Monoaminergic and Neuropeptidergic Neurons Have Distinct Expression Profiles of Histone Deacetylases

Monoaminergic and neuropeptidergic neurons regulate a wide variety of behaviors, such as feeding, sleep/wakefulness behavior, stress response, addiction, and social behavior. These neurons form neural circuits to integrate different modalities of behavioral and environmental factors, such as stress, maternal care, and feeding conditions. One possible mechanism for integrating environmental factors through the monoaminergic and neuropeptidergic neurons is through the epigenetic regulation of gene expression via altered acetylation of histones. Histone deacetylases (HDACs) play an important role in altering behavior in response to environmental factors. Despite increasing attention and the versatile roles of HDACs in a variety of brain functions and disorders, no reports have detailed the localization of the HDACs in the monoaminergic and neuropeptidergic neurons. Here, we examined the expression profile of the HDAC protein family from HDAC1 to HDAC11 in corticotropin-releasing hormone, oxytocin, vasopressin, agouti-related peptide (AgRP), pro-opiomelanocortin (POMC), orexin, histamine, dopamine, serotonin, and noradrenaline neurons. Immunoreactivities for HDAC1,-2,-3,-5,-6,-7,-9, and -11 were very similar among the monoaminergic and neuropeptidergic neurons, while the HDAC4, -8, and -10 immunoreactivities were clearly different among neuronal groups. HDAC10 expression was found in AgRP neurons, POMC neurons, dopamine neurons and noradrenaline neurons but not in other neuronal groups. HDAC8 immunoreactivity was detected in the cytoplasm of almost all histamine neurons with a pericellular pattern but not in other neuropeptidergic and monoaminergic neurons. Thus, the differential expression of HDACs in monoaminergic and neuropeptidergic neurons may be crucial for the maintenance of biological characteristics and may be altered in response to environmental factors.


Introduction
Monoaminergic and neuropeptidergic neurons regulate a wide variety of behaviors, such as feeding, sleep/wakefulness behavior, stress response, addiction, and social behavior. Feeding behavior is mainly regulated by orexigenic neurons containing neuropeptide Y (NPY) and agouti-related peptide (AgRP) and by anorexigenic neurons containing pro-opiomelanocortin (POMC). Both of these types of neurons are localized in the hypothalamic arcuate nucleus (ARC) [1]. Orexin (also known as hypocretin), which was originally identified as an orexigenic neuropeptide, is expressed in the lateral hypothalamic area (LHA) and plays a crucial role in sleep/wakefulness behavior, reward behavior, addiction, and body weight regulation [2][3][4]. Sleep/wakefulness behavior is also regulated by histamine neurons in the tuberomammillary nucleus (TMN), serotonin neurons in the dorsal raphe (DR), and noradrenaline neurons in the locus coeruleus (LC) [2,5]. Dopamine neurons in the ventral tegmental area (VTA), which is involved in the reward system, also alter feeding behavior, social behavior, and sleep/wakefulness [2,3,5,6]. The hypothalamic paraventricular nucleus (PVN) has neurons that contain corticotropin-releasing hormone (CRH), which constitutes the hypotha-lamic-pituitary-adrenal axis, a neuroendocrine system that controls stress response [3]. The PVN also contains neurons that produce oxytocin and vasopressin, which are important for social behavior [7].
Environmental factors such as stress, maternal care, and feeding conditions modulate the above-mentioned behaviors [8][9][10][11]. For instance, stressed animals decrease their food intake compared with non-stressed animals [8]. Male rats that had experienced neonatal maternal deprivation showed a decrease in total sleep time [11], and suppressed social interaction [12]. Recently, increasing reports suggest that the epigenetic regulation of gene expression plays a crucial role in behavioral change in response to environmental factors [13,14]. One possible mechanism for integrating environmental factors is through the epigenetic gene regulation of monoaminergic and neuropeptidergic neurons.
Acetylation of histones is a dynamic process that regulates gene expression in response to upstream cellular signaling and also regulates multiple behaviors, including addiction, depression, agerelated memory impairment, and memory recall [15][16][17][18]. Histone deacetylases (HDACs) remove an acetyl group from the lysine on the histone in a sequence-specific manner to repress, and in some cases enhance, gene transcription. HDACs also deacetylate nuclear and cytoplasmic proteins, including p53, STAT3, and atubulin [19][20][21][22]. The HDAC family comprises the following classes: class I (HDAC1, -2, -3, and -8), class IIa (HDAC4, -5, -7, and -9), class IIb (HDAC6 and -10), and class IV (HDAC11) [20,23,24]. HDAC4 and -5 show a nuclear-cytoplasmic shuttling in an activity-dependent manner in neural cells [25]. Recently, HDAC4 has been localized in the dendritic shaft and spines [26], suggesting that HDACs may work outside the cell body to regulate synaptic activity and dendritic transport. Consistently, overexpression of HDAC2 results in impaired memory formation [27]. In addition to memory formation, HDACs may play important roles in feeding and metabolism [28,29]. Increasing attention has been given to the versatile roles of HDACs in a variety of brain functions and disorders. However, only a few reports have detailed the localization of HDACs in the brain [26,29,30], which prompted us to examine the distribution of the HDAC family in monoaminergic and neuropeptidergic neurons.
In the present study, we examined the subcellular distribution of HDACs in the CRH, oxytocin, vasopressin, orexin, AgRP, POMC, histamine, dopamine, serotonin, and noradrenaline neurons in mice using the immunofluorescence method.

