Genetic and neuroendocrine regulation of the postpartum brain

Abstract

Changes in expression of hundreds of genes occur during the production and function of the maternal brain that support a wide range of processes. In this review, we synthesize findings from four microarray studies of different maternal brain regions and identify a core group of 700 maternal genes that show significant expression changes across multiple regions. With those maternal genes, we provide new insights into reward-related pathways (maternal bonding), postpartum depression, social behaviors, mental health disorders, and nervous system plasticity/developmental events. We also integrate the new genes into well-studied maternal signaling pathways, including those for prolactin, oxytocin/vasopressin, endogenous opioids, and steroid receptors (estradiol, progesterone, cortisol). A newer transcriptional regulation model for the maternal brain is provided that incorporates recent work on maternal microRNAs. We also compare the top 700 genes with other maternal gene expression studies. Together, we highlight new genes and new directions for studies on the postpartum brain.

Keywords: Addiction; Depression; Maternal; Mental health; MicroRNA; Oxytocin; Postpartum; Prolactin; Reward; Social behavior.

Fig. 1

Overview of possible genetic contributions to the maternal brain. While a subset of genes (~50) have been well-studied in the postpartum brain and additional genes (~150) have been identified as relevant to maternal processing via transgenic studies (left panel), most genes have not been examined in a maternal context (right panel). With over 20,000 genes in mice and humans and of those about 16–18,000 expressed in the CNS (right panel), over 99% of genes that could be involved in shaping the maternal brain still remain unexamined.

Fig. 2

Overview of synthesis of recent maternal microarray studies and outline of analysis of top 700 maternal genes. From microarray analysis of NAC, mPFC, LS, and MPOA with over 20,000 protein coding RNAs, we found high overlap of significant genes between each region and identified the top 700 maternal genes based on alterations in 3 or more regions (left part of figure; see text for additional details). Steps of analysis for the top 700 maternal genes in this review are provided in the right panel. Abbreviation: TF, transcription factor or transcriptional regulator. Overlap diagram drawn using Venny (Oliveros, 2007).

Fig. 3

Top 700 maternal genes show high enrichment with reward or addiction-related genes. (A) BigTest or MSET analysis (Eisinger et al., 2013a; Saul et al., 2016), where Y-axis represents the probability of X matches to database appearing in a randomly generated set of simulated results from the microarray background. The red arrow shows how many matches were found in the top 700 maternal genes changes and where that number falls on the probability density distribution. Enrichment is significant when evaluating top 700 maternal genes against either all reward/addiction genes (pooled from 5 databases) or against only reward/addiction genes appearing in at least two databases. When each maternal brain region is analyzed separately, enrichment is shown for all regions except mPFC, which is just above significance (top right). (B) Interaction network of matches between top 700 maternal genes and reward/addiction genes (in black) using GeneMania (see text for details). Genes in red circles have previously been associated with maternal reward, while genes in green circles are new candidates for maternal reward. GeneMania adds linking genes (in gray). Lines indicate type of evidence for interaction; yellow, predicted; red, physical interaction; purple, co-expression; gray, other; aqua blue, pathway; light blue, co-localization, gold, shared protein domain. (C) ToppCluster results showing significant enrichment of top 700 maternal genes with genes associated with six chemicals of abuse, namely nicotine, alcohol, dronabinol, methamphetamine, morphine, and cocaine. Graph on right shows the general pattern whereby some genes are only linked to one drug, but others are linked to two or three or more.

Fig. 4

Top 700 maternal genes show high enrichment with depression and (postpartum depression) PPD genes. Significant enrichment is found evaluating top 700 maternal genes against either depression genes found in two or more databases or against PPD genes within two databases using MSET analysis. The red arrow shows how many matches were found in the top 700 maternal genes changes and where that number falls on the probability density distribution (top panel). Interaction network of matches between top 700 maternal genes and depression or PPD genes (in black) using GeneMania (bottom panel). Genes in red circles are in PPD databases, while genes in green circles are new candidates for PPD. GeneMania adds linking genes (in gray).

Fig. 5

Top 700 maternal genes show high enrichment with mental health disorders that include social deficits. Significant enrichment is found evaluating top 700 maternal genes against either autism genes found in 3 or more databases, BPD genes found in 2 or more databases, or schizophrenia genes in 4 or more databases using MSET analysis. The red arrow shows how many matches were found in the top 700 maternal genes changes and where that number falls on the probability density distribution (left panel). All three disorders include social deficits (see text for details). Interaction network of matches between top 700 maternal genes and mental health genes (in black) using GeneMania (right panel). Genes in green circles are in all three disorders, while genes in red circles are in two of three disorders. GeneMania adds linking genes (in gray). Genes are tightly clustered, suggesting they may act as a functional unit, but whether they relate in general to social behavior (which is promoted in mothers and disrupted with certain mental health disorders) is not known.

Fig. 6

Overview of prolactin signaling intersection with top 700 maternal genes. Although neither prolactin (yellow circle) nor prolactin receptor (red circle), were in the top 700, both interact with a subset of top 700 maternal genes as assessed using STRING network analysis (see text for details; color of other circles are arbitrary and do not provide additional information). Cish, Socs2, and Socs3 are part of negative feedback on the prolactin receptor and thus expression changes in these genes could alter prolactin action in mothers or expression change could be due to alterations in prolactin signaling. Btrc is involved in ubiquitination events and could affect prolactin receptor action. Line colors indicate type of evidence connecting gene in STRING; blue, cooccurence; black, coexpression; purple, experiments; aqua blue, databases; light green, text mining; light blue, homology.

