Histone deacetylase-3: Friend and foe of the brain

Introduction

Gene transcription is dependent on the state of chromatin, which is determined, in large part, on posttranslational modifications of residues within the tails of histones, the protein component of nucleosomes. ¹ By regulating the associations of histones with other proteins and with DNA, these modifications determine whether chromatin is in a transcriptionally active or an inactive state. Over the past two decades, it has become clear that these chromatin-altering modifications may not only be dynamically regulated by environmental factors and experiences, but can also be heritable. Such regulation of gene expression, which is controlled not by the genetic “blueprint” but by histone modification-driven changes to chromatin is referred to as epigenetic regulation. Among the best studied of epigenetic modifications is the acetylation and deacetylation of lysine residues. ² Acetylation neutralizes the positive charge of lysine residues within histone tails resulting in a weakening of the electrostatic interaction between histones and the negatively charged DNA with which they associate. This causes the relaxation of chromatin permitting the binding of DNA polymerase and transcription factors promoting transcription. On the other hand, deacetylation of histone tails increases their interaction with DNA resulting in chromatin compacting and, consequently, a reduction of transcription. In addition to affecting chromatin structure, the acetylation and deacetylation of histone tails also regulate the recruitment of transcriptional regulatory factors to DNA. ² Depending on the proteins that are recruited, and complexes assembled to the modified histones, chromatin can be remodeled, and transcription can be activated or repressed.

Acetylation and deacetylation of histones are regulated by two families of enzymes – histone acetyltransferases (HATS) and histone deacetylases (HDACs). The HDAC family of deacetylases is subdivided into four groups.^3,4 Class I HDACs includes HDAC1, HDAC2, HDAC3, and HDAC8. These are ubiquitously expressed nuclear proteins, except for HDAC3 which is both nuclear and cytoplasmic. Class II HDACs are expressed more selectively and subdivided into two subgroups, with Class IIa consisting of HDAC4, HDAC5, HDAC7, and HDAC9, whereas Class IIb is made up of HDAC6 and HDAC10. Class II HDACs can shuttle between the cytoplasm and the nucleus based on their phosphorylation status, except for HDAC6 which is cytoplasmic. Class III HDACs, generally called sirtuins, include SIRT1—7. SIRTs can be nuclear, cytoplasmic, or mitochondrial. Finally, HDAC11, localized predominantly in the nucleus, is the sole member of the Class IV HDAC family. Class I, II, and IV HDACs are commonly referred to as classical HDACs. Classical HDACs share structural similarity in their catalytic domains and require zinc as cofactor. In contrast, the structurally distinct sirtuins are NAD⁺-dependent enzymes.^3,4

Although histones were their first identified substrates, HDACS (and HATs) regulate the acetylation status of a last number of non-histone proteins in the nucleus, cytoplasm, and mitochondria and in doing so, regulate their structural conformation, subcellular localization, stability, interactions with other proteins and, in the case of transcription factors, DNA-binding activity.^3,4 Through such actions, HDACs regulate diverse physiological and pathophysiological processes. Much of the information on the involvement of HDACs in various biological processes has been obtained through the utilization of pharmacological inhibitors. The most commonly used inhibitors of classical HDACs belong to five groups: hydroximates, benzamides, cyclic tetrapeptides, aliphatic acids, and electrophilic ketones. A major limitation of these inhibitors, however, is that they are generally non-selective with respect to the different HDAC proteins at the doses at which they are typically used.

Epigenetic mechanisms, such as histone acetylation/deacetylation, regulate brain development and hence important brain-related functions including activity-dependent transcription, synaptic plasticity, and learning and memory formation.^5–10 Not surprisingly, impaired epigenetic regulation affects proper functioning of the brain and has been implicated in in the etiology of complex brain disorders such as depression, autism schizophrenia, and bipolar disorder.^11–13 Epigenetic mechanisms also play a critical role in brain activities during normal adulthood and in the progressive cognitive decline associated with aging.^14–16 This review describes the actions of HDACs in the brain, focusing on a single member of the family, the Class I HDAC, HDAC3. As described below, HDAC3 plays both helpful and damaging roles in the brain. Specifically, HDAC3 is critical to the proper brain development but can also promote neurodegeneration in the adult brain. Results from several recent studies also implicate HDACs in the etiology of neuropsychiatric disorders. While this review focuses on HDAC3, the reader is referred to other recent reviews for information on the role of other Class I HDACs in brain development and degeneration.^5,6,17

