Temporal Dynam ics of the Neuregulin–ErbB Network in the Murine Prefrontal Cortex across the Lifespan

Abstract

Neuregulin–ErbB signaling is essential for numerous functions in the developing, adult, and aging brain, particularly in the prefrontal cortex (PFC). Mouse models with disrupted Nrg and/or ErbB genes are relevant to psychiatric, developmental, and age-related disorders, displaying a range of abnormalities stemming from cortical circuitry impairment. Many of these models display nonoverlapping phenotypes dependent upon the gene target and timing of perturbation, suggesting that cortical expression of the Nrg–ErbB network undergoes temporal regulation across the lifespan. Here, we report a comprehensive temporal expression mapping study of the Nrg–ErbB signaling network in the mouse PFC across postnatal development through aging. We find that Nrg and ErbB genes display distinct expression profiles; moreover, splice isoforms of these genes are differentially expressed across the murine lifespan. We additionally find a developmental switch in ErbB4 splice isoform expression potentially mediated through coregulation of the lncRNA Miat expression. Our results are the first to comprehensively and quantitatively map the expression patterns of the Nrg–ErbB network in the mouse PFC across the postnatal lifespan and may help disentangle the pathway’s involvement in normal cortical sequences of events across the lifespan, as well as shedding light on the pathophysiological mechanisms of abnormal Nrg–ErbB signaling in neurological disease.

Introduction

The prefrontal cortex (PFC) plays a fundamental role in the organization of higher cognitive functions including attention, memory, decision-making, and behavioral flexibility. These functions of the PFC develop and decline across the lifespan, concurrent with the protracted maturation profile of this brain region (Sowell et al. 2003; Gogtay et al. 2004; Tau and Peterson 2010), and are influenced by the tightly regulated temporal coordination of expression of thousands of genes (Colantuoni et al. 2011). The human PFC’s protracted period of maturation is conserved in nonhuman primates and rodents and undergoes dramatic anatomical and neurochemical changes across the postnatal lifespan in these species (Goldman-Rakic and Brown 1982; Liu et al. 2004; Wang and Gao 2009; Markham et al. 2013; Ueda et al. 2015). Given that the PFC represents a central brain region dysregulated in not only neurodevelopmental disorders but also disorders commonly associated with aging (Callicott et al. 2000; Salat et al. 2001; Brennan and Arnsten 2008; Schumann et al. 2010), understanding the temporal dynamics of biological signaling pathways that regulate molecular events in the PFC across the lifespan can allow for insight into normative brain development and aging, as well as how temporal changes may relate to neurological disease.

One such signaling pathway that acts as a critical mediator of numerous molecular events in the mammalian PFC across the lifespan is the Neuregulin–ErbB (Nrg–ErbB) signaling network (Meyer and Birchmeier 1995; Mei and Xiong 2008; Rico and Marin 2011; Mei and Nave 2014). While a large focus of research has highlighted the critical need for regulated Nrg–ErbB signaling in early neurodevelopmental processes including cell migration and proliferation (Anton et al. 2004; Flames et al. 2004), many studies have also demonstrated the importance of this signaling pathway in postnatal functions including dendritic growth and branching, synapse development and pruning, and myelination (Michailov et al. 2004; Li et al. 2007; Fazzari et al. 2010; Yang et al. 2013). Moreover, Nrg–ErbB signaling plays a vital neuroprotective role in the aging brain (Jiang et al. 2016; Xu et al. 2016). Previous studies have shown that the consequences of altered Nrg–ErbB signaling levels are dependent upon on the timing of the perturbation, suggesting that the expression of members of the Nrg–ErbB network undergo tight temporal regulation across the lifespan (Paterson and Law 2014; Loos et al. 2016; Wang et al. 2018). Furthermore, mice carrying targeted mutations altering the levels of Nrg–ErbB signaling display a range of neurobehavioral phenotypes relevant to neurodevelopmental, neuropsychiatric, and neurological disorders (Chen et al. 2008; Barros et al. 2009; Deakin et al. 2009; Paterson and Law 2014; Papaleo et al. 2016; Wang et al. 2018); however, these behavioral profiles are heavily dependent upon which member of the network has been modified. Many of these behavioral phenotypes stem from dysfunction of the PFC, providing convergent evidence for the necessity of controlled Nrg–ErbB signaling across the lifespan for prefrontal cortical development, homeostasis, and aging.

Both Neuregulin and ErbB gene families are complex, with six Neuregulin genes (NRG1–6) and four ErbB genes (ErbB1–4) having been identified to date (Carraway et al. 1997; Zhang et al. 1997; Harari et al. 1999; Kanemoto et al. 2001; Falls 2003; Kinugasa et al. 2004; Law et al. 2006). As an added level of complexity, many of these genes undergo extensive alternative splicing, for example, through alternative splicing and multiple promoter usage at least 30 isoforms of NRG1 exist (Law et al. 2006; Mei and Nave 2014). Notably, there is significant nonredundancy between members of these gene families and even between splice isoforms as demonstrated by nonoverlapping behavioral phenotypes in murine genetic models. Moreover, we, and others, have shown the differential involvement of altered expression of NRG and ErbB genes, as well as NRG1 and ErbB4 isoforms in multiple neuropsychiatric disorders (Hashimoto et al. 2004; Law et al. 2006, 2007; Chong et al. 2008; Nicodemus et al. 2009; Paterson et al. 2014, 2017; Chung et al. 2018). Furthermore, we have demonstrated that in the human PFC expression levels of NRG1 isoforms display nonoverlapping expression trajectories across the lifespan further suggesting that each member of the NRG–ErbB is differentially temporally regulated with a unique molecular signature that may elucidate the diverging roles of each gene in cortical functions at different life stages (Paterson et al. 2014).