Fig. 7

Overview of oxytocin/vasopressin signaling intersection with top 700 maternal genes. Genes for oxytocin signaling (yellow circles) and vasopressin signaling (red circles), were not in the top 700, but interact with a subset of top 700 maternal genes as assessed using STRING network analysis (see text for details; color of other circles are arbitrary and do not provide additional information). Receptors for hypocretin (orexin) (Hcrtr1 and Hcrtr2) are among those most tightly connected to oxytocin and vasopressin signaling and the HCRT system has previously been implicated in maternal care (see text for details). Neuromedin S (Nms) may modulate oxytocin release and/or act as an intermediate between enkephalin (Penk) and dynorphin (Pdyn) action on oxytocin and vasopressin signaling. Kalirin (Kalrn) regulates neuropeptide processing and G protein signaling and multiple mRNA variants and has been associated with a number of CNS processes, including plasticity and anxiety. Cacna2d3 is of interest as it has recently linked to attachment behavior (see text). Knockout of Gna11 (along with Gnaq) disrupts maternal care, so this gene could be critical in oxytocin action. Line colors indicate type of evidence connecting genes in STRING; red, gene fusion; blue, co-occurrence; black, coexpression; purple, experiments; aqua blue, databases; light green, text mining; light blue, homology.

Fig. 8

Overview of endogenous opioid signaling intersection with top 700 maternal genes. Penk, Pdyn, and Oprm1 are all top 700 maternal genes and interact with multiple other top 700 genes as assessed using STRING network analysis (see text for details; color of other circles are arbitrary and do not provide additional information). Among genes of interest are the somatostatin receptor (Sstr4), the acetylcholine receptor (Chrm4), the histamine receptor (Hrh3), the serotonin receptor (Htr1a), the glutamate receptor (Grm3), the adenosine receptor (Adora2a), and the dopamine receptor (Drd1a) (see text for more details on these genes). The neuropeptide, Nms (described above), could be interacting with Pdyn and Penk release or action. Recently, Grm3 was associated with dependence on the opioid (see text for details), so Grm3 could be critical in how endogenous opioids acts. Line colors indicate type of evidence connecting gene in STRING; red, gene fusion; blue, co-occurrence; black, coexpression; purple, experiments; aqua blue, databases; light green, text mining; light blue, homology.

Fig. 9

Overview of steroid receptor signaling (estradiol, progesterone, and glucocorticoid) intersection with top 700 maternal genes. Genes for estradiol signaling (yellow circles), glucocorticoid signaling (red circle), and progesterone signaling (green circle) were not in the top 700, but interact with a subset of top 700 maternal genes as assessed using STRING network analysis (see text for details; color of other circles are arbitrary and do not provide additional information). A high number of transcriptional regulators are identified, including Nr1d1, Hif1a, Nr4a1, Ncor2, Ep300, Egr1, Nr4a3, Npas2, Fkbp4, Rarb, Rxrg, and Creb1 (see text for information on known interactions with the steroid receptors). Line colors indicate type of evidence connecting gene in STRING; red, gene fusion; blue, co-occurrence; black, coexpression; purple, experiments; aqua blue, databases; light green, text mining; light blue, homology.

Fig. 10

Transcriptional regulation and the production of the maternal brain. In recent work directly examining miRNA expression across reproduction, we found 32 miRNAs consistently upregulated miRNAs (pregnant/postpartum relative to virgin) and 25 miRNAs consistently downregulated (Saul et al., 2016). Each miRNA can have multiple RNA targets and combined, the up and downregulated miRNAs target 4626 genes (top panel). The top 700 maternal genes showed extremely high enrichment for the miRNA targets with 41% of the top 700 being targets (BigTest p-value = 0.002; hypergeometric p-value <10⁻⁸). Within the 41% are 46 transcription regulators (bottom panel) that in turn can regulate miRNAs and/or other genes within the top 700 maternal genes. Thus, a complex interaction occurs whereby miRNAs target numerous maternal genes and a subset of these genes then can transcriptionally regulate other maternal genes as well as the miRNAs (middle diagram).

Fig. 11

Overview of top 700 maternal genes involved in development, differentiation, neurogenesis, and plasticity pathways. Interaction network of matches between top 700 maternal genes and genes associated with various CNS plasticity events (in black) using GeneMania. GeneMania adds the prolactin receptor (green circle) as a linking gene, which is consistent with its known role in cellular plasticity (see text). Because these genes represent adult changes in plasticity/development, we tested and found a significant overlap of top 700 maternal genes with the adult degenerative disorders, such as Alzheimer’s. The common genes are shown in red circles and may provide insight into how changes in the maternal brain could indicate a pathway for restoring the CNS from degeneration in adults.

Fig. 12

Top 700 maternal genes show high enrichment with top genes from new maternal RNA seq study. Significant enrichment (hypergeometric test) is found evaluating top 700 maternal genes against new RNA seq top 700 genes for postpartum day 3 (PPD3) or PPD10 relative to virgin. For comparisons, data from the maternal RNA seq study from hippocampus, cortex, hypothalamus, and cerebellum for PPD3 and PPD10 (Ray et al., 2016) was combined to provide a top 700 list (high rank in at least three of four regions; see text for details). The red arrow shows how many matches were found in the top 700 maternal genes changes and where that number falls on the probability density distribution (left panel). 108 top 700 genes were found in at least one of the two other lists (PPD3 or PPD10) and 34 genes were found in the top 700 of all three lists.

Fig. 13

Chromosome distribution of top 700 maternal genes. Blue dots indicate top 700 maternal genes and as shown, most are evenly distributed across chromosomes. The X chromosome had the lowest ratio of maternal genes per total genes. For miRNAs, one region on chromosome 12 (12qF1) is highly enriched for upregulated miRNAs (Saul et al., 2016) and this regions has been connected to cell differentiation events (see text for details). Background image taken from UCSC Genome Browser (http://genome.ucsc.edu) (Kent et al., 2002).