Animal
Male C57BL/6J mice (25-30 g, 12-16 week-old; n = 6) were obtained from Charles River Laboratories of Japan (Tokyo, Japan). Mice were provided food and water ad libitum, maintained on a 12-hour light/dark cycle (lights on, 0800-2000 h), and housed under controlled temperature (2561uC) and humidity conditions. All procedures were approved by the Institutional Animal Care and Use Committee of Toho University (Approved protocol ID #12-52-81).

Antibodies
The primary antibodies used in the current study are summarized in Table 1. As markers of the oxytocin and vasopressin neurons, we used antibodies for neurophysin I and copeptin, respectively. Neurophysin I is a peptide derived from an oxytocin precursor and is a marker for oxytocin neurons. Copeptin, a peptide derived from a vasopressin precursor, colocalizes with vasopressin and not oxytocin [31]. An antibody for tyrosine hydroxylase was used as a marker of dopamine and noradrenaline neurons. The secondary antibodies used for immunofluorescent visualization were Alexa 555-conjugated anti-goat IgG antibody (1:400, A21432, Invitrogen, CA, USA), Alexa 555-conjugated anti-guinea pig IgG antibody (1:400, A21435, Invitrogen), or Alexa 555-conjugated anti-mouse IgG antibody (1:400, A31570, Invitrogen), and Alexa 488-conjugated anti-rabbit IgG antibody (1:400, A21206, Invitrogen).

Immunofluorescent Staining
For immunofluorescent staining, mice (n = 3) were deeply anesthetized with sodium pentobarbital and perfused transcardially with phosphate-buffered saline (PBS, 0.1 M, pH 7.4) followed by phosphate-buffered 4% paraformaldehyde (PFA). Brains were rapidly removed, post-fixed overnight in phosphatebuffered 4% PFA, and equilibrated in 30% sucrose for 2 days. Brains were sectioned on a cryostat at 30 mm. Sections were stored in a cryoprotective tissue collection solution (25% glycerol, 30% ethylene glycol, 0.05 M phosphate buffer (PB)) at 220uC until use. Immunofluorescence was performed using a free-floating method. The brain sections were washed 2610 minutes in PBS and blocked for 1 hour in a blocking solution containing 0.1 M PB, 0.25% Triton X-100, and 5% normal donkey serum. The brain sections were then incubated with primary antibodies for CRH, orexin, neurophysin I, copeptin, AgRP, POMC, histamine, tyrosine hydroxylase, serotonin, MAP-2, or PSD-95 and with antibodies for HDAC1-11 with 0.25% Triton X-100 and 3% normal donkey serum overnight at 4uC.
After washing the sections 2620 minutes in PBS, the sections were incubated with a fluorescence-conjugated secondary antibody, and Hoechst 33342 (2 mg/ml, H21492, Invitrogen) for 2 hours at room temperature. After washing the sections 2610 minutes in PBS and 3 minutes in PB, the sections were mounted on a glass slide with Gel/Mount (BioMeda, CA, USA). For immunofluorescent staining of CRH and AgRP neurons, colchicine was intracerebroventricularly injected into the mice (n = 3) one day before decapitation. Colchicine inhibits axonal transport by suppressing microtubule polymerization but does not affect nuclear-cytoplasmic shuttling.

