Neurosecretory Proteins in Neuroprotection and Neurodevelopment

The intracellular sorting of pro-neuropeptides and neurotrophins to the regulated secretory pathway (RSP) is essential for processing, storage, and release of active proteins and peptides in the neuroendocrine cell. We investigated the sorting of proopiomelanocortin (POMC, also known as pro-ACTH/endorphin), proinsulin, and brain-derived neurotrophic factor (BDNF) to the RSP. Our studies showed that these pro-proteins undergo homotypic oligomerization as they traverse the cell from the site of synthesis in the endoplasmic reticulum (ER) to the trans-Golgi network (TGN). In the TGN, the pro-proteins are sorted into the dense-core granules of the RSP for processing by prohormone convertases and carboxypeptidase E (CPE) and then secreted. We showed that the sorting of prohormones to the RSP occurs by a receptor-mediated mechanism. Site-directed mutagenesis studies identified a 3-D consensus sorting motif consisting of two acidic residues found in POMC, proinsulin, and BDNF. We identified the transmembrane form of CPE as an RSP–sorting receptor that is specific for the sorting signal of these pro-proteins.

We also investigated the role of secretogranin III (SgIII) as a surrogate sorting receptor for membrane CPE in targeting POMC to the RSP. Using RNA interference (siRNA) to knock down SgIII or CPE expression in pituitary AtT20 cells (a pituitary corticotrope tumor cell line), we demonstrated in both cases that POMC secretion via the constitutive secretory pathway was elevated. In double CPE–SgIII knock-down cells, elevated constitutive secretion of POMC and stimulated secretion of ACTH (adrenocorticotropic hormone) were perturbed. Thus, CPE mediates trafficking of POMC to the RSP; SgIII may play a compensatory role for CPE in POMC sorting to the RSP.

Transport of vesicles containing hormone or BDNF to the plasma membrane for activity-dependent secretion is critical for endocrine function and synaptic plasticity. We showed that the cytoplasmic tail of a transmembrane form of CPE in hormone- or BDNF–containing dense-core secretory vesicles plays an important role in their transport to the vesicles' release site. Overexpression of the CPE tail inhibited the movement of BDNF– and POMC/CPE–containing vesicles to the processes in hippocampal neurons and pituitary cells, respectively. The transmembrane CPE tails on the POMC/ACTH and BDNF vesicles interact with dynactin and the microtubule-based motors KIF1A/KIF3A to effect anterograde vesicle movement to the plasma membrane for activity-dependent secretion. Additionally, we showed that another player, snapin, a promoter of vesicle docking, binds directly to the cytoplasmic tail of CPE and connects with the microtubule motor complex, which consists of kinesin-2 and kinesin-3, to mediate post-Golgi anterograde transport of POMC/ACTH vesicles to the process terminals of AtT20 cells for secretion. Knockdown of snapin reduced stimulated ACTH secretion, while protein kinase A (PKA) activation by forskolin significantly increased the interactions of kinesin-2 and kinesin-3 with CPE and levels of ACTH vesicles at the terminus, and enhanced secretion of ACTH in AtT20 cells. Thus, our study uncovered a novel complex, consisting of the CPE–cytoplasmic tail snapin–kinesin-2 and -3, which mediates anterograde transport of ACTH/POMC vesicles to the process terminals for secretion in a PKA–dependent manner in neuroendocrine cells.