Previous studies have shown in human, nonhuman primate, and rodent brain that multiple members of the NRG–ErbB network are expressed in the brain across the lifespan (Corfas et al. 1995; Meyer et al. 1997; Pinkas-Kramarski et al. 1997; Steiner et al. 1999; Longart et al. 2004; Mechawar et al. 2007; Thompson et al. 2007; Liu et al. 2011). However, these studies are predominantly qualitative by histological analysis, investigating sole members of the NRG–ErbB network, and across limited sample ages. Therefore, our understanding and interpretation of murine models of altered NRG–ErbB function are incomplete. To aid in the understanding of the divergence of Nrg–ErbB postnatal functions, we performed a comprehensive quantitative expression mapping analysis of the Nrg–ErbB gene network during PFC development and aging in male C57BL/6 mice, examining gene expression of multiple members of the Nrg–ErbB network (including splice isoforms) across numerous developmental time points, from postnatal day (PND) 1 through 2 years of age.

We identify that each member of the Nrg–ErbB signaling network displays a unique cortical expression profile across the mouse postnatal lifespan. In many cases, these nonoverlapping expression signatures correspond with the diverse range of known biological functions of NRG–ErbB signaling and can provide complementary evidence for which distinct members of the network are involved in these functions during discrete life phases. Furthermore, we observed that splice isoforms of Nrg1 and ErbB4 display marked differences in their expression trajectories and identify the long noncoding RNA Miat as a potential temporal regulator of ErbB4 splice isoform shifting during mouse postnatal life. Together, our findings not only serve to highlight the pleiotropy within Nrg and ErbB genes but can also serve as a resource for the development and interpretation of mouse models of altered Nrg–ErbB signaling.

Materials and Methods

Animals

C57BL/6 mice breeding pairs were purchased from Charles Rivers and were bred, reared, and housed in standard ventilated cages with ad libitum access to chow and water under standard 12 h light/dark cycle. Pups were weaned at PND 21 and group-housed by sex. Only male mice were used for the current study. Male mice were euthanized by decapitation at multiple different time points spanning the lifespan, including daily time points throughout PNDs 1–10, weekly time points from PND14 to PND49, and monthly time points from 2 to 24 months of age. Each time point assessed was composed of samples obtained from ≥2 independent litters as to minimize the potential confound of litter effects within a single time point. All animal procedures were carried out in accordance with and approved by the University of Colorado Institutional Animal Care and Use Committee.

RNA Extraction and cDNA Synthesis

PFC tissue was dissected from n = 3–6 mice per time point. For consistency in PFC tissue collection, all dissections were carried out by an individual investigator using Paxinos and Franklin mouse atlas for anatomical reference (Paxinos and Franklin 2012). In neonatal mice, dissections were performed under magnification to achieve the most comparative tissue collection across ages utilizing neuroanatomical hallmarks. Total RNA was isolated from the tissue using TRIzol extraction method according to manufacturer’s recommendations (RNeasy Lipid tissue kit, Qiagen). RNA quality and concentration were assessed with Nanodrop 2000 spectrophotometer (Thermo Scientific). About 2 μg of each sample in a final reaction volume of 20 μL was used for first strand cDNA synthesis using Multiscribe Reverse Transcriptase kit with random primers according to manufacturer’s recommendations (Life Technologies).

Quantitative Reverse Transcription Polymerase Chain Reaction

mRNA levels of members of the Neuregulin–ErbB network were measured by Inventoried TaqMan Gene Expression Assays (Life Technologies) (pan-Nrg1, Nrg2, Nrg3, ErbB2, ErbB3, pan-ErbB4 and Miat), custom-designed primers and internal TaqMan probes (Life Technologies) (ErbB4-JMa, ErbB4-JMb, ErbB4-CYT1, ErbB4-CYT2 and Nrg1-III), or custom-designed SYBR assays (Nrg1-I and Nrg1-II) using an ABI 7900HT fast real-time PCR system in 384-well plate format (Life Technologies). Assay sequences for each transcript examined are detailed in Table 1. Where custom-designed assays were used, PCR products were initially analyzed for verification of expected amplicon size and sequence. For each target gene, each sample was performed in triplicate and relative quantities were determined from a 6-point standard curve derived from pooled samples of each lifespan time point. For each sample, the expression levels of each gene of interest were normalized to the geometric mean of two internal control housekeeping genes (Gapdh and β-Actin). Importantly, the geometric mean of these housekeeping genes did not significantly correlate with age (P > 0.05).

Table 1.

Details of reverse transcriptase quantitative PCR assays used in the current study

Inventoried TaqMan assays	Assay ID
Gapdh	Mm99999915_g1
β-Actin	Mm00607393_s1
Pan ErbB4	Mm01256793_m1
ErbB2	Mm00658541_m1
ErbB3	Mm01159999_m1
Pan Nrg1	Mm01212135_m1
Nrg2	Mm01158087_m1
Nrg3	Mm01209104_m1
Miat	Mm01196418_g1
Custom-designed TaqMan assays	Sequence
ErbB4 JMa	Forward primer: CCACCCTTGCCATCCAAA
	Reverse primer: CCAATGACTCCGGCTGCAATCA
	Internal probe: ATGGACGGGCCATTCCACTTTACCA
ErbB4 JMb	Forward primer: CCACCCTTGCCATCCAAA
	Reverse primer: CCAATGACTCCGGCTGCAATCA
	Internal probe: TCAAGCATTGAAGACTGCATCGGCCTGAC
ErbB4 CYT1	Forward primer: CAACATACCTCCTCCCATCTACAC
	Reverse primer: GGCATTCCTTGTTGTGTAGCAA
	Internal probe: GAAATTGGACACAGCCCTCCTCCTG
ErbB4 CYT2	Forward primer: CAACATACCTCCTCCCATCTACAC
	Reverse primer: GGCATTCCTTGTTGTGTAGCAA
	Internal probe: ATTGACTCCAATAGGAATCAGTTTGTGTACCAAGAT
Nrg1-III	Forward primer: CACCCAAGTCAGGAACTCAG
	Reverse primer: TCCTTCACCATGAAGCACTC
	Internal probe: CTGCTCCTAAACTTTCCACATCTACATCC
Custom-designed SYBR assays	Sequence
Nrg1-I	Forward primer: AAAGAAGGCAGAGGCAAGG
	Reverse primer: TCTGAGTGAGGAGTACTCAG
Nrg1-II	Forward primer: CAAGGAGGACAGCAGGTACA
	Reverse primer: TCTGAGTGAGGAGTACTCAGAGC