Confocal Laser Scanning Microscopy
Samples were observed using confocal laser scanning microscopy as previously described [32]. In brief, immunofluorescent images were captured using a scanning confocal microscope (LSM510 META, Zeiss, Oberkochen, Germany) with objectives of 406, 636, or 1006. The pinhole size was adjusted, so the optical thickness of the sections was 0.6-1.0 mm. To obtain immunofluorescent images, each channel was collected separately with single wavelength excitation and then merged to produce a composite image. Experimental controls were prepared in which one or both of the primary antibodies were omitted from the reaction solution. Confocal laser scanning microscopy showed no immunolabeling of omitted antibodies in the control sections. Photoshop CS5 (Adobe Systems, Mountain View, CA) was used to combine drawings and digital images into plates. The contrast and brightness of images were adjusted. HDAC immunoreactivities localized in dendrites were carefully confirmed using z-stack images.

Image Analysis
To assess the double immunofluorescence data, we observed all CRH, oxytocin, and vasopressin neurons of the bilateral PVN of three brain sections (0.7-0.9 mm posterior to bregma; corresponding to Figures 37,38 in Franklin and Paxinos [33]); all orexin neurons of the bilateral LHA, and all AgRP and POMC neurons of the bilateral ARC of three brain sections (1.5-1.8 mm posterior to bregma; corresponding to Figures 44-46 in Franklin and Paxinos [33]); all histamine neurons of the bilateral TMN of three brain sections (2.5-2.8 mm posterior to bregma; corresponding to Figures 52-54 in Franklin and Paxinos [33]); all dopamine neurons of the bilateral VTA of three brain sections (3.0-3.2 mm posterior to bregma; corresponding to Figures 56, 57 in Franklin and Paxinos [33]), all serotonin neurons of the DR of three brain sections (4.5-4.7 mm posterior to bregma; corresponding to Figures 68-70 in Franklin and Paxinos [33]); and all noradrenaline neurons of the bilateral LC of three brain sections (5.4-5.6 mm posterior to bregma; corresponding to Figures 76-78 in Franklin and Paxinos [33]) per animal. The percentage of HDAC-immunoreactive cells among the monoaminergic and neuropeptidergic neuron groups was determined. To verify the localization of HDACs at the subcellular level, all brain sections were counterstained with the nuclear dye Hoechst 33342. Nuclear or cytoplasmic localization of the HDACs was determined based on the colocalization with Hoechst 33342. To assess the extent of the HDACs immunoreactivities, we used a four-point scale based on the intensity and area of the immunoreactivities in the following manner: +++, strong HDACs immunoreactivity; ++, moderate HDACs immunoreactivity; +, weak HDACs immunoreactivity; and 2, no HDACs immunoreactivity above background. The assessment was independently performed by two observers.

HDAC1
In general, HDAC1 immunoreactivity was recognized mainly in the nuclei of neurons and glial cells (data not shown), with punctate immunoreactivity in neuropils throughout the brain. A few HDAC1-immunoreactive puncta were also observed in the cytoplasm and dendrites. Nuclear  Table 2). HDAC1-immunoreactive puncta were uniform in size and widely distributed in the PVN, LHA, ARC, TMN, DR and LC. HDAC1-immunoreactive puncta in the   neuropils were often found in dendrites and only a few HDAC1immunoreactive puncta were colocalized with PSD95-immunoreactive puncta, which correspond to the postsynaptic area.

HDAC2
Similar to HDAC1, strong HDAC2 immunoreactivity was mainly found in the nuclei of neurons, glial cells (data not shown). Only a few HDAC2-immunoreactive puncta were observed in the hypothalamus, VTA, DR, and LC, and there were fewer HDAC2immunoreactive puncta than for HDAC1. HDAC2 expressions was recognized in a broad range of neuron groups, including the CRH neurons (10060%), oxytocin neurons (9861%), and vasopressin neurons (10060%) of the PVN; the orexin neurons (10060%) of the LHA; the AgRP neurons (10060%) and POMC neurons (9561%) of the ARC; the histamine neurons of the TMN (10060%); the dopamine neurons (10060%) of the VTA; the serotonin neurons (10060%) of the DR; and the noradrenaline neurons (10060%) of the LC (Figures 1J-P, Table 2). A small number of HDAC2-immunoreactive puncta were colocalized with PSD95 immunoreactive puncta of dendrites.