HDAC3 is required for proper development of the brain

Conventional knockout of Hdac3 results in early embryonic lethality. ¹⁸ Norwood et al. ¹⁹ were the first to describe the effects of conditional genetic ablation of Hdac3 on the development of the CNS (Table 1). Mice in which Hdac3 was deleted specifically in neural progenitor cells (NPCs) survived for less than 24 h after birth. ¹⁹ Examination of the cortex revealed severe lamination defects in the cortex along with loss of neurons in the superficial layers, which are the last cortical layers to be formed. The dentate gyrus was poorly formed. Similarly, the cerebellum of these Hdac3-deficient mice was underdeveloped and largely devoid of foliation. Purkinje neurons of the cerebellum were improperly positioned. ¹⁹ During brain development, the production of neurons (neurogenesis) precedes the production of glial cell generation (gliogenesis). Within the gliogenic phase, astrocytes are generated first and after a cell fate switch, NPCs differentiate to produce oligodendrocytes.^20–22 Examination of the status of glial cells in the cortex of Hdac3 mutant mice on the day of birth suggested a defect in glial cell fate determination resulting in the overproduction of astrocytes at the cost of oligodendrocyte precursor cells. ¹⁹ Mice in which Hdac3 was selectively ablated in forebrain neurons displayed progressive motor deficits and died within six weeks demonstrating an essential role in neuronal function. Although Hdac3 ablated in developing neurons in these mice, the expression of GFAP, a marker of astrocytes, was highly elevated in the cortex indicating overproduction of astrocytes. ¹⁹ The disruption of glial cell fate determination was confirmed and extended by Zhang et al. ²³ (Table 1). These authors found that HDAC3 antagonized astrogliogenesis by inhibiting the acetylation of STAT3, a transcription factor necessary for astrogenesis, while activating the oligodendrocyte lineage. ²³ Deletion of Hdac3 in oligodendrocyte progenitors resulted in reduced production of oligodendrocytes demonstrating the requirement for HDAC3 for the generation of this glial cell type in proper numbers. ²³ Oligodendrocytes that were produced were severely deficient in myelin. ²³ In striking contrast to its role in oligodendrocyte development, HDAC3 inhibits the maturation and myelination of Schwann cells, the cellular equivalent of oligodendrocytes in the peripheral nervous system (PNS). ²⁴ Deletion of Hdac3 in the Schwann cell lineage resulted in premature myelination ending up in their hyper-myelination of Schwann cells. Moreover, Hdac3 deletion stimulated re-myelination after nerve transection. An inhibitory role for HDAC3 in myelin formation as well as its regeneration after nerve injury was confirmed using pharmacological inhibitors of HDAC3. ²⁴ Nerve conductivity after injury was enhanced in mice both by HDAC3 deletion and following administration of HDAC3 inhibitors. ²⁴ In vitro experiments using pharmacological inhibition and genetic knockdown confirmed that HDAC3 inhibits Schwann cell differentiation and maturation and revealed that the inhibitory action of HDAC3 was mediated by inhibition of myelination genes through the antagonization of the NRG1-PI 3 kinase-Akt signaling pathway. ²⁴ Although some of the findings of He et al. were confirmed by another group, other findings were not. Specifically, Rosenberg et al. ²⁵ found that HDAC3 is not required for the production of myelinated Schwann cells during development, but plays a critical role in maintaining a proper level of myelination through adulthood. Deletion of Hdac3 led to hypertrophy and hyper-myelination of the Schwann cells during adulthood resulting in progressive neuropathy. ²⁵ The different conclusion with regard to the generation of properly myelinated Schwann cells is unclear but may be related in part to the utilization of different Cre constructs for Hdac3 ablation in the two studies. While He et al. used 2′,3′-cyclic nucleotide 3′ phosphodiesterase (Cnp) promoter, Rosenberg et al. utilized the P0 promoter, which like the Cnp promoter is also expressed in Schwann cell progenitors.

OPEN IN VIEWER

Table 1.

Roles of HDAC3 in neurodevelopment.

Role	References
Regulation of NPC proliferation and differentiation in the neocortex and hippocampus	26–28
Experience-dependent NPC proliferation in the Xenopus tectum	29
Neocortical lamination, survival of neurons in the superficial layers of the cortex	19
Cerebellar morphogenesis and hippocampal formation	19
Auditory organ morphogenesis in zebrafish	32
Glial cell fate determination, production of oligodendrocytes	19,23–25
Development and myelination of Schwann cells	24,25

Note: Unless otherwise mentioned, all studies listed in the tables were performed in rodents or in vitro systems derived from rodent or human.

