HTR6) gene. Specifically, the HTR6 gene contained a CpG site that was hypomethylated (MD = −0.16) in PNS compared with Control bulls, suggesting increased gene expression (Table 2). In the postsynaptic neuron, HTR6 is known to increase activation of adenylate cyclase (Unsworth and Molinoff, 1994); therefore, increased activity of HTR6 predicts potential increased activation of adenylate cyclase (Figure 4). Increased adenylate cyclase is known to regulate cell activity by increasing cAMP needed for activation of the postsynaptic neuron. Serotonin signaling is well known to interact with dopamine and GABA signaling pathways (Fischer and Markus Ullsperger, 2017). Like dopamine and GABA signaling pathways, components of the serotonin signaling pathway have been associated with behavior in humans, nonhuman primates, and cattle (Raleigh et al., 1980; Knutson et al., 1998; Garza-Brenner et al., 2017). Dysfunctions in serotonin signaling have been associated with disorders such as anxiety (Marcinkiewcz et al., 2016), aggression (Olivier, 2004), autism spectrum disorder (Chugani, 2002), and schizophrenia (Hashimoto et al., 1991). Single-nucleotide polymorphisms within the genes 5-hydroxytryptamine receptor 2A (HTR2A) and solute carrier family 18 member A2 (SLC18A2; a monoamine transporter) have been associated with temperament in cattle (Garza-Brenner et al., 2017). Prenatal stress has been associated with alterations in serotonin receptor binding, serotonin synthesis, and associated behavioral alterations (Peters, 1986a; Van den Hove et al., 2006; Miyagawa et al., 2011). Newborns whose mothers experienced anxiety or depression during pregnancy and took serotonin reuptake inhibitor medication exhibited differential methylation of a CpG site within the SLC6A4 gene (encodes for a serotonin transporter) in cord blood (Non et al., 2014). Furthermore, mice whose dams were exposed to daily 6-h periods of restraint between 5.5 and 17.5 d of gestation exhibited increased anxiety-like behavior and an increase in cells within the raphe nuclei (Miyagawa et al., 2011), the region of the brain that contains the majority of serotonin neuron cell bodies which reach throughout diverse regions of the brain (Wilkinson and Brown, 2015). Rats whose dams experienced noise and light stress during gestation exhibited decreased binding of the receptor 5-HT1A in the ventral hippocampus (Van den Hove et al., 2006). Like predicted alterations to PNS bulls, rats whose dams were exposed to a prenatal stressor (daily crowding and saline injections) were more sensitive to serotonin administration, exhibiting increased behavioral and physiological responses compared with Control rats (Peters, 1986b). It has been established that the serotonin signaling pathway is subject to changes induced by the prenatal environment; specific results from this study predict increased neuron sensitivity to serotonin.

Figure 4.

Canonical pathway (Modified from IPA) for Serotonin Signaling [−log(P-value) = 2.04; Z-score = NaN].