Western blotting

Western blotting was carried out on fresh frozen PFC samples obtained at PND1, PND35, 3 months, and 24 months as previously described (Papaleo et al. 2016). In brief, PFC samples were homogenized and protein extracted with T-PER lysis buffer supplemented with protease and phosphatase inhibitors (Thermo Scientific). Protein concentrations were quantified using a BCA method. About 35 μg of protein was electrophoresed through 4–12% gels and transferred onto PVDF membranes (Invitrogen). The following primary antibodies were diluted in 5% milk in TBST (Tris-Buffered Saline 1% Tween 20): anti-Nrg1 (1:500; Santa Cruz Biotechnology sc-348), anti-Nrg3 (1:5000; Abcam ab77597), anti-ErbB4 (1:1000; Santa Cruz Biotechnology sc-283), and anti-β-Actin-HRP (1:10000; Sigma). HRP-conjugated secondary antibodies (Santa Cruz Biotechnology) against rabbit were used at a 1:5000 dilution when needed. Immunoreactive bands were visualized using a FluorChem Q image analyzer (Protein Simple). The optical density of the protein bands corresponding to full-length protein products was quantified using Image J (NIH). Expression levels of each protein were normalized to the respective levels of that sample’s β-actin expression derived from the same blot.

Statistical analysis

All statistical analyses were performed using SPSS statistics package version 25. Primary analysis of Pearson’s correlations followed by Bonferroni multiple comparison correction was used to explore the relationship between the expression of each gene of interest and age across the lifespan. In a secondary analysis, we performed Pearson correlation analysis for each gene expression “within” lifestages that reflect critical windows of development, including early postnatal, prepuberty, postpuberty, and adulthood through aging (Rice and Barone 2000; Tau and Peterson 2010; Markham et al. 2013). Correlations deriving a P ≤ 0.003 (to account for Bonferroni multiple comparisons correction) were determined statistically significant for primary comparisons across the lifespan, and P ≤ 0.05 were determined statistically significant for secondary within lifestage correlation analysis. For western blot analysis, significant effects of age on protein expression were examined using one-way ANOVA. Data from all samples were included for statistical analysis unless otherwise noted.

Results

We sought to examine the cortical expression trajectories of a comprehensive list of Nrg and ErbB genes and their splice isoforms across the entire mouse postnatal lifespan from PND 1 to 2 years of age (n = 95). Furthermore, to provide insight into the potential roles of each member of the Nrg–ErbB network within specific lifestages, we mapped the molecular signatures of each gene within defined time frames that represent critical windows of postnatal development (early postnatal PND1–7, n = 25; prepubertal PND8–35, n = 21; postpubertal PND42–3 months, n = 14, and adulthood through aging 4–24 months, n = 35).

Murine Neuregulin Genes Display Nonoverlapping Cortical Expression Profiles across the Lifespan

Of the Neuregulin genes identified, NRG1–3 are best characterized for their roles in the central nervous system (CNS) and association to neurological dysfunction. While Nrg1, Nrg2, and Nrg3 expression is abundant in the mouse brain and has been localized to the mouse cortex (Longart et al. 2004), their expression profiles have not simultaneously been quantitatively mapped across the postnatal murine lifespan.

We did not find a significant correlation between age and expression of Nrg1 in the mouse PFC (r = −0.175, P > 0.05), whereby the expression remained relatively stable from PND1 to 2 years of age, with no individual lifestage displaying a significant correlation within the age range (early postnatal r = −0.230, P = 0.269; prepubertal r = −0.111, P = 0.632; postpubertal r = 0.304, P = 0.291; adulthood through aging r = 0.151, P = 0.386; Fig. 1A). Similarly, Nrg2 expression was also stable across murine PFC postnatal development and aging (r = 0.137, P > 0.05); however, significant trajectories of Nrg2 expression were observed within lifestages, including a significant developmental increase in Nrg2 within the early postnatal period (r = 0.427, P = 0.033) followed by a significant decrease during prepubescence (r = −0.613, P = 0.003), before significantly increasing again during the postpubertal period (r = 0.553, P = 0.040), and stabilizing to constant levels across adulthood and aging (r = 0.302, P = 0.078) (Fig. 1B). Unlike Nrg1 and Nrg2, the expression of Nrg3 was significantly positively correlated with age (r = 0.661, P < 0.001). This significant expression increase of Nrg3 in the mouse PFC across aging was due to a steadily increasing trajectory across all lifestages analyzed (early postnatal r = 0.134, P = 0.522; prepubertal r = 0.087, P = 0.709; postpubertal r = −0.109, P = 0.710; adulthood through aging r = −0.193, P = 0.268) (Fig. 1C).

Figure 1.