As described above, mice in which Hdac3 is deleted in neural progenitors show impaired cerebellar morphogenesis ¹⁹ (Table 1). Another study described that knockdown of Hdac3 in the zebrafish using morpholinos impaired morphogenesis of the auditory system which was accompanied by decreased cell proliferation and increased apoptosis. ³⁰ The effect of HDAC3 in enabling auditory organ morphogenesis was found to be mediated by the stimulation basic FGF/FGF2. Administration of FGF2, a mitotic factor and morphogen, compensated for the lack of HDAC3 and promoted proper auditory organ development. ³⁰ Even in the mouse, FGF2 stimulates proliferation of NPCs in the cortex as well as in other areas including the hippocampus, cerebellum, and spinal cord, regions of the brain which are all affected in mice lacking Hdac3 in NPCs.^31–36 Whether FGF2 mediates the morphogenetic actions of HDAC3 in the mammalian brain has not yet been examined. It is possible, however, that the absence of HDAC3 reduces FGF2-mediated proliferation NPCs that generate late-developing neuronal populations in the cortex (upper-layer neurons), hippocampus (dentate gyrus), and the cerebellum. This would explain the selective impact on neurons in these brain regions.

Other studies have also found that HDAC3 stimulates NPC proliferation, which could then contribute to the regulation morphogenetic actions. Using neural NPCs cultured from the adult mice lacking HDAC3 in the hippocampus, Jiang and Hsieh described that HDAC3 was necessary for cell cycle progression acting at the G2/M boundary. This stimulatory effect on cell cycle progression in these adult was found to be due to the posttranslational stabilization of the mitotic cyclin-dependent kinase, CDK1. ²⁶ The reduction in NPC proliferation was confirmed in mice in which Hdac3 was inducibly deleted in the hippocampus. ²⁶ A similar proliferation defect leading to apoptotic death or increased differentiation was described in embryonic cortical NPCs cultured from mice lacking Hdac3 and normal NPCs in which Hdac3 was knocked down.^27,28 A role for HDAC3 in the stimulation of NPC proliferation was also reported in studies of the zebrafish retina suggesting that this proliferative activity of HDAC3 is conserved. ²⁹

In sum, based on the defective migration in the cortex, the loss of later-born neurons in the cortex, the underdeveloped cerebellum and mispositioning of Purkinje neurons, ¹⁹ the impaired glial cell fate determination,^19,23 and this regulation of myelination of oligodendrocytes and Schwann cell,^23–25 it is clear that HDAC3 is a critical mediator of brain development regulating multiple neurodevelopmental processes possibly through distinct signaling partners and mechanisms.

HDAC3 as a key contributor to neurodegenerative disease

The first indication that HDACs contribute to neurodegenerative disease came from a study by Steffan et al. ³⁷ in 2001 which showed that pharmacological inhibition of HDACs suppressed polyglutamine-dependent neurotoxicity in Drosophila (Table 2). Since then a large number of studies have demonstrated the ability of structurally distinct HDAC inhibitors to protect in diverse cell culture, invertebrate, and mammalian models of neurodegenerative conditions.^17,38–40 However, because these pharmacological inhibitors are generally non-selective with respect to the HDAC family, the identity of the specific HDAC(s) that trigger neurodegeneration and that are targeted by the inhibitors to afford protection has been unresolved. A growing and compelling body of evidence points to HDAC3 being a particularly important player in the promotion of neurodegeneration in a variety of disease models. The first evidence of HDAC3 as a promoter of neurodegeneration came from C. Elegans in which it was shown that knockdown Had-3, considered to be the ortholog of mammalian Hdac3, suppressed toxicity resulting from the expression of a human huntingtin (Htt) fragment with an expanded polyglutamine tract (Htt-Q150). ⁴¹ PolyQ-expansion of Htt is the cause of Huntington’s disease (HD), an inherited neurodegenerative disorder characterized by neuronal loss in the striatum and to a lesser extent, the cortex.^42,43 Interestingly, the knockdown of other HDAC genes in C. Elegans caused an enhancement of Htt-Q150 neurotoxicity. This is consistent with the finding that some HDAC proteins, particularly those belonging to the Class IIa family, are neuroprotective in mammals.^44–46 The first compelling report that HDAC3 is neurotoxic in mammalian neurons came from a study conducted in cultured cerebellar granule neurons (CGNs) and cortical neurons in which it was shown that elevating HDAC3 expression resulted in cell death ⁴⁷ (Table 2). Interestingly, this toxicity was selective for neurons and cell lines of neuronal origin, as the viability of other cell types was unaffected by elevated HDAC3. Knockdown of endogenous HDAC3 protected against death expression resulting from oxidative stress, low potassium treatment, and overexpression of polyQ-Htt, indicating that HDAC3 is required for neuronal death in response to diverse neurotoxic stimuli. ⁴⁷

OPEN IN VIEWER

Table 2.

HDAC3 in the promotion of neurodegeneration.