HDAC5
HDAC5 immunoreactivity was generally observed in both the nucleus and cytoplasm. HDAC5 expression was observed in the   (Figures 3A-D). The noradrenaline neurons (10060%) showed very intense immunoreactivity for HDAC5 in the cytoplasm ( Figure 3E). HDAC5 was detected in the cytoplasm and dendrites of orexin neurons. Only a subset of oxytocin neurons (563%), vasopressin neurons (563%) and POMC neurons (1561%) did not show any immunoreactivity for HDAC5 (Table 2). HDAC5-immunoreactive puncta were variable in size and distributed in the PVN, LHA, ARC, TMN, and DR ( Figures 3A-E). HDAC5-immunoreactive puncta were observed in the MAP2-positive dendrites ( Figure 3F). A very small population of HDAC5-immunoreactive puncta in the neuropil was overlapped with PSD95-immunoreactive puncta.

HDAC8
HDAC8 immunoreactivity showed a clear difference among neuronal groups (Figures 4I-O). Surprisingly, all histamine neurons (10060%) of the TMN showed HDAC8 immunoreactivity around and adjacent to histamine-immunoreactive cytoplasm, suggesting that HDAC8 was localized adjacent to the plasma membrane ( Figure 4L). In addition, a small population of oxytocin neurons (463%) showed HDAC8 immunoreactivity in a pericellular pattern, similar to the pattern observed in the histamine neurons. In contrast, HDAC8 immunoreactivity was not recognized in any of the CRH neurons, vasopressin neurons, orexin neurons, AgRP neurons, POMC neurons, dopamine neurons, serotonin neurons, or noradrenaline neurons (Figures 4I-K, M-O, Table 2). Importantly, the majority of HDAC8-positive neurons of the brain, including the PVN, ventromedial hypothalamus, and LHA, showed HDAC8 immunoreactivity throughout the cytoplasm without a pericellular staining pattern, as previously reported ( Figure 4I) [29]. HDAC8-immunoreactive puncta were uniform in size and widely distributed in the PVN, LHA, ARC, TMN, VTA, DR, and LC. The pericellular immunoreactivity was also observed along dendrites ( Figure 4P). A very small population of HDAC8immunoreactive puncta in the neuropil was overlapped with PSD95-immunoreactive puncta.

Discussion
In the present study, we examined the expression profile of the HDAC protein family in monoaminergic and neuropeptidergic neurons. The expression patterns of HDAC1,-2,-3,-5,-6,-7,-9, and -11 were very similar among all monoaminergic and neuropeptidergic neurons, while the HDAC4, -8, and -10 immunoreactivity patterns were clearly different among the neuronal groups.

Differential Expression of HDAC10 among Neuron Groups
HDAC10 expression was observed in AgRP neurons, POMC neurons, dopamine neurons and noradrenaline neurons but not in neurons containing CRH, oxytocin, vasopressin, orexin, histamine, or serotonin. Nuclear HDAC10 immunoreactivity was consistent among HDAC10-positive neurons, while cytoplasmic HDAC10 immunoreactivity was clearly observed in the dopamine neurons and noradrenaline neurons.
HDAC10, a member of the class IIb HDAC family, has a catalytic domain in the amino terminal half and is leucine-rich in the carboxyl terminal half. Although the function of HDAC10 remains largely unknown, HDAC10 is associated with hsc70, Pax3, and KAP1 and interacts with histones to enhance the deacetylated status of target molecules [34]. Another member of the class IIb HDAC family, HDAC6, has a highly similar catalytic domain as HDAC10 and deacetylates histones and cytoplasmic proteins such as alpha-tubulin, actin-binding protein, contactin, and heat shock chaperone protein HSP90 [20,35]. Thus, HDAC10 could alter the acetylation status of a variety of nuclear and cytoplasmic molecules of neurons, resulting in a change in gene transcription and cellular function. The clear difference in HDAC10 expression among neuronal groups suggests that HDAC10 regulates gene expression levels, which are pivotal for the functional and biological specificity of AgRP, POMC, dopamine, and noradrenaline neurons.