Several null and homozygous nonsense mutations in the CPE (also known as neurotrophic factor-alpha1 [NF-alpha1]) gene were identified in seven children and a young adult woman from five different families. They display clinical features such as childhood-onset obesity, type 2 diabetes, intellectual disabilities, hypogonadotrophic hypogonadism, and infantile hypotonia, underscoring the importance of CPE in the normal development and function of the endocrine and nervous systems in humans. To study the physiological role of CPE/NF-alpha1 in vivo, we generated a Cpe knock-out (KO) mouse. The KO mouse exhibited obesity, infertility, and diabetes. Further analysis of Cpe–KO mice in the Morris water maze and by object-preference tests revealed defects in learning and memory and depressive-like behavior in the forced swim test. Electrophysiological measurements showed a defect in the generation of long-term potentiation in hippocampal slices. We discovered that a major cause of the neurological defects in such mice was the loss of CA3 neurons in the hippocampus after weaning stress. Hippocampal neurons in the CA3 region are enriched in CPE. Cpe–KO mice showed normal hippocampal cytoarchitecture at three weeks of age just before weaning, indicating that the defect was not the result of aberrant development. Rather, we hypothesized that the degeneration of the CA3 neurons was likely caused by glucocorticoid-induced epileptic-like neuronal firing of the granule cells in the dentate gyrus, releasing large amounts of glutamate during the weaning-stress paradigm, which includes emotional stress from maternal separation and physical stress from ear tagging and tail clipping for genotyping. The hypothesis was supported by the finding that treatment with carbamazepine, an anti-epileptic drug, prior to weaning prevented the stress-induced degeneration of the CA3 neurons in the Cpe–KO mice. Hence, CPE/NF-alpha1 is important for the survival of CA3 neurons during severe stress. To determine whether the neuro-protective effect of CPE/NF-alpha1 depends on the BDNF–TrkB pathway (TrkB: tropomyosin receptor kinase B), we treated mice with ANA12, a trkB inhibitor. Interestingly, downregulation of the BDNF–TrkB pathway had no detrimental effect on the survival of the CA3 neurons after the weaning-stress paradigm, unlike the Cpe–KO mice, which showed complete degeneration, suggesting that CPE/NF-alpha1 is more critical than BDNF in protecting CA3 neurons from severe stress–induced cell death [Reference 1].

Furthermore, we showed that a mutant mouse expressing an enzymatically inactive form of CPE/NF-alpha1 (E342Q) had a normal hippocampus and learning and memory after the weaning-stress paradigm, indicating that the neuroprotective action of CPE/NF-alpha1 is independent of its enzymatic activity [Reference 1]. We showed that CPE/NF alpha1 (E342Q), either overexpressed or applied externally to cultured hippocampal or cortical neurons, protected the neurons from apoptosis induced by oxidative stress with hydrogen peroxide or glutamate treatment. Likewise, the enzymatically inactive form of CPE/NF-alpha1, applied extracellularly, had the same neuroprotective effect. We thus demonstrated that CPE/NF-alpha1 acts extracellularly as a signaling molecule to mediate neuroprotection. To this end, we showed that ¹²⁵I-CPE/NF-alpha1 bound specifically to the cell surface of HT22 cells, an immortalized hippocampal neuronal cell line, in a saturable manner, suggesting the existence of a receptor. K235a, a Trk family inhibitor, and PD16285, a fibroblast growth-factor receptor (FGFR1-3) inhibitor, did not prevent the neuroprotective action of CPE/NF-alpha1 in hippocampal neurons treated with hydrogen peroxide, suggesting that CPE/NF-alpha1 likely uses a different class of receptors than those of the Trk family or FGFRs. The mechanism of action of CPE/NF-alpha1 in neuroprotection in mouse and rat hippocampal neurons involves the activation of the ERK1/2 (extracellular signal–regulated kinases)–signaling pathway during stress, which then leads to enhanced expression of a pro-survival mitochondrial protein, BCL2, inhibition of caspase 3 activation, and promotion of neuronal survival [Reference 1]. We then screened a human GPCR (G protein–coupled receptor) library using CPE/NF-alpha1 as a ligand, and identified the serotonin receptor 5-HTR1E as a binding partner, an interaction that was confirmed by co-immunoprecipitation and pulldown assays. In collaboration with William Goddard, we carried out molecular-dynamics studies followed by experimental validation, which demonstrated that CPE/NF-alpha1 interacts with 5-HTR1E via four salt bridges stabilized by several hydrogen bonds, and is independent of the serotonin-binding pocket. Immunohistochemistry revealed co-localization of 5-HTR1E and CPE/NF-alpha1 on the surface of hippocampal neurons. Signal transduction studies in HEK293 cells stably transfected with HTR1E showed that HTR1E–CPE/NF-alpha1 interaction activated the ERK1/2–CREB pathway via recruitment of beta-arrestin. In turn, this activated the BCL2 pro-survival pathway to mediate protection against oxidative stress. We then showed that the 5-HTR1E–CPE/NF–alpha1 interaction mediated neuroprotection of human primary neurons against hydrogen peroxide–induced cytotoxicity and glutamate-induced neurotoxicity [Reference 2]. The Goddard team then designed and synthesized small molecules that we showed bound to HTR1E with higher affinity and protected human neurons against oxidative stress with greater potency than did CPE/NF-alpha1. Upon binding to HTR1E, these CPE/NF-alpha1 mimetrics activated the ERK1/2–BCL2 pathway to promote neuronal survival [Reference 3]. We also demonstrated that the CPE/NF-alpha1 mimetrics protected mouse neurons from oxidative stress via activation of the ERK–Bcl2 pathway. Work is now in progress to test the efficacy of these CPE/NF-alpha1 mimetrics in vivo in protecting against neurodegeneration in Alzheimer’s disease and in mild traumatic brain injury mouse models, as observed after CPE/NF-alpha1 gene therapy (see below).