Gene expression profiles of Neuregulin genes Nrg1 (A), Nrg2 (B), and Nrg3 (C) in the postnatal mouse PFC from PND 1 through to 24 months of age. The geometric mean of Gapdh and βActin was used for normalization. Each data point represents an individual mouse PFC sample. N = 3–6 samples/age, n = 95 total lifespan. ^*P < 0.05, ^**P < 0.01 significant correlation with age within lifestage (early postnatal n = 25, prepubertal n = 21, postpubertal n = 14, adulthood through aging n = 25). ^###P < 0.001 significant correlation across postnatal age. Lines represent best fit expression trajectory within lifestage (gray) or across entire postnatal period (black). M, month.

Isoform-Specific Expression Trajectories of Neuregulin 1 Transcripts in the Murine PFC

Numerous isoforms of the NRG1 gene have been identified through alternative splicing and differential promoter use (Steinthorsdottir et al. 2004; Tan et al. 2007). These isoforms can be broadly grouped into six “Types” (Nrg1 type I-VI) each differing in their N-terminal exon and through the presence of an immunoglobulin-like domain (Types I, II, IV, V, and VI) or cysteine-rich domain (Type III) which impacts cellular localization and cleavage susceptibility of the membrane bound growth factor (Mei and Xiong 2008). In vitro and in vivo studies have demonstrated that Nrg1 isoforms harbor distinct neurological functions and exhibit nonoverlapping expression patterns in the rodent brain (Meyer and Birchmeier 1994; Corfas et al. 1995; Meyer et al. 1997; Liu et al. 2011); however, these histological studies have predominantly focused on the embryonic developmental period and little is known regarding Nrg1 isoform signatures postnatally, specifically in the mouse PFC.

NRG1 types I–IV have been most extensively studied in the context of neurological function, while the expression of Nrg1 type IV in the postnatal rodent brain is still under debate (Tan et al. 2007; Liu et al. 2011); therefore, we focused on mapping the expression trajectories of Nrg1 types I–III in the mouse PFC across the entire postnatal lifespan.

Like Nrg1, alternative splicing of the murine Nrg2 and Nrg3 genes has been reported (Britto et al. 2004; Howard et al. 2005); however, unlike Nrg1 the splicing of other Neuregulins is comparatively less complex, with the presence or functionality of these splice isoforms in the murine central nervous system remaining largely unknown. Moreover, while conservation of Nrg1 splicing is highly conserved between mouse and human genomes, splicing of mouse Nrg3 does not appear to impact the 5′ region of the gene, which is highly transcriptionally variable and responsible for generating temporally regulated subclasses of NRG3 splice isoforms in the human genome (Kao et al. 2010; Paterson et al. 2017). For these reasons, we restricted our mapping of Neuregulin splice isoforms to the Nrg1 gene.

Nrg1 type I expression was significantly negatively correlated with age (r = −0.846, P < 0.001), whereby Nrg1 type I levels dramatically declined both during the early postnatal (r = −0.679, P < 0.001) and prepubertal periods (r = −0.797, P < 0.001), stabilized during post puberty (r = −0.084, P = 0.774), before continuing to steadily decline throughout adulthood into aging (r = −0.449, P = 0.007) (Fig. 2A). Conversely, Nrg1 type II expression in the mouse PFC displayed a significant positive correlation with age (r = 0.352, P < 0.001), stemming from a gradual incline in expression levels across each lifestage (early postnatal r = 0.089, P = 0.674; prepubertal r = 0.145, P = 0.531; postpubertal r = 0.028, P = 0.925; adulthood through aging r = 0.251, P = 0.146) (Fig. 2B). While Nrg1 type I and type II displayed opposing patterns of expression in the mouse PFC across postnatal life decreasing and increasing with age, respectively, Nrg1 type III expression presented an nonoverlapping signature remaining stable across the entire lifespan (r = −0.090, P > 0.05), with only the postpubertal lifestage showing a modestly significant increase in expression levels (early postnatal r = 0.000, P = 0.999; prepubertal r = −0.234, P = 0.308; postpubertal r = 0.542, P = 0.045; adulthood through aging r = 0.234, P = 0.183) (Fig. 2C).

Figure 2.

Gene expression profiles of Neuregulin 1 alternative splice isoforms Nrg1 type I (A), Nrg1 type II (B), and Nrg1 type III (C) in the postnatal mouse PFC from PND 1 through to 24 months of age. The geometric mean of Gapdh and βActin was used for normalization. Each data point represents an individual mouse PFC sample. N = 3–6 samples/age, n = 95 total lifespan. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 significant correlation with age within lifestage (early postnatal n = 25, prepubertal n = 21, postpubertal n = 14, adulthood through aging n = 25). ^###P < 0.001 significant correlation across postnatal age. Lines represent best fit expression trajectory within lifestage (gray) or across entire postnatal period (black). M, month.

ErbB Receptor Tyrosine Kinase Genes Diverge in Expression across the Murine Postnatal Lifespan

Neuregulin function in the central nervous system is largely mediated through binding to and activation of the ErbB tyrosine kinase family of receptors (ErbB2–4). Upon growth factor binding ErbB receptors form dimers inducing autophosphorylation and subsequent canonical initiation of intracellular signaling cascades (Birchmeier 2009). ErbB2–4 display unique binding profiles and affinities for varying neuregulins and thus can exert nonoverlapping functions in the CNS, which is evidenced by targeted mutations in ErbB receptors displaying differing viability and behavioral profiles (Erickson et al. 1997; Gerlai et al. 2000; Roy et al. 2007). Moreover, preliminary investigations show that ErbB receptors have nonoverlapping temporal and spatial expression trajectories in the rodent brain (Meyer et al. 1997; Pinkas-Kramarski et al. 1997; Ozaki et al. 1998; Steiner et al. 1999; Fox and Kornblum 2005; Abe et al. 2009); however, these studies failed to investigate, at a quantitative level, the mouse PFC across the lifespan. We sought to address this gap by studying the expression profiles of ErbB2–4 in the mouse PFC from birth to 2 years of age.