GABA Signaling

The GABA Receptor Signaling pathway was altered in PNS compared with Control bull calves (Table 1). Gamma-aminobutyric acid (GABA), the most prevalent neurotransmitter in the brain, is an inhibitory amino acid neurotransmitter, spanning diverse brain regions (Young and Chu, 1990). The glutamic acid decarboxylase (GAD) enzyme converts the precursor glutamic acid (glutamate) to GABA (Wilkinson and Brown, 2015; Figure 5). The GAD gene family, specifically the glutamic acid decarboxylase 2 (GAD2) gene, contained a CpG site within the promoter region that was hypomethylated (MD = −0.12; Table 2; Figure 5) in PNS compared with Control bulls, suggesting increased gene expression. As a result, IPA predicted increased synthesis of GABA within the presynaptic neuron (Figure 5). It is established that GABA acts on GABAA receptors on the postsynaptic neuron, which consist of 5 protein molecules to form an ion channel that opens upon interaction with GABA to inhibit electrophysiological activity (Wilkinson and Brown, 2015). In addition to GABA, tranquilizers, anesthetics, barbiturates, and alcohol elicit their effects by acting on GABAA, thereby suppressing anxious behaviors (Wilkinson and Brown, 2015). The γ-aminobutyric acid receptor subunit alpha (GABR-A) gene family encodes for receptors within the GABAA family. Within the GABR-A family, the GABRA4 gene contained a CpG site within the promoter region that was hypermethylated (MD = 0.15), and the GABRQ gene contained two CpG sites within the promoter region that were hypermethylated (MD = 0.13 and 0.10), suggesting suppressed gene expression of GABR-A receptors (Figure 5). Suppression of GABR-A predicts suppression of the inhibitory effects of GABA on the postsynaptic neuron, which might have played a role in the phenotype of increased temperament scores in the larger population of calves in which PNS calves in this study were derived (Littlejohn et al., 2016). Furthermore, due to the aforementioned differences in methylation (hypermethylation) of the MR gene in PNS bulls, it is interesting to note that MR and GR have been found to be differentially expressed in GABA-ergic (inhibitory) and glutamatergic (excitatory) neurons in the temporal cortex of humans (Koning et al., 2019). Specifically, intermediate to low expression of MR was observed in both GABA-ergic and glutamatergic neurons and high expression of GR was observed in glutamatergic neurons (Koning et al., 2019). The GABA signaling pathway is involved in the physiological control of cognition, motor learning, fear, and behavior (Stagg et al., 2011; Earnheart et al., 2007; Paredes and Aġmo, 1992; Stratton et al., 2014). Dysfunctions in GABA signaling have been associated with neural and neurobehavioral disorders, such as depression (Stratton et al., 2014), anxiety (Berger et al., 2002; Stratton et al., 2014), and Attention-Deficit/Hyperactivity Disorder (Bollman et al., 2015). Previous reports suggest an influence of prenatal stress on various components of the GABA signaling pathway (Ehrlich et al., 2015; Lussier and Stevens, 2016; Vangeel et al., 2017). Children born to mothers with high anxiety levels during gestation exhibited genome-wide differences in DNA methylation, including GABBR1 (GABA-B receptor subunit 1 gene; Vangeel et al., 2017). Mice whose dams underwent restraint and light stress from embryonic day 12 until parturition exhibited increased anxiety-like characteristics and developmental delays in GABAergic cell counts (Lussier and Stevens, 2016). Furthermore, rats whose dams were exposed to various gestational stressors between days 9 and 20 of gestation exhibited anxious behaviors and altered gene expression of the GABAergic regulators, KCC2 and NKCC1 (chloride transporters; Ehrlich et al., 2015). Although differential methylation of GAD2 predicted increased synthesis of GABA in PNS bulls, differential methylation of 3 CpG sites within GABR-A receptors suggests potential suppression of the inhibitory effects of GABA.

Figure 5.

Canonical pathway (Modified from IPA) for GABA Receptor Signaling [−log(P-value) = 5.07; Z-score = NaN].

Other neural, neuroendocrine, and endocrine systems

Beyond the genes, signaling pathways, and function terms discussed in the preceding paragraphs, many other systems related to neural, neuroendocrine, and endocrine systems controlling behavior and stress response phenotypes were predicted to be altered in PNS compared with Control bulls due to differential methylation (Tables 1–3). For example, a CpG site within the promoter region of the Secretogranin V (SCG5) gene was hypermethylated (MD = 0.12) in PNS compared with Control bulls, suggesting decreased gene expression. Likewise, Cao-Lei et al. (2014) reported multiple CpG sites within the promoter region of SCG5 to be hypermethylated in children whose mothers experienced the Quebec ice storm of 1998. The SCG5 gene encodes for a chaperone protein that is known to transport and activate prohormone convertase 2 (PC2), an important enzyme involved in the synthesis of endogenous opioids. Furthermore, PC2 knockout mice are known to exhibit increased MOR in the brain (Lutfy et al., 2016). Also, the Retinoic Acid Receptor (RAR) activation pathway was predicted to be altered in PNS compared with Control bulls (Table 1). Specifically, the cellular retinoic acid binding protein 1 (CRABP1) gene contained a CpG site that was hypermethylated (MD = 0.10) in PNS compared with Control bulls, suggesting decreased gene expression. Specifically, CRABP1 is known to increase metabolism and degradation of retinoic acid; therefore, increased methylation of CRABP1 might suggest decreased retinoic acid (Noy, 2000). Retinoic acid is involved in nervous system development (Maden and Holder, 1991) and has been associated with behavioral alterations, such as hyperactivity in rodents (Cai et al., 2010). Additionally, the Reelin Signaling in Neurons pathway was predicted to be altered by differential methylation in PNS compared with Control bulls (Table 1). Likewise, rats whose dams underwent daily restraint stress for 2-h periods between embryonic days 11 and 20 exhibited lower numbers of reelin-positive neurons, lower reelin gene expression, and altered DNA methylation within the promoter region of the reelin gene (Palacios-García et al., 2015). The reelin gene is expressed during cortical development in Cajal-Retzius cells and is involved in cortical lamination and synaptic maturation (Palacios-García et al., 2015). In general, IPA software predicted activation of the following function terms: Quantity of Neurons (P < 0.001; Z-score = 0.861); Quantity of Nervous Tissue (P < 0.001; Z-score = 0.954); Migration of Neurons (P < 0.001; Z-score = 1.530); Proliferation of Neuronal Cells (P < 0.001; Z-score = 1.252); Outgrowth of Neurons (P < 0.001; Z-score = 0.843); Development of Neurons (P < 0.001; Z-score = 1.783); and Quantity of Neurotransmitter (P < 0.001; Z-score = 0.922). Prenatal and early life stressors have been associated with diverse differences in neural function and physiology. Rats between 80 and 120 d of age whose dams were randomly handled in a new environment and received daily saline injections during gestation exhibited increased volume, neuronal density, number of neurons, and number of glial cells in the lateral nucleus of the amygdala (Salm et al., 2004). Contrasting results were reported in rats at 25 d of age whose dams were also randomly handled in a new environment and received daily saline injections during gestation; prenatally stressed rats exhibited decreased volume, number of neurons, and number of glial cells in the basolateral, central, and lateral nuclei of the amygdala (Kraszpulski et al., 2006). Rats from the same study at 45 d of age exhibited no difference in volume, number of neurons, or number of glial cells (Kraszpulski et al., 2006). Prenatal stress has been associated with decreased neurogenesis in the dentate gyrus portion of the hippocampus (Lemaire et al., 2000; Coe et al., 2003), decreased hippocampal volume (Schmitz et al., 2002; Coe et al., 2003), decreased hippocampal weight (Szuran et al., 1994), and decreased number of granule cells (Lemaire et al., 2000) in the hippocampus. Increasing amounts of literature are revealing differential methylation of DNA as a regulator of neural differences in prenatally stressed individuals. For example, Provençal et al. (2019) reported altered in vitro neurogenesis in response to dexamethasone administration that was associated with altered DNA methylation and transcript abundance. Furthermore, DNA methylation enrichment analyses revealed alterations in Neurogenesis and Neuronal Differentiation function terms (Provençal et al., 2019). A DNA methylation enrichment analyses in mice whose dams were exposed to a viral challenge on day 9 or 17 of gestation revealed Neuronal Differentiation to be the most enriched gene ontology term, significant subterms of which included Gamma-aminobutyric Acidergic Differentiation, Central Nervous System Differentiation, Noradrenergic System Differentiation, and Dopamine Differentiation (Richetto et al., 2017). In the present study, prenatal transportation stress was associated with vast differences in methylation of DNA within genes related to neural, neuroendocrine, and endocrine pathways controlling behavior and stress response.