Examination of the pathway during stress in vivo revealed that, after mild chronic restraint stress (CRS) for one hour per day for seven days, mice showed significantly elevated levels of CPE/NF-alpha1 mRNA and protein, as well of as the anti-apoptotic protein Bcl2, in the hippocampus. In situ hybridization studies indicated particularly elevated CPE/NF-alpha1 mRNA levels in the CA3 region but no gross neuronal cell death after mild CRS. Furthermore, primary hippocampal neurons in culture showed elevated CPE/NF-alpha1 and Bcl2 expression and a decline in Bax, a pro-apoptotic protein, after dexamethasone (a synthetic glucocorticoid) treatment. The up-regulation was mediated by glucocorticoid binding to glucocorticoid-regulatory element (GRE) sites on the promoter of the Cpe gene. Thus, during mild CRS, when glucocorticoid is released, CPE/NF-alpha1 and Bcl2 expression are coordinately up-regulated to mediate neuroprotection of hippocampal neurons. The importance of CPE as a neuroprotective agent was demonstrated by the absence of an increase in Bcl2 in the hippocampus of Cpe–KO mice after CRS and degeneration of the CA3 neurons. Furthermore, CRS also elevated expression of the signaling protein FGF2 (fibroblast growth factor 2). We demonstrated that mouse primary hippocampal neurons treated with CPE/NF-alpha1 increased FGF2 expression. Thus, another action of CPE/NF-alpha1 is to increase FGF2, which has known neuroprotective and neurogenesis effects. In summary, CPE/NF-alpha1 is a critical neurotrophin for protecting neurons against stress-induced cell death via the Erk–Bcl2 signaling pathway and neurogenesis.

The relevance of CPE/NF-alpha1 in neuroprotection in humans was underscored by our studies on a mutation of the CPE gene found in an Alzheimer’s disease (AD) patient. Our search of the GenBank EST database identified a sequence entry from the cortex of an AD patient that had three adenosine inserts in the CPE gene, thereby introducing nine amino acids, including two glutamines, into the mutant protein, herein called CPE-QQ. Overexpression of CPE-QQ into neuroblastoma cells indicated that the mutated protein aggregates with the wild-type (WT) protein in the ER, causing ER (endoplasmic reticulum) stress, reduced Bcl2 levels, and neuronal cell death. We generated transgenic mice overexpressing CPE-QQ and showed that, at 50 weeks, but not at 11 weeks of age, the animals exhibited memory deficits and depressive-like behavior compared with WT mice, but that their spatial learning ability was unimpaired. The CPE-QQ mice were neither obese nor diabetic, as there was some CPE activity in these mice, given that the endogenous WT gene was not deleted. However, they had significantly fewer neurites in the CA3 region and the dentate gyrus of the hippocampus and in the medial prefrontal cortex, indicative of neurodegeneration. Moreover, they exhibited reduced neurogenesis in the subgranular zone and hyperphosphorylation of the microtubule-associated protein tau at ser³⁹⁵, a hallmark of AD. These studies thus identified a human mutation in the CPE gene resulting in expression of a CPE-QQ protein, which caused neurodegeneration and impairment of memory function, as well as depressive-like behavior in a mouse model, linking the gene for the first time with neurodegenerative disease and depression. We therefore explored whether CPE could rescue cognitive dysfunction and memory loss in AD mice. We injected AAV (adeno-associated virus)–CPE bilaterally into the hippocampus of AD mouse model with genetic defects (3xTg-AD), at the pre- and post-symptomatic stages. AD pathology such as memory loss, neurodegeneration, tau hyperphosphorylation, and increase in Abeta 1-42 amyloid peptides did not develop in pre-symptomatic mice and were reversed in post-symptomatic mice [Reference 4]. AAV-CPE treatment also down-regulated the expression of a microglia neuro-inflammatory protein, Card14, which may contribute mechanistically to the lower microglia activation and neuroinflammation compared with control AD mice. Additionally, expression of Plin4, a mitophagy-inhibitor protein, was attenuated [Reference 4]. A proteomic study revealed up- and down- regulation of many proteins after AAV-CPE gene therapy [Reference 5]. Up-regulated proteins included BCL2, PSD95, and Beclin 1 to promote neuroprotection, synaptogenesis and autophagy, respectively, which are defective in AD mice, concomitant with deficits of these proteins. Thus, CPE-NF-alpha1, overexpressed in the hippocampus, targets many regulatory components to prevent the development of AD pathology in AD mice [Reference 5]. Recently, we also showed that injection of AAV-CPE-NF-alpha1 into the cortex followed by mild traumatic brain injury prevented subsequent neurodegeneration. Thus AAV-CPE-NF-alpha1 gene therapy offers a promising new approach for treating neurodegenerative diseases such as mild concussion, especially in children, and Alzheimer's disease in old and in young adults with Down’s syndrome.

