Biochemical analyses demonstrated that Thy1.2 Nrg1 mice (included as a reference in the present study) displayed an almost 6-fold overexpression of Ig-NRG1 and 5-fold higher ErbB4 hyperphosphorylation compared to STOP-Nrg1*NEX-Cre mice. These findings suggest that ErbB4 is not a rate limiting factor for Ig-NRG1 signaling in the brain, consistent with analyses of heterozygous NRG1 and ErbB2/ErbB3 mutants in the PNS, which show that partial elimination of NRG1, but not of ErbB2/ErbB3 receptors, causes hypomyelination (Michailov et al., 2004). On the other hand, ErbB4 activity can be downregulated by dephosphorylation mediated by receptor tyrosine phosphatase β (RPTPβ) or tyrosine-protein phosphatase non-receptor type 21 (PTPN21) (Buxbaum et al., 2008; Plani-Lam et al., 2015). Ligand-induced
57 proteolytic cleavage and ectodomain shedding by TACE has also been proposed as a mechanism for ErbB4 downregulation. Subsequently, levels of the resulting intracellular fragment, which contains the tyrosine kinase activity, are controlled by proteasome function (Rio et al., 2000; Vecchi et al., 1996; Vecchi and Carpenter, 1997; Zhou and Carpenter, 2000).
However, since Ig-NRG1 transgenic mice display permanently increased steady-state levels of hyperphosphorylated ErbB4, these mechanisms for downregulation appear not to compensate for Ig-NRG1-mediated hyperactivation. Furthermore, growth factor receptors are internalized upon ligand binding to modulate receptor signaling, in addition to target activated receptors to other intracellular substrates. Nevertheless, ligand-dependent receptor internalization is notably low for ErbB receptors (except ErbB1) (Baulida et al., 1996).
Moreover, hippocampal interneurons predominantly express the TACE-mediated cleavage-resistant JMb ErbB4 receptor isoforms, and thus, maintain a high level of ErbB4 at the cell surface (Longart et al., 2007), which can also explain the persistently higher levels of Ig-NRG1-mediated ErbB4 phosphorylation.
In an attempt to analyze the impact of Ig-NRG1-mediated ErbB4 hyperstimulation on two major signaling pathways downstream of NRG1/ErbB4 signaling, biochemical analyses of MAPK and AKT pathways were performed. The MAPK signaling pathway regulates astrocyte proliferation, neuronal survival, synaptic plasticity, dendritic spine morphology, and cognitive processes such as learning and memory (Thomas and Huganir, 2004; Thongrong et al., 2016; Wu et al., 2001). Interestingly, ErbB4 hyperphosphorylation was identified using an antibody directed against Tyr1284, which is the binding site for the SHC1 adapter protein, involved in MAPK signaling (Schulze et al., 2005). Nevertheless, only pan-neuronal massive overexpression of Ig-NRG1 was sufficient to significantly increase levels of phosphorylated MAPK. In STOP-Nrg1*NEX-Cre mice, although there was a tendency towards increased activation of MAPK, this was not significant, which implies that feedback mechanisms seem to effectively downregulated activation of this signaling pathway.
On the other hand, the level of phosphorylated AKT, a protein acting directly downstream of PI3K, was significantly increased in both conventional and conditional transgenic mice. The PI3K/AKT signaling pathway mediates several processes necessary for the proper function of cortical networks, such as interneuron migration, primary neurite formation and dendritic arborization (Junttila et al., 2000b; Kainulainen et al., 2000; Krivosheya et al., 2008; Polleux et al., 2002). Furthermore, the PI3K/AKT pathway has been implicated in several neurodevelopmental diseases, such as SZ (Emamian, 2012; Emamian et al., 2004; Law et al., 2012; Wang et al., 2017), and an increase of phosphorylated AKT levels in the PFC of schizophrenic patients was recently reported (Hino et al., 2016). Additionally, a genetic association study reported epistasis between SNPs in Nrg1, ErbB4 and AKT1 that may increase genetic liability for schizophrenia (Nicodemus et al., 2010).