Role	References
Polyglutamine-induced neurotoxicity in Drosophila and C. Elegans	37,41
Low potassium and HCA-induced neuronal death in cultured cerebellar granule and cortical neurons	48–50
Huntington’s disease	48,51–57
Alzheimer’s disease	58 ––61
Parkinson’s disease	62
Friedreich’s ataxia	63–65
Spinocerebellar ataxia-3	66
Ischemic stroke	67–69
Spinal cord injury	70
Retinal ganglion cell degeneration	71,72

Note: Unless otherwise mentioned, all studies listed in the tables were performed in rodents or in vitro systems derived from rodent or human. Not shown in the table but described are the negative effects of HDAC3 in synaptic plasticity, learning and memory.^73–77

Pharmacological studies using more recently developed HDAC3-selective inhibitors confirm its requirement for neurodegeneration in HD. The Thomas lab was the first to demonstrate that administration of the HDAC3/HDAC1-selective inhibitor, HDACi 4 b, improved disease-associated body weight loss, motor dysfunction, and cognitive decline in different models of HD.^51–53 Interestingly, the protective effect was transgenerational in that it was also seen in first filial generation (F1) offspring of HDACi 4 b-administered male HD transgenic mice. ⁵⁴ The transgenerational protection was attributed to changes in the DNA methylation pattern elicited from the administration of the inhibitor resulting in elevated expression of several genes including Gadd45b, Parp, Mbd3, and Rnf4. Protection in transgenic mouse models of HD through HDAC3 inhibition was confirmed in a follow-up study by the same group as well as by another group using a different HDAC inhibitor, RGFP966.^48,55 More recently HDAC3 has been shown to reduce degeneration and improve behavioral performance in a variety of other serious neurodegenerative conditions. This includes Alzheimer’s disease (AD),^58,59,73 Parkinson’s disease (PD), ⁶² Friedreich’s ataxia,^63,64 ischemic stroke,^67–69 and spinal cord injury ⁷⁰ (Table-2). In AD, the beneficial effect of pharmacological inhibition was confirmed using lentivirally expressed shRNA against Hdac3, which reduced spatial memory deficits, decreased amyloid plaque load and Aβ levels, and increased dendritic spine density. ⁷⁸ ShRNA-mediated knockdown of Hdac3 inhibition also reduces lipid peroxidation, and oxidation of protein and nuclei acid in the hippocampus of AD mice. ⁶⁰ Similarly, in optic nerve crush-induced retinal ganglion cells (RGC) degeneration, protection by pharmacological HDAC3 inhibition was confirmed using RGC-specific conditional knockout of Hdac3.

Neuroinflammation, which exacerbates neuronal loss in neurodegenerative diseases as well as other neurological conditions such as stroke and spinal cord injury, is also suppressed by pharmacological HDAC inhibition.^68,70,78 For example, in a rat spinal cord injury model, Chen et al. ⁷⁹ described an upregulation of HDAC3 expression in glial cells after injury which was blocked by valproic acid, a pan-HDAC inhibitor. These authors showed that pharmacological inhibition of HDAC3 inhibited the transformation of microglia from the anti-inflammatory M2 phenotype to the pro-inflammatory M1 phenotype, thus reducing neuroinflammation. ⁷⁹ The anti-inflammatory effect was mediated by acetylation of NF-κB, which reduced its inflammatory activity. Additionally, nuclear localization of HDAC3 in neurons was reduced by HDAC3 inhibition. ⁷⁹

Surprisingly in view of the effectiveness of pharmacological inhibitors in HD, hemizygous deletion of Hdac3 in the R6/2 mouse model of HD does not reduce neuropathology or improve behavioral performance. ⁸⁰ One possibility is that, at least in this aggressive mouse model of HD, a 50% reduction in HDAC3 level is insufficient to provide significant benefit. Alternatively, and given that neurotoxicity by HDAC3 requires altered posttranslational modifications, protein–protein interactions, and subcellular localization (see below), these pathological alterations could be enhanced in the Hdac3 hemizygous mice to compensate for the reduction of expression. In other words, although present at reduced levels, the remaining HDAC3 may have enhanced neurotoxic activity.

Although knockdown of Hdac3 is protective in cell culture models of HD and other neurotoxic stimuli, neither Hdac3 mRNA or protein levels are altered in dying neurons. ⁴⁷ A similar lack of alteration of HDAC3 expression has been reported in other cell culture and in vivo models of neurodegeneration.^47,60,71 This suggests that rather than increased expression, the posttranslational modification of existing HDAC3 converts it to a neurotoxic protein. Indeed, the phosphorylation of HDAC3 by GSK3β is necessary for its neurotoxic activity. ⁴⁷ This finding is interesting because increased activity of GSK3β has been described in a variety of neurodegenerative diseases, although exactly how this activation triggers neuronal loss is unclear.^81–84 The findings of Bardai et al. suggest that HDAC3 is a substrate and effector of GSK3β in its neurodegeneration-promoting action. Another kinase that stimulates the neurotoxic activity of HDAC3 through direct phosphorylation is leucine-rich repeat kinase 2 (LRRK2), mutations of which are the major cause of inherited PD. ⁶² This finding raises the possibility that neurodegeneration in PD in patients with LRRK2 mutations could be mediated by HDAC3.