In the mouse PFC, ErbB2 displayed a significantly negatively correlated expression trajectory with age (r = −0.341, P < 0.001). Secondary examination of individual lifestages identified that this negative correlation originated from significant changes in ErbB2 expression during the early postnatal (r = 0.410, P = 0.042) and prepubertal periods (r = −0.527, P = 0.014), while there were no expression correlations in older lifestages (postpubertal r = −0.333, P = 0.245; adulthood through aging r = 0.172, P = 0.322) (Fig. 3A). In contrast, ErbB3 expression levels were significantly positively correlated with age (r = 0.892, P < 0.001), displaying a unique cortical expression trajectory across the mouse lifespan whereby the lowest levels of expression were observed during PND1–PND9 (Fig. 3B). Specifically, we identified that ErbB3 expression significantly increased across the early postnatal period (r = 0.520, P = 0.008), followed by a marked increase in expression throughout the prepubertal period (r = 0.820, P < 0.001) stabilizing during post puberty (r = −0.355, P = 0.213) and again increasing within adulthood into aging (r = 0.540, P = 0.001). Comparable to ErbB3, ErbB4, expression was also significantly positively correlated with age (r = 0.604, P < 0.001); however, unlike ErbB3, cortical ErbB4 expression profile displayed a gradual incline throughout the 2-year lifespan studied with only the prepubertal lifestage displaying a significant positive correlation (early postnatal r = 0.307, P = 0.136; prepubertal r = 0.617, P = 0.003; postpubertal r = 0.412, P = 0.144; adulthood through aging r = 0.151, P = 0.385) (Fig. 3C).

Figure 3.

Gene expression profiles of v-erb homolog receptor tyrosine kinase genes ErbB2 (A), ErbB3 (B), and ErbB4 (C) in the postnatal mouse PFC from PND 1 through to 24 months of age. The geometric mean of Gapdh and βActin was used for normalization. Each data point represents an individual mouse PFC sample. N = 3–6 samples/age, n = 95 total lifespan. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 significant correlation with age within lifestage (early postnatal n = 25, prepubertal n = 21, postpubertal n = 14, adulthood through aging n = 25). ^###P < 0.001 significant correlation across postnatal age. Lines represent best fit expression trajectory within lifestage (gray) or across entire postnatal period (black). M, month.

ErbB4 Splice Isoform Expression Ratios Shift across the Murine Lifespan and Correlate with the Putative Splicing Regulator and Long Noncoding RNA Miat

Given that we identified temporal regulation of ErbB4 expression in the mouse PFC across the lifespan and that ErbB4 is the predominant signaling partner of Neuregulins in the CNS, in that it is the only ErbB sufficient to both bind, and becomes activated by neuregulins without heterodimerization (Birchmeier 2009), and is implicated in neuropsychiatric illness (Law et al. 2007; Chong et al. 2008; Chung et al. 2016, 2018); we investigated this gene in further detail. Similar to NRG1, ErbB4 is subject to alternative splicing, with transcripts differing in their susceptibility to proteolytic cleavage and PIK3 docking domain having been identified in both human and rodent brain (Mei and Xiong 2008). These splice isoforms (termed JMa, JMb, CYT1, and CYT2) have been differentially associated with schizophrenia and additional neuropsychiatric disorders (Law et al. 2007; Chung et al. 2016, 2018). The temporal expression profiles have never been mapped in the typically developing mouse brain. Therefore, next we quantified the cortical expression of ErbB4 JMa, JMb, CYT1, and CYT2 across mouse postnatal lifespan.

While the murine cortical expression of the ErbB4 splice isoform JMa was significantly negatively correlated with age (r = −0.585, P < 0.001) (Fig. 4A), examination of specific lifestages revealed that the expression of JMa fluctuated dramatically across different stages of the lifespan. Specifically, JMa expression significantly decreased during the early postnatal period (r = −0.533, P < 0.001), increased during prepuberty (r = 0.624, P = 0.002), significantly decreasing during post puberty (r = −0.612, P = 0.007) before becoming subsequently stable until 2 years of age (r = 0.280, P = 0.104). JMb expression on the other hand displayed an expression profile significantly positively correlated with age (r = 0.636, P < 0.001), exhibiting a trajectory of gradual and nonsignificant expression incline across all lifestages until 2 years of age (early postnatal r = 0.147, P = 0.483; prepubertal r = 0.285, P = 0.210; postpubertal r = 0.469, P = 0.091; adulthood through aging r = 0.262, P = 0.128) (Fig. 4B). CYT1 expression trajectory exhibited a strong negative correlation with age (r = −0.743, P < 0.001), with steep declines in CYT1 expression observed during both early postnatal (r = −0.711, P < 0.001) and prepubertal periods (r = −0.641, P = 0.002). CYT1 expression stabilized across the postpubertal stage (r = 0.114, P = 0.699), before beginning to increase during adulthood and aging (r = 0.622, P < 0.001) (Fig. 4C). In a very similar pattern to JMb, and contrary to CYT1, CYT2 expression was significantly positively correlated with aging, progressively increasing until 2 years of age (r = 0.554, P < 0.001) (Fig. 4D). Expression levels of CYT2 in the murine PFC gradually increased within each lifestage examined, with the exception of the prepubescent stage where CYT2 levels significantly increased between PND8–35 (early postnatal r = 0.058, P = 0.784; prepubertal r = 0.519, P = 0.016; postpubertal r = 0.425, P = 0.130; adulthood through aging r = 0.168, P = 0.336). These results indicate that in the developing murine PFC JMa/CYT1 represents the major ErbB4 splice isoforms, whereas, in the adult and aging murine brain, there is a conversion to JMb/CYT2 becoming the major splice isoforms present.