Discussion
58 PI3K/AKT signaling is activated by the CYT1 isoform of ErbB4 that binds the PI3K regulatory subunit p85 through its phosphorylated Tyr1056. This interaction leads to the activation of the catalytic subunit of PI3K and subsequent conversion of phosphatidylinositol-4,5-bisphosphate (PI-4,5-P2) to the second messenger phosphatidylinositol-3,4,5-trisphosphate (PI-3,4,5-P3) at the inner plasma membrane. Interaction with these phospholipids recruits AKT to the inner membrane, where it is phosphorylated at Tyr308 and Ser473 by 3'-phosphoinositide-dependent kinases (PDK) 1 and 2, respectively, and subsequently regulates downstream proteins, such as GSK3, Mdm2, and mTOR (Elenius et al., 1999; Fresno Vara et al., 2004;
Schulze et al., 2005). Thus, Tyr1056 in ErbB4 CYT1 together with Tyr308 and Ser473 in AKT are key residues involved in the activation of the AKT signaling pathway, whose higher levels may indicate increase stimulation of this pathway. The antibody used to analyze the levels of phosphorylated AKT was directed against Ser473, and full activation of AKT requires phosphorylation of this residue (Wang et al., 2017). However, as mentioned above, ErbB4 hypersphosphorylation was detected at Tyr1284, not at Tyr 1056. Thus, the extent by which Ig-NRG1-mediated ErbB4 activation can stimulate PI3K/AKT pathway is currently unknown, which also prevents an exact molecular explanation for the similar levels of phosphorylated AKT in conditional and conventional Ig-NRG1 transgenic mice, despite distinct levels of phosphorylated ErbB4. Further biochemical analyses of proteins upstream and downstream AKT, eg. PI3K and GSK3, are required to validate the preferential recruitment of this signaling pathway.
To further address a selective recruitment of the AKT pathway under conditions of adult Ig-NRG1 overexpression, STOP-Nrg1 mice were also bred to a tamoxifen inducible version of the Cre driver line (NEX-CreERT2) (Agarwal et al., 2012). Hippocampi were collected four weeks after the last tamoxifen injection. This approach resulted in reduced levels of overexpression (by approximately 40-50%) compared to STOP-Nrg1*NEX-Cre mice, which could be due, at least in part, to absence of Cre recombinase activity in dentate gyrus granule cells of adult mice (Agarwal et al., 2012; Goebbels et al., 2006). No changes in the levels of phosphorylated MAPK were found in this experimental paradigm, consistent with the idea that only massive overexpression of Ig-NRG1 in Thy1.2 Nrg1 mice is sufficient to stimulate this pathway. In contrast, although at moderate levels, a significant increase of phosphorylated AKT in STOP-Nrg1*NEX-CreERT2 mice was present, consistent with a preferential recruitment of the AKT signaling pathway by Ig-NRG1 overexpression. These findings contrast with similar experiments in which CRD-NRG1 overexpression was induced in adult mice. Here, adult activation of CRD-NRG1 expression led to a preferential stimulation of the MAPK signaling pathway (T. Unterbarnscheidt, personal communication). These results suggest isoform-specific, preferential recruitment of AKT versus MAPK signaling pathways. However, as for mice with pan-neuronal overexpression of Ig-NRG1, transgenic mice with strong
59 expression of a CRD-NRG1 variant that mimics BACE1 processing at the juxtamembrane
“stalk” region, stimulate both signaling pathways (T. Unterbarnscheidt, personal communication). Thus, expression levels of Ig- and CRD-NRG1 above a certain threshold permit hyperstimulation of both signaling pathways, and this contrasts with more physiological relevant overexpression in conditional transgenic mice.
Remarkably, compound heterozygous mutants for AKT and NRG1 showed impaired episodic-like memory and impaired sociability (Huang et al., 2015). In line with this finding, hyperactivation of GSK3β (normally inactivated by AKT) in the brain of mice leads to impaired social interaction that could be ameliorated by chronic lithium (GSK3 inhibitor) treatment (Mines et al., 2010). This suggests that impaired inhibition of GSK3β by AKT may affect social behavior.Furthermore, AKT and GSK3β play a critical role in ethanol-induced suppression of kainate-induced -oscillations in CA3 (Wang et al., 2017), consistent with the idea that activation of this signaling pathway by elevated NRG1/ErbB4 signaling may be involved in impaired -oscillation and subsequent hippocampal-dependent social cognitive deficits.
3.5 Ig-NRG1 accumulates in the somatodendritic, not the presynaptic compartment of