Besides phosphorylation, HDAC3 neurotoxicity requires its interaction with HDAC1, another Class-I HDAC. ⁴⁹ A large body of experimental data describe that HDAC1 and HDAC3 are components of distinct multi-protein complexes.^3,4 Whereas HDAC1 (and HDAC2) are components of the Co-REST (co-repressor for element-1-silencing transcription factor), NuRD (nucleosome remodeling and deacetylation) and Sin3 complexes Kelly and Cowley 2013), HDAC3 associates with either NCOR1 (Nuclear Receptor Corepressor-1) or NCOR2 (also known as silencing mediator of retinoid and thyroid receptors, or SMRT) together with the WD40 protein, TBL1, and GPS2, a signaling protein.^85–87 Consistent with the large number of studies performed in non-neuronal cell types, HDAC1-HDAC3 interaction is not detectable in healthy neurons. ⁴⁹ However, robust interaction is seen in cultured neurons primed to die. ⁴⁹ Robust HDAC1-HDAC3 interaction is also observed in the striatum of HD mice, but not in other brain regions that are not affected in HD. Furthermore, HDAC1-HDAC3 interaction in HD mice coincides with the appearance of neuropathology and behavioral deficits. Demonstrating the requirement for HDAC1 in HDAC3-mediated neuronal death is the finding that shRNA-mediated knockdown of HDAC1 protects against the neurotoxic effect of mut-Htt and other stimuli. It deserves pointing out that RGFP966, the HDAC3 inhibitor shown to be neuroprotective in many models of neurodegenerative disease, also inhibits HDAC1 significantly. ⁵⁵ HDAC1/HDAC3 selective inhibition also protects in a mouse model of Friedreich’s ataxia, a neurological disorder caused by the silencing of the (FXN) frataxin gene.^63–65 In this case, administration of the inhibitors resulted in the upregulation on FXN expression. Likewise, genetic knockdown of either HDAC1 or HDAC3 increases FXN mRNA synthesis in Friedreich’s ataxia cells. Interestingly, the transgenerational protections provided by HDACi 4 b in HD mice are not observed with RGFP966. A difference between these two HDAC3 inhibitors is that while HDACi 4 b also inhibits HDAC1 effectively, RGFP966 is much more selective for HDAC3. ⁷⁴ This suggests that transgenerational protection requires efficient inhibition of the HDAC1-HDAC3 complex.

While expression of HDAC3 does not increase during neuronal death in cell culture models or in mouse models of HD or AD, HDAC1 expression does increase in these models. ⁴⁹ Whether the increase in HDAC1 facilitates the interaction with HDAC3 is not known, but possible. Increased expression of HDAC1 has also been described in the spinal cord in a mouse model nerve injury-induced neuropathic pain. ⁸⁸ In this study, increased HDAC1 expression stimulated the JNK-c-jun signaling pathway resulting in increased phosphorylation of c-jun, a transcription factor involved in promoting neuronal death, and an interaction between c-jun and HDAC1. This interaction was not detectable in the brains of control mice. Pharmacological inhibition using an HDAC1-selective inhibitor protects against spinal cord injury and prevents activation of JNK-c-jun signaling. ⁸⁸ Whether HDAC3 is part of the HDAC1-c-jun complex in the injured spinal cord has not been looked at. Direct evidence that HDAC3 inhibits recovery from nerve injury comes from a study that described that either pharmacological inhibition of HDAC3 or shRNA-mediated knockdown robustly stimulates regeneration of the peripheral axons of dorsal root neurons (DRGs) within damaged sciatic nerve. ⁸⁹ Although the kinase responsible was not identified, the repressive action of HDAC3 on peripheral axonal regeneration depends on its phosphorylation. Axonal regeneration requires the dephosphorylation by protein phosphatase-4 (PP4), which inactivates HDAC3. ⁸⁹ In contrast to the sciatic nerve, axonal regeneration of CNS nerves is known to be poor because of inhibitory signals. It was found that the central axonal branch of the DRG neurons that lie within spinal cord do not regenerate following spinal cord injury because PP4 is not activated and consequently HDAC4 remains phosphorylated inhibiting the regeneration program. ⁸⁹