Figure 4.

Gene expression profiles of ErbB4 splice isoforms JMa (A), JMb (B) CYT1 (C), and CYT2 (D) as well as the long noncoding RNA Miat (E) in the postnatal mouse PFC from PND 1 through to 24 months of age. The geometric mean of Gapdh and βActin was used for normalization. Correlation analysis of normalized Miat expression to the ratio of JMa:JMb (white data points) and CYT1:CYT2 (black data points) ErbB4 transcripts (F). Each data point represents an individual mouse PFC sample. N = 3–6 samples/age, n = 95 total lifespan. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 significant correlation with age within lifestage (early postnatal n = 25, prepubertal n = 21, postpubertal n = 14, adulthood through aging n = 25). ^###P < 0.001 significant correlation across postnatal age. Lines represent best fit expression trajectory within lifestage (gray) or across entire postnatal period (black). M, month.

To further address this shift in ErbB4 splice isoform expression across the lifespan, we sought to investigate the expression profile of the lncRNA Miat, which has been previously demonstrated to regulate ErbB4 splicing (Barry et al. 2014), has itself been implicated in schizophrenia (Barry et al. 2014; Chung et al. 2016), and has important functions in the developing murine brain (Mercer et al. 2008). Interestingly, Miat displayed a similar expression profile to that of JMb and CYT2 trajectories, significantly increasing in expression levels gradually across the murine lifespan (r = 0.672, P < 0.001) (Fig. 4E). Identical to CYT2, examination of Miat expression trajectories within life stages revealed that levels of Miat in the murine PFC during the prepubescent stage significantly increased between PND8–35 (r = 0.623, P = 0.003), but not during the other lifestages (early postnatal r = −0.021, P = 0.922; postpubertal r = 0.092, P = 0.756; adulthood through aging r = −0.304, P = 0.076). These findings begin to provide evidence to suggest that increasing levels of Miat may preferentially induce splicing of ErbB4 JMa/CYT2 transcripts in the adult mouse PFC. Furthermore, in support of this finding, when we examined all samples (regardless of age), we found that the samples with the highest JMa:JMb or CYT1:CYT2 expression ratios also had the lowest expression of Miat (Fig. 4F) (significant negative correlation between JMa:JMb and Miat expression, or CYT1:CYT2 and Miat expression r = −0.571, P < 0.001 and r = −0.586, P < 0.001, respectively).

Cortical Nrg–ErbB Protein Expression across the Murine Lifespan

To gain insight into the cortical expression profiles of the Nrg–ErB network across the murine lifespan at the translational level, we performed western blot analysis to quantify protein levels of Nrg1, Nrg3, and ErbB4 in the mouse PFC at select time points across the lifespan (PND1, PND35, 3 months, and 24 months of age). Utilizing antibodies that we have previously used and validated in human and murine brain tissue that detect all isoforms of each protein, we observed, in agreement with our previous studies, the expected full-length protein products of 130, 77, and 180 kDa for Nrg1, Nrg3, and ErbB4, respectively (Fig. 5) (Chong et al. 2008; Kao et al. 2010.

Figure 5.

Expression profiles of Nrg–ErbB proteins across murine cortical development and aging. Densitometry analysis of western blots for full-length protein products of Neuregulin 1 (A), Neuregulin 3 (B), and ErbB4 (C) in the postnatal mouse PFC at PND 1, PND 35, 3 months (M), and 24 M. Expression of β-Actin was used for normalization. Data represent mean ± SEM. n = 3 samples/age.

Full-length Nrg1 protein expression significantly changed across the developmental time points studied, F(3,11) = 29.239, P < 0.001, increasing dramatically between PND1 and PND35 after which protein levels remained steady (Fig. 5A). Cortical protein expression levels of Nrg3 were also significantly associated with age, F(3,11) = 7.423, P = 0.01; however, Nrg3 displayed a unique expression profile to Nrg1 peaking in expression at 3 months of age (Fig. 5B). In a similar pattern to Nrg3, full-length ErbB4 protein expression was temporally regulated across the developmental time points studied, F(3,11) = 17.179, P = 0.001 (Fig. 5C).

Discussion

Here, we provide a comprehensive map of gene expression in the mouse PFC from the early postnatal period through aging, focusing on the Nrg–ErbB network. We report for the first time using quantitative expression mapping over the entire postnatal lifespan that each member of the Nrg–ErbB signaling network displays nonoverlapping expression trajectories. Additionally, we provide novel evidence that ErbB4 splice isoforms are differentially temporally regulated and present findings to suggest that the long noncoding RNA Miat may regulate the switching of ErbB4 splice isoforms across postnatal life. Together, our findings have important implications for our understanding of how the diverse range of functions attributed to Nrg–ErbB signaling may be imparted by distinct Nrg and ErbB splice isoforms at discrete epochs of cortical development.

Expression mapping of Nrg and ErbB genes in the mouse postnatal PFC revealed that this network is temporally regulated, with 11 of the 14 gene targets studied displaying statistically significant dynamic expression profiles across postnatal life. Interestingly, when we examine which critical windows of cortical development exhibit expression changes, 75% of the observations occur within the early postnatal and prepubertal life stages, corresponding with Nrg–ErbB’s known roles in cortical migration, synaptic development, and myelination all of which occur during these early postnatal periods (Rice and Barone 2000). Moreover, these periods represent epochs where GABAergic, glutamatergic, cholinergic, and dopaminergic neurotransmitter systems thought to be influenced by Nrg–ErbB signaling (Mei and Xiong 2008; Mei and Nave 2014) are evolving into their mature state (Ben-Ari et al. 1997; Liu et al. 2004; Ueda et al. 2015). Notably, we saw very few changes in expression of the Nrg–ErbB network specifically within “Postpubertal” and “adulthood through aging” stages suggesting that maintenance and homeostasis of mature prefrontal cortical circuitry may be more restricted to a limited number of members of the network.

