Background
p23 (also termed TMP21, p24c, or p24δ
1) belongs to the p24 family of type-I transmembrane proteins, which predominantly localize to cis-Golgi and coated protein complex I (COPI)-coated vesicles. The mammalian p24 family contains 10 proteins that can be grouped into four subfamilies, termed p24α, β, γ, and δ [
1,
2], whose main function is to regulate anterograde and retrograde transport in the early secretory pathway between the endoplasmic reticulum (ER) and Golgi apparatus [
3‐
7]. Eight of the p24 family proteins, including p23, are ubiquitously expressed in mouse tissues. p23/p24δ
1 is the only member of the p24δ subfamily in vertebrates, with the exception of an amphibian-specific p24δ
2 [
2]. Embryos with targeted deletion of
Tmed10 (the gene that encodes p23) die at E4.5 prior to implantation, demonstrating that p23 function is essential for mouse embryonic development [
8].
Like other p24 proteins, p23 has four recognizable domains, a Golgi dynamics domain at the N-terminus followed by a coiled-coil domain, a transmembrane domain, and a short cytoplasmic tail. The role of p23 in the budding process of COPI vesicle from the cis Golgi membrane is well characterized [
9]. In addition, p24 family proteins are involved in the export of glycosylphosphatidylinositol anchored proteins, and likely other select cargo, into COPII vesicles [
7]. Finally, p24 proteins are important for maintenance of the ribbon-like Golgi morphology. When overexpressed in mammalian cells, exogenous as well as endogenous p23 mislocalized to the ER, causing expansion and clustering of smooth ER membranes and fragmentation of Golgi apparatus [
5,
6,
10]. Partial loss of p23 expression in kidney cells of p23
+/- mice also caused Golgi apparatus to dilate [
8]. Thus, both diminution and overexpression of p23 induce changes in Golgi morphology.
A potential new function for p23 in Alzheimer's disease (AD) pathogenesis emerged from a study that identified p23 as a binding partner of γ-secretase [
11]. γ-secretase is a multi-transmembrane enzyme complex made of four essential subunits, presenilin, nicastrin, PEN2, and APH1 [
12]. Sequential cleavage of amyloid precursor protein (APP) by β-site APP cleaving enzyme BACE1 and then by γ-secretase generates β-amyloid (Aβ) peptides, which are deposited in senile plaques found in brains of individuals with AD [
13]. siRNA knockdown of p23 expression in cultured neuronal and non-neuronal cells enhances secretory trafficking of APP as well increased secretion of soluble APP derivatives and Aβ, suggesting that p23 is a negative modulator of Aβ production [
11].
Previously we reported that p23 is widely expressed in major brain areas, and co-localizes in neurons with γ-secretase core subunits presenilin 1 and nicastrin [
14]. Interestingly, the steady-state p23 levels decline during postnatal development in rat and mouse brain, and are also reduced in the brains of individuals with AD. Based on these findings, we suggested that age-related decline in p23 expression may be an intrinsic factor that significantly impacts on APP processing and Aβ burden in the aging nervous system [
14]. To test whether cerebral Aβ levels can be manipulated by neuronal overexpression of p23, we generated multiple lines of transgenic mice that express human p23 under the control of a neuron-specific promoter. Here, we report that neuronal overexpression of p23 causes severe motor dysfunction, growth retardation, infertility, and early death. Although overt neuronal loss was not detected, marked proliferation of astrocytes, activation of microglia, and thinning of myelinated bundless in brainstem of Hup23 mice indicate that neurons overexpressing p23 were indeed subject to crisis or insult. Our data reveal that regulated p23 expression in neurons is critically important for neuronal function, and our study highlights the risks of neuronal overexpression of p23.
Discussion
We report here the generation and characterization of transgenic mice overexpressing human p23 in neurons throughout the central nervous system, under the control of Thy-1.2 promoter. We found that neuronal overexpression of p23 in mice causes a complex set of neurological problems with onset in the first few weeks after birth, progressive motor deficits, post-natal growth retardation, infertility, and premature death. Despite this morbid phenotype, the major brain structures appeared normal in Hup23 mice, and neurons throughout the brain and spinal cord had normal morphology when analyzed by staining with neuronal markers and showed no overt sign of enhanced apoptosis. However, we observed striking astrogliosis, activation of microglia and defects in myelination in brainstem of Hup23 mice. Furthermore, p23 overexpression led to increased burden of protein misfolding in the ER as evidenced by the activation of integrated stress response in brainstem. These results suggest that p23 function in the early secretory pathway is critical for neuronal function and even modest overexpression of this protein is sufficient to cause deleterious cellular dysfunction, which ultimately manifests as a complex neurological phenotype in transgenic mice.
p24 family proteins exist and function as stable oligomeric complexes, and the stability of each p24 protein depends on successful heteromeric assembly with other family members. For example, p23 can exist as monomers, homodimers, and heterodimers with p24, p25, and p27, but not p26 [
34,
35]. Knockdown of p23 expression reduces steady-state levels of p24, p25, p27, and tp24, supporting their existence as heteromers and their dependency on each other to form stable complexes [
8,
36‐
38]. Our findings that even modest neuronal overexpression of p23 (~1.4 fold endogenous levels) elicits a range of phenotype in transgenic mice, and the severity of the phenotype correlated well with the level of p23 overexpression, suggest that a delicate balance exists in the relative abundance of p24 proteins in neurons. In contrast, mice with targeted deletion of one allele of the gene encoding p23, which resulted in 70% loss of steady-state p23 expression, were viable and free of any neurological deficits [
8]. Notably, the steady-state levels of other p23 family proteins were concomitantly reduced in heterozygous p23 animals [
8]. However, in Hup23 mice there was virtually no compensatory upregulation of p24 following overexpression of p23, even when p23 levels were elevated more than 10-fold in the brainstem of line 61 (Figure
2). This observation is in agreement with lack of changes in the steady-state levels of endogenous p24 proteins following transgenic expression of p24δ1 (p23 orthologue) under the control of the proopiomelanocortin promoter in the melanotrope cells of the intermediate pituitary [
39]. Therefore, it is possible that the relative ratio of different p24 family proteins in certain cell types, such as post-mitotic neurons, is more critical than the absolute level of individual p24 proteins.