Bardai et al. ⁵⁶ described that in healthy neurons HDAC3 associates with Htt, which is predominantly a cytoplasmic protein. The HDAC3-Htt interaction is reduced in cultured neurons induced to die likely freeing HDAC3 to translocate to the nucleus. HDAC3-Htt interaction is also reduced in the striatum, and to a lesser degree the cortex, of symptomatic HD mice. ⁵⁶ In contrast HDAC3-Htt interaction was elevated in the cerebellum. This is noteworthy because neurodegeneration in HD is highest in the striatum with a lesser amount in the cortex, whereas the cerebellum is largely unaffected. ⁴³ Interestingly, polyQ-Htt does not bind to HDAC3. However, the presence of polyQ-expanded Htt disrupts the binding between HDAC3 and wild-type Htt. ⁵⁶ This finding along with the correlation between the reduction in HDAC3-Htt interaction and neuropathology suggests that the loss of interaction is causally involved in HD pathogenesis. It may be noted that like HDAC3, conventional knockout of Htt leads to embryonic lethality, whereas conditional deletion in the brain causes abnormalities including neurodegeneration.^90,91 Taken together, these findings indicate that along with its phosphorylation, the replacement of Htt with HDAC1 as a binding partner converts HDAC3 to a neurotoxic protein.

Whereas HDAC1 is normally nuclear in neurons and most other cell types, HDAC3 localizes to both the nucleus and the cytoplasm. ⁹² This would suggest that the interaction with HDAC1 involves HDAC3 that normally localizes to the nucleus and/or cytoplasmic HDAC3 that translocates to the nucleus in response to neurotoxic stimuli. Evidence from studies that have analyzed subcellular localization of HDAC3 indicate that its neurotoxicity is mediated in the nucleus as a result of translocation from the cytoplasm.^57,71,72,78 For example, in an in vivo model of retinal ganglion cell (RGC) degeneration triggered by excitotoxicity, the total level of HDAC3 was unchanged through the degenerative process, although nuclear localization of HDAC3 was significantly elevated concomitant with neuronal loss. ⁷¹ This translocation is normally blocked by its cleavage within the C-terminus region by calpain1/calpain2, which results in the retention of HDAC3 in the cytoplasm. A similar nuclear translocation of HDAC3 in hippocampal neurons, coincident with neuropathology, was described in the APP/PS1 transgenic mouse model of AD. ⁶⁰ It is likely that translocation of HDAC3 may by itself be insufficient for neuronal death. One study described that c-fos, an immediate-early gene regulated by neuronal activity, interacts with HDAC3 and can inhibit its neurotoxic action. ⁵⁰ Under conditions of neuronal death, c-fos expression is reduced. ⁵⁰ Such a reduction would permit HDAC3 to interact with HDAC1 unhindered by c-fos. C-fos expression is also reduced in the striatum of HD mice. ⁵⁰ Interestingly, the interaction of c-fos with HDAC3 does not require its transcriptional activity but involves a 20 amino acid region at the C-terminus of the protein. Overexpression of this HDAC3-interacting motif of c-fos is sufficient to protect against HDAC3 toxicity in cultured neurons. ⁵⁰

Two other that HDAC3 interacts with is ataxin-3 and ataxin-7, abnormal polyglutamine expansion of which causes spinocerebellar ataxia-7 (SCA3) and SCA7, respectively.^66,93 In contrast to its inability to interact with polyQ-Htt however, HDAC3 interacts similarly with normal and mutant ataxin-3 and ataxin-7.^66,93 Although stabilizing both wild-type and mutant ataxin-7, HDAC3 caused an accumulation of mutant ataxin-7 in the nucleus and enhanced its neurotoxicity. ⁶⁶ While interacting with HDAC3, ataxin-3 also regulates HDAC3 expression. Mouse embryonic fibroblasts from mice ataxin-3 knockout mice show reduced expression of HDAC3 as well as NCoR1, with which it complexes. ⁹⁴ In this study, it was shown that HDAC3 bound and repressed the promoter of the Eph receptor A3(Efna3) gene, a key component of Eph-ephrin signaling. The reduction of HDAC3 in ataxin-3 knockout mice leads to the de-repression of Efna gene transcription leading to a robust increase in its expression which likely alters Eph-Ephrin signaling pathways. Besides playing a key role in brain development, Eph-Ephrin signaling regulates neurodegeneration. ⁹⁵