While nonoverlapping spatial and temporal expression profiles of Nrg1 isoforms have been previously been reported in mouse, histologically and at a limited number of postnatal ages (Corfas et al. 1995; Meyer et al. 1997), we extend on these findings by providing the quantitative expression signatures of the isoforms across the entire postnatal murine lifespan. The demonstration that Nrg1 type I cortical expression decreases across the postnatal lifespan, most dramatically during the first month of life is consistent with a previous study in rat cerebral cortex (Liu et al. 2011). Nrg1 type I has been most extensively studied for its roles in central nervous system myelination (Michailov et al. 2004; Brinkmann et al. 2008). Consistent with the importance of white matter integrity in cognitive processes, mice overexpressing Nrg1 type I display a range of neurocognitive deficits reliant upon intact cortical function (Deakin et al. 2009). Moreover, transcriptional profiling of mice overexpressing Nrg1 type I show extensive myelin-related gene expression changes in brains collected at a young age (consistent with Nrg1 type I’s high expression during early postnatal period); however, in older transgenic mice, gene expression changes were limited to genes related to inflammation and immunity (Deakin et al. 2018). Given that we found Nrg1 type I’s expression significantly declines during adulthood into aging, this suggests that Nrg1 type I may regulate the neuroimmune system in the mature brain. Surprisingly, we found that expression of Nrg1 type III was relatively stable across the entire postnatal lifespan. Given that both overexpression and knockdown of Nrg1 type III result in impairments in sensorimotor gating, social cognition, learning, and memory (Chen et al. 2008; Olaya et al. 2018), we posit that tightly regulated expression of Nrg1 type III is required across the entire lifespan to ensure typical cortical development, maturation, and function.

Next, our data lend support for Nrg2 as an important and underappreciated member of the Neuregulin family that may have particular significance for postnatal brain function. Our data are in line with a previous immunohistochemical study showing that Nrg2 expression increases over the first postnatal week (Longart et al. 2004). Interestingly, we find that while Nrg2 expression does not show a correlation with aging across the lifespan, the expression profile is highly dynamic in early postnatal stages, increasing dramatically in the first postnatal week before decreasing to initial levels and stabilizing by adulthood. Of note, recent studies indicate that Nrg2-ErbB4 signaling alters NMDA receptor function, specifically GluN2B containing NMDARs (Vullhorst et al. 2015; Yan et al. 2018). Our finding of Nrg2 expression peaking at PND7 coincides with the developmental switching of NMDAR subunits from GluN2B to GluN2a in early postnatal life (Liu et al. 2004), suggesting that perinatal Nrg2 may be essential for this switch and thus the early shaping of cortical microcircuits through regulation of NMDA kinetics, and thus why Nrg2 hypomorphic mice display PFC-mediated working memory impairments (Yan et al. 2018).

Convergent with previous studies, we identified that Nrg3 expression was abundant in the postnatal rodent cortex (Anton et al. 2004; Longart et al. 2004), and confirm recent reports demonstrating in whole rat brain extracts Nrg3 expression steadily increases into adulthood (Rahman et al. 2019). However, we extend these findings to show that specifically within the murine PFC, Nrg3 expression increases gradually across postnatal lifestages into aging. Nrg3 has critical roles during embryonic and perinatal cortical development regulating cortical laminar patterning (Assimacopoulos et al. 2003; Flames et al. 2004; Bartolini et al. 2017), and our prior research indicates perinatal disruption of Nrg3 expression has life-long neurobehavioral consequences (Paterson and Law 2014). However, Nrg3 is also thought to perform additional roles including in oligodendrocyte survival (Carteron et al. 2006), and given that ongoing maturation of white matter structures has been identified in the mouse brain into the second month of life (Baloch et al. 2009), we propose that this may be a principal function of Nrg3 in the murine brain throughout adulthood and into aging. Additionally, temporally regulated manipulation of Nrg3 in the adult mouse PFC impacts impulsive behaviors, emphasizing the importance of Nrg3 signaling in the mature mouse PFC (Loos et al. 2014).

Significantly, we also identify that Nrg3 and its cognate receptor ErbB4 (Zhang et al. 1997) have overlapping expression profiles in the mouse PFC across the lifespan. Many studies have demonstrated that ErbB4 is essential for synapse formation and maturation (Fazzari et al. 2010). ErbB4 expression in the postnatal mouse cortex is largely restricted to GABAergic interneurons and enriched within parvalbumin-positive interneurons (Neddens et al. 2011). Interestingly, we identify that while there is an overall positive correlation of cortical ErbB4 expression with postnatal age, this expression increase is highly significant between PNDs 7 and 35, strikingly similar to the peak expression profile of parvalbumin in mouse PFC (Ueda et al. 2015), Highlighting the importance of ErbB4 postnatally in inhibitory cortical circuitry (Rico and Marin 2011). Of note ErbB2, which we have identified as having an opposing lifespan expression trajectory to ErbB4, has been shown to be important in pyramidal cell regulation (Barros et al. 2009). ErbB3 expression meanwhile presented the most dynamic expression profile of the ErbB genes dramatically increasing in the second postnatal week. Convergent to our findings, a hybridization-based study identified that in mouse brain ErbB3 signal emerged after PND 7 and in the adult brain displayed a high-density signal predominantly within white matter tracts supportive of ErbB3’s role in CNS myelination and postnatal glial function (Fox and Kornblum 2005). Moreover, in the mouse PFC, genes encoding myelin components dramatically increase in expression between PND7 and PND21 (Ueda et al. 2015) mirroring our findings of ErbB3.