Previous studies have shown that at least two p24 family proteins are essential for mouse embryonic development: p23 knockout embryos die at E4.5 prior to implantation, and p24 knock out embryos die at E10.5 [
8,
40]. In either case, expression of other p24 family proteins was severely compromised, in agreement with the notion that steady-state levels of p24 proteins are interdependent and complex formation regulates their stability. While it has not been firmly established, it is very likely that lethality of p23 and p24 knockout embryos is the consequence of abnormal protein trafficking due to loss of function of p24 family proteins during early embryogenesis. In this regard, the
Thy-1.2 promoter used to generate Hup23 mice is known to drive transgene expression in subsets of neurons as early as E13, although robust neuronal expression is consistently observed shortly after birth [
15,
17]. Our observation that Hup23 mice are normal at birth and showed no symptoms for the first two weeks after birth indicates that neuronal overexpression of p23 is not deleterious during late stages of embryonic and early post-natal development (Figure
4B). This agrees well with our previous observations that the steady-state level of endogenous p23 in mouse brain is high during embryonic development and on P0, but declines by 50% in the first four weeks after birth [
14]. High-level p23 expression during embryogenesis and the time around birth may also be an indication of a higher demand of select trafficking facilitated by p23 at early stages in life. However, sustained high level neuronal p23 expression during post-natal development in Hup23 mice, especially at time when endogenous p23 expression begins to decline [
14], was deleterious to neuronal functions as evidenced by the complex neurological phenotype observed in Hup23 mice.
Confocal microscopy analysis confirmed that p23 is overexpressed only in neurons in Hup23 mice; proliferating astrocytes and activated microglia do not show evidence of p23 overexpression (Figures
6 and
8). Consistent with immunoblot analyses where we quantified higher steady-state levels of p23 in brainstem of multiple Hup23 lines, we observed overall higher p23 staining in neurons in brainstem as compared with pyramidal neurons in cortex and hippocampus. Moreover, while overexpressed p23 localized to Golgi apparatus in pyramidal neurons in cortex and hippocampus as well as Purkinje neurons in cerebellum, a subset of neurons in brainstem showed p23 staining as clusters that were not labeled by the cis-Golgi marker GM130 (Figure
11). This later observation is highly reminiscent of mislocalization of overexpressed p23 in expanded ER membranes in cultured cells [
5,
6,
10]. We believe that clustering of p23 in neurons localized in brainstem is simply an outcome of the overall higher p23 expression in this brain region. Consequently, we observed adverse cellular reactions more readily in brainstem, such as extensive astrocyte proliferation, microglial activation and partial loss of myelination. How neuronal overexpression instigates these complex cellular reactions is an interesting question. Nevertheless, premature death and the inability to generate Hup23 F2 offspring curtailed our investigation. Mapping with greater precision the neuronal population responsible of the phenotype will require the generation of additional mouse models to achieve p23 overexpression in a Cre-dependent manner by crossing transgenic mice harboring "floxed" p23 expression constructs with specific driver lines (such as the HB9
cre line to direct p23 overexpression selectively in motoneurons).
While the effect of p23 on protein trafficking has been primarily investigated in cells where endogenous p23 expression has been silenced using siRNA, overexpression of p23 in culture cells has also shown to affect protein trafficking [
6,
10]. Since overexpression of p23 displaces endogenous p23 from cis-Golgi into clustered smooth ER membranes, it is likely that normal p23 function in protein trafficking is compromised in cells upon p23 overexpression. Accumulation of misfolded protein is a common feature in several neurodegenerative diseases including Huntington's disease, which is characterized by motor abnormalities [
41]. Furthermore, ER stress has been implicated in pathogenesis of various myelin disorders, including Charcot-Marie-Tooth disease and multiple sclerosis [
42]. Therefore we examined whether UPR is activated following high-level p23 expression in Hup23 animals. Although real-time PCR analysis showed activation of CHOP expression in two lines, immunoblot analysis of all six lines failed to demonstrate steady-state accumulation of UPR associated proteins such as GRP78, ATF4 and ARP. Precisely what effect p23 overexpression has on the neuronal early secretory pathway remains to be investigated. Nevertheless, results of our analysis clearly shows that the complex phenotype observed in Hup23 mice can not be attributed to accumulation of misfolded proteins to the extent that it triggers robust UPR activation.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PG, JR, CGF, KSV and ATP designed experiments, performed the molecular, immunohistochemical, and behavioral characterization of p23 transgenic mice, and participated in data analysis. PG, VPB, SK, and GT participated in image acquisition and analysis of immunostaining data. LAZ participated in characterizing the disease phenotype. PG and GT wrote the manuscript. SK, LAZ, and ATP helped to draft the manuscript. GT conceived of the study, designed experiments, coordinated data analysis, and prepared the final manuscript. All authors read and approved the final manuscript.