Stress induces depression. FGF2 has been reported to be an anti-depressant. We found that prolonged (six hours per day for 21 days) restraint stress reduced CPE/NF-alpha1 and FGF2 in the hippocampus of mice and induced depressive-like behavior. However, after short-term restraint stress (one hour per day for seven days), mice did not show depressive-like behavior despite the elevated corticosterone levels indicative of stress. Moreover, hippocampal CPE/NF-alpha1, FGF2, and doublecortin, a marker for neurogenesis, were elevated in these mice, suggesting that the anti-depressive effects of CPE/NF-alpha1 are mediated, at least in part, through increased neurogenesis. Indeed, we found that exogenously applied CPE/NF-alpha1 could up-regulate FGF2 mRNA and protein expression in cultured hippocampal neurons, indicating that CPE/NF-alpha1 regulates FGF2 expression. CPE/NF-alpha1–KO mice exhibited severely reduced hippocampal FGF2 levels and immature neuron numbers in the subgranular zone. The mice displayed depressive-like behavior, which was rescued by FGF2 administration. Thus, we propose that CPE/NF-alpha1 prevents stress-induced depressive-like behavior by up-regulating hippocampal FGF2 expression, which leads to enhanced neurogenesis and anti-depressive activity. Furthermore, we showed that rosiglitazone, an anti-diabetic drug, can trigger the CPE/NF-alpha1-FGF2 pathway leading to anti-depressive like activity.

Embryonic mouse brains express three forms of CPE/NF-alpha1 mRNA (2.1kb, 1.9kb, and 1.73kb in size), which encode a 53kD WT CPE/NF-alpha1, and two terminal-truncated isoforms of CPE/NF-alpha1-ΔN (47kD and 40kD). The three mRNAs are expressed as early as E8.5 and increase significantly in two waves at E10.5 and postnatal day 1. Interestingly, CPE/NF-alpha1-ΔNs are expressed much less in adult mouse brain. In situ hybridization studies indicate that CPE/NF-alpha1 is expressed primarily in the forebrain in mouse embryos, suggesting that it plays a role in neurodevelopment. We examined the effect of CPE/NF-alpha1 on E13.5–neocortex-derived neurospheres, which contain stem cells and neuro-progenitors. Application of recombinant CPE/NF-alpha1 reduced the number and size of the neuro-spheres formed, suggesting inhibition of proliferation and maintenance of the 'stemness' of the stem cells in the neurospheres. CPE/NF-alpha1 down-regulated the wnt signaling pathway in the neurospheres, leading to reduced levels of beta-catenin, a protein known to enhance proliferation, suggesting that CPE/NF-alpha1’s inhibitory effect on proliferation is brought about by negatively regulating the wnt pathway.