HDAC8 Expression in Histamine Neurons
Although HDAC8 was not expressed in the neuropeptidergic and monoaminergic neurons we examined in the current study, the histamine neuron is a unique exception. HDAC8 immunoreactivity was found in the cytoplasm of all histamine neurons with a pericellular pattern. Surprisingly, the HDAC8 immunoreactivity within the histamine neurons was confined to the cytoplasmic periphery, sparing a cytoplasmic region positive for histamine. We observed that the subcellular immunoreactive pattern for HDAC8 in neurons of the mouse brain displays nucleocytoplasmic, cytoplasmic, and pericellular distribution patterns. In the amygdala, cerebral cortex, hippocampus, and hypothalamus, a small population of neurons showed moderate to strong HDAC8 immunoreactivity in the cytoplasm and dendrites [29]. Confocal observation of HDAC8 immunoreactivity also identified cytoplasmic expression of HDAC8 in the amygdala, hippocampus, and cerebral cortex (data not shown).
HDAC8 belongs to the class I HDAC family; can deacetylase all core histones; and is associated with EST1B, Hsp70, Hsp90 and STIP [20,36]. In smooth muscle cells, HDAC8 is co-localized with alpha-smooth muscle actin filament [37] and is interacts directly with it [38]. Although the function of HDAC8 in neurons remains unknown, the abundant HDAC8 localization in the peripheral region of the cytoplasm and dendrites of a specific subset of neurons, including histamine neurons, suggests that an undiscovered role of HDACs is in intracellular signaling rather than gene transcription. Although we previously found that a subset of neurons in the anterior parvicellular and periventricular subdivisions of the PVN changed HDAC8 immunoreactivity in response to fasting and high-fat diet feeding [29], none of the PVN neurons that expressed CRH, oxytocin, or vasopressin were positive for HDAC8. This result is consistent with the localization of only a few neurons containing CRH, oxytocin, or vasopressin in the anterior parvicellular and periventricular subdivisions of the PVN [39].

Differential Expression Profiles among Neuron Groups
All groups of monoaminergic and neuropeptidergic neurons showed immunoreactivity for HDAC4, but the subcellular distribution and intensity varied. Cytoplasmic immunoreactivity for HDAC4 was not observed in AgRP neurons, POMC neurons, or dopamine neurons; however, cytoplasmic HDAC4 was detected in other neuronal groups. Interestingly, almost all neurons showing cytoplasmic HDAC4 immunoreactivity did not have immunoreactivity for HDAC10 ( Table 2). The only exception was in the noradrenaline neurons, which showed both HDAC10 immunoreactivity and cytoplasmic HDAC4 immunoreactivity.
Thus, based on the HDAC4, -8, and -10 immunoreactivities, the monoaminergic and neuropeptidergic neurons we examined were classified into four groups: 1) HDAC8-positive: histamine neurons; 2) HDAC10-positive and cytoplasmic HDAC4-negative: AgRP neurons, POMC neurons and dopamine neurons; 3) HDAC10-positive and cytoplasmic HDAC4-positive: noradrenaline neurons; and 4) HDAC10-negative and cytoplasmic HDAC4positive: CRH neurons, oxytocin neurons, vasopressin neurons, and serotonin neurons ( Figure 6). Although this classification is valid for the adult male mouse under basal conditions, it could differ based on gender, age, stress, and nutrition. Importantly, HDAC4 showed activity-dependent translocation from the nucleus to the cytoplasm in vitro [25], suggesting that the subcellular localization of HDAC4 could be dynamically regulated in response to environmental stimuli. Generally, most monoaminergic and neuropeptidergic neurons express HDACs in an all-ornone manner, but POMC neurons showed a variable population of HDAC-positive cells, especially for HDAC3 and HDAC10. This variable expression may be associated with the differential expression of the anorexigenic genes of POMC neurons, by modulating the acetylation status of genes important for feeding and body weight regulation when a mouse is fasted or fed a highfat diet.

HDACs Immunoreactivity in the Neuropils
We observed HDAC-immunoreactive puncta in the neuropil. The number of puncta varied among both HDACs and brain regions. Double immunofluorescent observation of HDACs with MAP2 showed that HDACs-immunoreactive puncta were fre-quently localized in the dendrites. Puncta immunoreactive for HDAC4 and -11 showed frequent colocalization with PSD95immunoreactive puncta [26], but the majority of PSD95immunoreactive puncta were negative for HDACs. Although the substrates and functions of HDACs in the dendrites or spines are unknown, punctate distribution suggests that the HDACs exit in a functional compartment which may be involved in molecular traffic between the cell body and spines, or spine activity [40,41].