We show that ErbB4 splice isoform expression changes across the murine lifespan. Our findings that JMa transcripts show an opposing expression trajectory to that of JMb transcripts is in agreement with a previous report that JMa and JMb receptors display antagonistic cellular responses (Sundvall et al. 2010) and lends further support that individual ErbB4 isoforms display differing cellular properties in neurons (Gambarotta et al. 2004; Fregnan et al. 2014; Fornasari et al. 2016). Our findings that JMb and CYT2 transcripts increase with aging in the murine PFC align with previous studies in human and rodent brain that JMb/CYT2 is the major isoform in the adult brain (Fornasari et al. 2016; Chung et al. 2018; Erben et al. 2018). Interestingly, our finding that JMa/CYT1 transcripts are expressed predominantly in early life is consistent with previous studies implicating these as the disease-associated transcripts in schizophrenia (Law et al. 2007; Chung et al. 2016, 2018), a disorder characterized by alterations in transcripts enriched during early life (Birnbaum et al. 2014). Moreover, our findings suggest that previous findings of the involvement of ErbB4 in aging and age-related disease may be due specifically to JMb/CYT2 dysfunction.

While the function of the long noncoding RNA MIAT has been predominantly studied with regard to its roles in early brain development including in neuronal differentiation and neurogenic commitment and survival (Mercer et al. 2008; Aprea et al. 2013), emerging studies reveal that MIAT expression increases with age in the human subventricular zone (Barry et al. 2015) and mouse hippocampus (Stilling et al. 2014). We extend these findings to show that Miat expression is also increased with aging in the mouse PFC. Moreover, we demonstrate that MIAT expression is positively correlated with that of JMb and CYT2 ErbB4 splice isoforms across the murine lifespan, providing initial support that Miat function is also important in the adult and aging brain that may involve the regulation of JMb/CYT2 function. Our findings both support and provide a temporal context to a previous study which demonstrated that overexpression of MIAT has been associated with increased production of JMb and CYT2 ErbB4 splice isoform transcripts (Barry et al. 2014). While the exact mechanism by which Miat regulates ErbB4 splicing is not fully understood, converging evidence suggests that a binding motif in the 3′ region of Miat results in the sequestering of key spicing factors including splicing factor 1 (SF1) inhibiting their binding potential to ErbB4 transcripts (Tsuiji et al. 2011; Barry et al. 2014; Ishizuka et al. 2014).

Finally, we show that in addition to transcriptional temporal regulation of the Nrg–ErbB network across the murine lifespan, expression levels of multiple members of the network, namely Nrg1, Nrg3, and ErbB4, are also temporally regulated at the level of translated protein products. In keeping with our findings of transcriptional mapping, we identify that protein levels of Nrg3 and ErbB4 display overlapping expression patterns and both RNA and protein levels increase across cortical development. Conversely, Nrg1 protein levels across the lifespan did not correlate with the RNA profiles observed using a Nrg1 “pan” assay. This disparity between protein and RNA expression is likely due to the complexity of Nrg1 splicing, as well as protein levels being impacted by multiple regulatory processes occurring after RNA is transcribed, including protein stability, turnover, and proteolytic processing. Moreover, poor transcript–protein correlation is common, especially in brain tissue (Vogel and Marcotte 2012; Bauernfeind and Babbitt 2017; Moritz et al. 2019). Importantly, the commercial antibodies available for Nrg1 and ErbB4 do not distinguish between isoforms; therefore, comparisons between transcript and protein patterns for splice isoforms are not possible (Table 2).

Table 2.

Summary of cortical expression trajectories of the Nrg-ErbB network across the murine lifespan

Gene	Time frame
	Lifespan PND 1–24M	Early PND 1–7	Pre-pubertal PND 8–35	Post-pubertal PND 42–3M	Adult aging 4M–24M
Nrg1 “pan”	0	0	0	0	0
Nrg1-I	−	−	−	0	−
Nrg1-II	+	0	0	0	0
Nrg1-III	0	0	0	+	0
Nrg2	0	+	+	+	0
Nrg3	+	0	0	0	0
ErbB2	−	+	−	0	0
ErbB3	+	+	+	0	+
ErbB4	+	0	+	0	0
JMa	−	−	+	−	0
JMb	+	0	0	0	0
CYT1	−	−	−	0	+
CYT2	+	0	+	0	0
Miat	+	0	+	0	0

0 = no correlation, + = significantly positive correlation, and − = significantly negative correlation between age and cortical gene expression of given transcript within time frame in the mouse PFC.

Together our findings underscore the molecular diversity of the Nrg–ErbB signaling network and emphasize the importance of the complex coordination of all members of the Nrg–ErbB network in the regulation of cortical circuitry development and maintenance. We provide a comprehensive resource of Nrg and ErbB expression profiles that can be utilized both in the context of interpreting previous genetic murine models of disrupted Nrg–ErbB signaling and assisting the development of studies novel models. Our results highlight the importance of consideration of splice isoforms, especially in the cases of Nrg1 and ErbB4 when developing murine models, and support that cortical circuitries could be differentially disrupted dependent upon the timing of Nrg–ErbB perturbation.

Funding

Developmental Psychobiology Research Group Endowment at the University of Colorado Anschutz Medical Campus (to C.P.); National Institutes of Mental Health (Award Number R01MH103716 to A.J.L.), http://grantome.com/grant/NIH/R01-MH103716-05. Access to raw data is available upon request.

Notes

We would like to thank Prof. Susan Mikulich, Director of the Biostatistics Core, Department of Psychiatry, University of Colorado Anschutz Medical Campus for their assistance in statistical analyses. Conflict of Interest: None declared.