We also carried out differentiation studies using neurospheres from seven-day cultures that were dissociated into single cells and cultured for an additional five days. We observed an increase in astrocytes after CPE/NF-alpha1 treatment, without alteration in the percentage of neuronal and oligodendrocyte populations. We also observed this phenomenon when the cultured embryonic stem cells were treated with a non-enzymatic form of CPE, indicating that the effect was independent of enzymatic activity. Interestingly, dissociated cells from neurospheres derived from Cpe/NFalpha1–KO mouse embryos showed fewer astrocytes but more neurons, which was reversed by CPE/NF-alpha1 application. In vivo, Cpe/NF-alpha1–KO mouse cortex (at postnatal P1, the time of astrocytogenesis) showed about half the astrocyte numbers of those in WT animals, confirming the ex vivo data. Our results suggest a novel role for CPE/NF-alpha1 as an extracellular signal to inhibit proliferation and induce differentiation of neural stem cells into astrocytes, thus playing an important role in neurodevelopment.

Neurite outgrowth is key to the formation of synapses and the neural network during development. We found that CPE/NF-alpha1 prevented Wnt-3a (a canonical Wnt ligand) inhibition of nerve growth factor (NGF)–stimulated neurite outgrowth in PC12 cells, a neuroendocrine cell line derived from a rat pheochromocytoma, and in cortical neurons. Moreover, CPE/NF-alpha1 augmented Wnt-5a–mediated neurite outgrowth. Thus, the interplay between NGF neurite outgrowth, which is inhibited by Wnt-3a, and augmenting neurite outgrowth, which is mediated by the signaling protein Wnt-5a and CPE/NF-alpha1, could play an important role in regulating these positive and negative cues, which are critical for neurodevelopment. Analysis of the brain of 6- to 14-week-old Cpe–KO mice revealed poor dendritic pruning in cortical and hippocampal neurons, which could affect synaptogenesis, resulting in cognitive dysfunction in these mice.

We also studied the function of 40kD CPE/NF-alpha1-ΔN and showed that it is translocated from the cytoplasm into the nucleus of rat embryonic neurons. Overexpression of 40kD CPE/NFalpha1-ΔN in HT22 cells, a hippocampal cell line, resulted in an increase in the expression of IGF–binding protein2 (IGFBP2), death-associated protein (DAP1), and Ephrin 1A mRNAs and proteins (receptor protein tyrosine kinases), which are involved in neuronal proliferation, programmed cell death, and neuronal migration, respectively. We demonstrated that IGFBP2 is involved in proliferation in a CPE/NF-alpha1-ΔN–dependent manner in HT22 and mouse embryonic cortical neurons. Thus, 40kD CPE/NF-alpha1-ΔN functions to regulate expression of genes important in neurodevelopment [Xiao L, Yang X, Loh YP. Front Neurosci 2019;13:243]. To further delineate the role of 40kD CPE/NF-alpha1-ΔN versus WT CPE in neurodevelopment in vivo, we generated a knock-in mouse model expressing CPE/NF-alpha1-ΔN (CPE-DN) but not WT CPE. Although CPE-DN is not expressed beyond post-natal day1 (P1) in mice, it is expressed in adult human brain. We analyzed the neurological behavior of 12 week–old adult mice as well as the brain biochemistry and morphology between 3–12 weeks of age. CPE-DN mice showed deficits in learning in the Morris water maze test, and memory loss in the probe test. However, these mice did not exhibit any memory loss in the novel object recognition test. Nissl staining showed that there was partial degeneration in the hippocampal CA3 region, unlike in CPE–KO mice, which showed complete degeneration of the CA3 region; this could account for the partial deficits in cognitive function observed. However, the most striking morphology was the robustness of doublecortin staining in the dentate gyrus of these mice which was comparable to WT mice, consistent with our in vitro studies that showed that CPE-DN promotes neurogenesis, independent of WT CPE. In contrast, CPE–KO mice showed a deficit in neurogenesis in the dentate gyrus. Our study suggests that CPE-DN plays an important role in neurogenesis not only during development but also in the adult brain. Nevertheless, the newborn neurons that have migrated from the dentate gyrus to different areas of the hippocampus appear not to survive as well in the absence of WT CPE, despite having levels of IGFBP2 and FGF2 comparable to those in WT mice.