How does nuclear HDAC3 (likely is association with HDAC1) promote neurodegeneration? Given that both HDAC1 and HDAC3 are transcriptional repressors, it is likely that neurotoxicity is triggered by the repression of genes which have critical functions, including the promotion of neuronal survival. Several potential targets of HDAC3 were recently identified using ChIP-Seq in cultured cerebellar granule neurons. ⁹⁶ Among these were Bdnf and Npas4 genes. ⁹⁶ Conventional ChIP assays confirmed that the interaction of HDAC3 with the Bdnf and Npas4 gene promoters was greatly increased in neurons primed to die. ⁹⁶ Binding of HDAC3 represses the promoters of these two genes. Whereas overexpression of HDAC3 inhibits Bdnf and Npas4 expression, treatment of cells with RGFP966 stimulates expression confirming that these two genes are transcriptional targets of HDAC3-mediated repression. ⁹⁶ The neuroprotective effects of BDNF are well-established.^97–103 A number of studies have found Npas4, best known as a gene expressed specifically in neurons and in response to neuronal activity, also promotes neuroprotection.^104–108 Using mass spectrometric-based proteomic analyses, Qu et al. identified several proteins that were deregulated by HDAC3 expression in cultured cerebellar granule neurons. Among these were Nptx1 (neuronal pentraxin-1), which was upregulated and Hip1r, and Hdac9, which were repressed by HDAC3. The deregulation of these genes by HDAC3 expression was confirmed by standard ChIP assays and by qPCR. ¹⁰⁹ Interestingly, NPTX1 is involved in promoting neuronal death in different models of human neurodegenerative conditions, including in in vivo models of hypoxic-ischemic brain injury and a cell culture model of AD.^110–119 In contrast, HIP1R is involved in dendritic and synaptic development, whereas an alternatively spliced form of HDRP is neuroprotective.^45,120,121

Using ChIP assays to probe specific promoters bound by HDAC3 in a cell culture model of ischemic stroke, Hsp1a, Bcl2l1, and Prdx2 were found to be targets of HDAC3-mediated transcriptional repression. ⁶⁹ Both treatments with RGFP966 or pre-conditioning, which also has a strong protective effect against subsequent ischemic insult, reduced recruitment of HDAC3 to the promoters of these genes and increased their expression. On the other hand, pharmacological inhibition of calpains promoted nuclear localization and reduced the protective effect of pre-conditioning. ⁶⁹

A recent study described that the protective effect of enriched environment (EE) against the synaptotoxic effect of oligomeric Aβ protein was mediated by the suppression of HDAC3 expression. ⁶¹ This suppressive effect involved an upregulation of miRNA-132. Overexpressing miR-132 or injecting an HDAC3 inhibitor into mice mimicked the protective effect of EE. ⁶¹ Another recently published study conducted in mice described that upregulation of HDAC3 expression is involved in the cognitive decline and neurodegenerative changes caused by diabetes mellitus. ¹²²

HDAC3 as a contributor to memory impairment and neurodevelopmental and neuropsychiatric disorders

HDAC3 negatively regulates memory formation in the mature brain. This has been demonstrated by focal deletion of Hdac3 in the hippocampus of adult mice as well as by the administration of RGFP966.^75,76 The negative regulation of memory was described to be due to the transcriptional repression of the immediate-early genes (IEGs) c-fos and Nr4a2, both of which are known to be involved in memory formation and promoting synaptic plasticity.^123–126 HDAC3 associates with the promoters of the c-fos and Nr4a2 gene promoters and deletion of Hdac3 increases the expression of these genes. ⁷⁶ While ablation of Hdac3 in hippocampal neurons improves memory, another group described that focal ablation of Hdac3 in neurons within the lateral hypothalamus, a part that communicates with the hippocampus to regulate memory, causes learning and memory impairment indicating that HDAC3 can also facilitate memory formation. ⁷⁷ Interestingly, it was found that the enhancement of memory by HDAC3 requires its interaction with NCOR1/SMRT but that its deacetylase activity was not necessary. ⁷⁷ Recent studies have demonstrated that many functions of HDAC3 outside of the nervous, particularly during development, are independent of its enzymatic activity. ¹²⁷ Given this, it remains to be rigorously determined whether the neurodegeneration-promoting action of HDAC3 is dependent on its enzymatic activity.

HDAC3 has also been implicated in different neurodevelopmental and neuropsychiatric disorders. One of these disorders is schizophrenia. Several studies have shown that the expression of dysbindin-1, a schizophrenia-susceptibility protein that can localize to both the nucleus and cytoplasm, is reduced in the prefrontal cortex and hippocampus of schizophrenia patients.^128–131 Schizophrenia-like symptoms and cognitive impairment have also been described in dysbindin-1-deficient mice.^132–135 HDAC3 interacts with two of the three isoforms of dysbindin-1 in human neuroblastoma cells and in mouse brain. ¹³⁶ These authors showed that dysbindin-1B expression was increased in the nucleus in the presence of HDAC3, while the phosphorylation of HDAC3 increased in the presence of dysbindin-1B without a change in expression. ¹³⁶

Although there is no indication that the expression of HDAC3 is deregulated in rodent models of schizophrenia or in patients with the disorder, HDAC1 expression is substantially higher in the prefrontal cortex (PFC) in mouse models of schizophrenia.^137,138 HDAC1 expression is also increased in the PFC and hippocampus of patients with schizophrenia.^137,139,140 Whether this increase is followed by interaction with HDAC3, as observed in models of neurodegeneration, has not been specifically examined. The well-documented alterations in gene expression in the brains of patients with schizophrenia would be consistent with such an interaction.