Conclusions

Prenatal transportation stress was associated with differential methylation of DNA within genes related to the physiological control of behavior and the stress response, which are multifaceted manifestations of complex interactions between neural, neuroendocrine, and endocrine systems. Altered pathways included but were not limited to opioid signaling, CRH signaling, dopamine signaling, serotonin signaling, and GABA signaling, all of which are known to interact to influence aspects of behavior and the stress response. Association of prenatal transportation stress with genes in behavior and stress response–related signaling pathways coincided with previously reported elevations in temperament scores and circulating concentrations of cortisol observed in the larger population of calves from the which bull calves in this study were derived (Littlejohn et al., 2016). Differential methylation of DNA in different brain regions has been associated with differences in temperament in cattle (Cantrell et al., 2019). In our study, leukocytes were used as a generalized surrogate cell type. The use of peripheral tissues as surrogate cells for neural tissues in DNA methylation assessment has been widely utilized, especially in human studies in which it is not feasible to collect neural tissues. A high concordance of DNA methylation between blood cells and brain tissues across CpG sites has been repeatedly reported (Horvath et al., 2012; Tylee et al., 2013; Braun et al., 2019), though the correlation of individual CpG site methylation in blood compared with brain tissues has been reported to be variable and potentially dependent on factors such as gene function, gene region, CpG abundance within the gene region, and developmental stage in which differential methylation is established (Davies et al., 2012; Bediaga et al., 2017; Braun et al., 2019). More specifically, the degree of differential methylation associated with prenatal or early life stressors has been observed to be similar in leukocytes compared with neural cells (Provençal et al., 2012; Seifuddin et al., 2017). Rhesus monkeys that experienced early life rearing stress exhibited a significant overlap of differentially methylated CpG sites in T cells compared with prefrontal cortex cells (Provençal et al., 2012). Although the literature supports the use of leukocytes as a surrogate cell to evaluate differential methylation of neural tissue, it is important to acknowledge that differential methylation in leukocytes serves as only a partial representation of differential methylation in neural tissues. These data from bovine leukocytes warrant future evaluation of the influence of prenatal transportation stress on specific neural tissues. Future studies will evaluate methylomic and transcriptomic differences in specific neural tissues of cattle exposed to prenatal stressors. Understanding how the prenatal environment shapes postnatal behavior and stress response phenotypes may provide novel opportunities for beef cattle improvement.

Conflict of interest statement

The authors declare the absence of economic or other types of conflicts of interests.