In developing NPCs, HDAC3 has been found to colocalize and associate with Ankrd11, mutations or deletion of which results in cognitive dysfunction and autism spectrum disorder in humans.^141–143 Like HDAC3, Ankrd11 is a transcriptional co-repressor and can localize to both the nucleus and cytoplasm and acts as a repressor of transcription.^144,145 Also like HDAC3, Ankrd11 is essential for brain development and function. ¹⁴⁶ In the embryonic mouse cortex, Ankrd11 promotes proliferation of neural progenitors a function that involves association with HDAC3. ¹⁴⁶ This raises the possibility that the mutations/deletion of Ankrd11 affects the functioning of HDAC3 in brain development resulting in cognitive impairment and autism. Consistent with such a possibility, HDAC3 also interacts with TBL1XR1, a component of the NCOR1 complex and a protein that also causes cognitive impairment and autism when mutated.^147–149 HDAC3 may also contribute to depressive disorder. At least two studies utilizing distinct anti-depressants described that these drugs downregulate HDAC3 activity.^150,151 In one of these studies, the reduced HDAC3 activity resulted in increased expression of NR2B, a subunit of the NMDA receptor. ¹⁵¹ However, another study found that Hdac3 was one of many genes that was upregulated following treatment with a subset of anti-depressants. ¹⁵²

HDAC3 in aging

Gene expression is reduced in the brain during normal aging. Although the precise reasons for this are not known, the adoption a more repressive chromatin state is likely to be an important factor. For example, it is known that learning and memory require new gene expression, a requirement that would not be adequately met in the aging brain. It has been suggested that this impairment results from the epigenetic modification of chromatin in aging neurons. Indeed, work done by Kwapis et al. ¹⁵³ has shown that this is the case and that HDAC3 plays a key role in driving the decline in learning and memory. Specifically, it was shown that deletion of Hdac3 in the dorsal hippocampus of mice ameliorates age-related impairments in both long-term memory and synaptic plasticity. The findings were confirmed using a dominant-negative form of HDAC3 to disrupt normal HDAC3 activity. Furthermore, deletion of Hdac3 reversed age-related reduction of LTP in hippocampal slices. Interestingly, rather than restoring the gene expression profile of the young brain, deleting HDAC3 in the aging brain prevented the reduction in expression of a few key genes that are critically important for long-term memory formation. One of these genes is circadian clock gene Period1 (Per1) gene. HDAC3 binds to the promoter of the Per1 gene repressing its transcription. ¹⁵³ The importance of Per1 for memory formation in the younger brain has been documented.^154,155 Another gene that is transcriptionally repressed by HDAC3 in the aged hippocampus is Nr4a2, which as described above, is also targeted by HDAC3 in the non-aged adult brain. ¹⁵⁶ Overexpression of Nr4a2 in the dorsal hippocampus ameliorated age-related memory impairment memory. Pharmacological inhibition of Class I HDACs, which includes HDAC3, stimulates Nr4a expression and enhanced long-term memory.^157,158 This enhancement by HDAC inhibition was prevented by blocking Nr4a signaling supporting the idea that HDAC3 impairs long-term memory by inhibiting Nr4a expression. ¹⁵⁷

Conclusions

HDAC3 is necessary for the regulation of critical events in the development of the mammalian CNS (Figure 1). The absence or reduction of HDAC3 levels results in disrupted regulation of the proliferation of NPCs, mislocalization of neurons, abnormal cell death, deregulated production of astrocytes and oligodendrocytes, and morphogenetic abnormalities in the brain. Additionally, emerging evidence suggests that HDAC3 dysfunction can cause memory impairment and likely contribute to neurodevelopmental and neuropsychiatric disorders in humans (Figure 1). Compelling evidence from experimental models points to HDAC3 being a key player in promoting neurodegeneration in the mature brain (Figure 1). HDAC3 also inhibits regeneration after nerve injury. The detrimental effects of HDAC3 on brain function and health are caused not by changes in its expression, but by posttranslational modifications, altered subcellular localization, and abnormal protein–protein interactions. Therefore, a better understanding of how HDAC3 modification, subcellular localization, and interaction with other proteins is regulated could provide the knowledge necessary for developing therapeutic approaches for nerve injury and neurodegenerative disorders.

OPEN IN VIEWER

Figure 1.

Figure depicts stimulatory and inhibitory actions of HDAC3 during brain development, on learning and memory and in neurodegenerative diseases. NPC: neural progenitor cell; AD: Alzheimer’s disease, HD: Huntington’s disease; PD: Parkinson’s disease. (A color version of this figure is available in the online journal.)