Anatomical and functional characterization of clock gene expression in neuroendocrine dopaminergic neurons

Michael T. Sellix, Marcel Egli, Maristela O. Poletini, De'Nise T. McKee, Matthew D. Bosworth, Cheryl A. Fitch, Marc E. Freeman

Abstract

Oscillations of gene expression and physiological activity in suprachiasmatic nucleus (SCN) neurons result from autoregulatory feedback loops of circadian clock gene transcription factors. In the present experiment, we have determined the pattern of PERIOD1 (PER1), PERIOD2 (PER2), and CLOCK expression within neuroendocrine dopaminergic (DAergic) neurons (NDNs) of ovariectomized (OVX) rats. We have also determined the effects of per1, per2, and clock mRNA knockdown in the SCN with antisense deoxyoligonucleotides (AS-ODN) on DA release from NDNs. Diurnal rhythms of PER1 and PER2 expression in tuberoinfundibular DAergic (TIDA) and periventricular hypophyseal DAergic (PHDA) neurons, peaked at circadian time (CT)18 and CT12, respectively. Rhythms of PER1 expression in tuberhypophyseal neuroendocrine DAergic (THDA) neurons were undetectable. Rhythms of PER2 expression were found in all three populations of NDNs, with greater levels of PER2 expression between CT6 and CT12. AS-ODN injections differentially affected DA turnover in the axon terminals of the median eminence (ME), neural lobe (NL) and intermediate lobe (IL) of the pituitary gland, resulting in a significant decrease in DA release in the early subjective night in the ME (TIDA), a significant increase in DA release at the beginning of the day in the IL (PHDA), and no effect in the NL (THDA). AS-ODN-treatment induced a rhythm of DA concentration in the anterior lobe, with greater DA levels in the middle of the day. These data suggest that clock gene expression, particularly PER1 and PER2, within NDNs may act to modulate diurnal rhythms of DA release from NDNs in the OVX rat.

  • prolactin
  • dopamine
  • suprachiasmatic nucleus
  • hypothalamus

dopamine (DA) of hypothalamic origin exerts tonic inhibitory control over prolactin (PRL) secretion (for review see Ref. 17). DA is released directly into hypothalamo-hypophyseal portal blood from three populations of neuroendocrine DAergic neurons (NDNs) (5, 17). These subpopulations include the tuberoinfundibular DAergic (TIDA) and tuberohypophyseal neuroendocrine DAergic (THDA,A12) neurons with cell bodies in the arcuate nucleus (ARN) and periventricular hypophyseal DAergic (PHDA,A14) neurons with cell bodies in the periventricular region (5, 17). THDA and PHDA axons traverse the pituitary stalk and terminate on fenestrated short portal vessels within the neural (NL) and intermediate (IL) lobes of the pituitary gland (21). TIDA axons terminate on fenestrated capillary beds within the external zone of the median eminence (ME) that drain into long portal vessels, transporting DA to the anterior lobe (AL) of the pituitary gland (8, 10). TIDA neurons are well established as the primary PRL inhibitory neurons, although growing importance has been assigned to both THDA and PHDA neurons in the regulation of PRL secretion (13, 21, 43, 52, 61).

We have observed diurnal rhythms of DA turnover in the nerve terminals of TIDA, THDA, and PHDA neurons in the ovariectomized (OVX) rat (57, 58). Whereas both TIDA and PHDA neurons display circadian rhythms of DA release, THDA neurons failed to exhibit a free-running rhythm. In mammals, circadian rhythms are coordinated by the master biological clock located in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus. The SCN contains several thousand individual autonomous oscillators shown to express circadian rhythms of gene expression and physiological activity. The molecular substrates for these sustained endogenous rhythms are a series of tightly regulated transcriptional/translational negative feedback loops of circadian clock gene transcription factor expression (15, 55). Several circadian clock gene transcription factors have been cloned in mammals, including the period gene family, clock and bmal1 genes (15, 55). Recent data suggest that neural targets of the SCN, the primary circadian oscillator, express the putative clock genes with a circadian rhythm (1, 22, 23, 75, 76). These tissues appear to function as “damped” oscillators that are both entrained by photic cues transduced by the SCN and actively stimulated to maintain rhythmicity via SCN afferents (76). Therefore, in the absence of SCN input, these tissues cannot sustain a circadian rhythm of clock gene expression for more than 3–5 days, or alternatively, display desynchronized rhythms of gene expression masked by SCN-mediated coordination. These areas include the ARN, paraventricluar nucleus (PVN), pineal gland, and pituitary gland (76). However, a notable exception to this pattern is the olfactory bulb, which appears to maintain free-running oscillations of PERIOD1 (PER1) expression in SCN lesioned rats (22). Moreover, recent experiments suggest that cultured NIH3T3 fibroblasts express the gene Rev-erbα in a self-sustained cell-autonomous rhythm that continues in daughter cells after mitosis (48).

Recent reports suggest that NDNs within the rostral ARN (RARN) express PER1 with a diurnal rhythm (33). Because several studies indicate that clock genes play an active role in cellular physiology through regulation of protein-protein interactions and gene transcription, it seems plausible that clock genes may act within NDNs to drive expression of enzymes responsible for DA synthesis within the presynaptic cell and/or release of DA from the presynaptic terminal (18, 24, 69). A corresponding rhythm of clock gene expression within NDNs would suggest that NDNs may act as semiautonomous or damped circadian oscillators. In fact, a rudimentary examination of the hypothalamic-specific promoter region of the tyrosine hydroxylase (TH) gene reveals a significant (>10) number of E-box (CACGTG)-like sequence motifs (Ref. 42 and Sellix MT and Freeman ME, unpublished observation). CLOCK:BMAL1 heterodimers drive gene transcription through DNA recognition at the E-box motif; E-box sequences within the TH promoter may suggest a level of clock-controlled gene expression (CCG) in NDNs.

Given the novel and diverse rhythms of DA turnover we have observed from the TIDA, THDA, and PHDA neurons under a standard 12:12-h light-dark cycle, we hypothesized that rhythms of clock gene expression in NDN neurons correspond to the rhythmic synthesis and release of DA from nerve terminals. Using Western blot analysis and immunocytochemistry, we have attempted to verify the expression of several of the clock gene products within tissues of the hypothalamo-pituitary-gonadal (HPG) axis and to further characterize the rhythmic expression of these gene products within each of the three populations of hypothalamic DA neurons in the OVX rat. We have used antibodies against the DA synthesis rate-limiting enzyme TH to identify DAergic neurons within the PVN (PeVN, PHDA neurons), RARN (THDA neurons), and dorsomedial ARN (DMARN; TIDA neurons) and specific primary antibodies against PER1/2 and CLOCK to identify those DAergic neurons expressing clock genes in their nucleus under a 12:12-h light-dark cycle.

Several reports have shown that mutations affecting the core molecular clock network, including PER1/2, CLOCK, and BMAL1, have dramatic and varied effects on circadian rhythms of locomotor activity (3, 25, 35, 59). Therefore, we hypothesize that acute transient knockdown of clock gene expression in the SCN may have a significant effect on the timing and/or amplitude of the diurnal rhythms of NDN activity. The inability of SCN-targeted antisense treatment to disrupt diurnal rhythms of DA turnover would suggest that NDNs maintain the capacity to release DA with an endogenous rhythm in the absence of timing cues from the SCN. In the current experiment, we determined the effects of short-duration knockdown of per1, per2, and clock mRNA on the diurnal rhythms of NDN activity, as well as PRL and corticosterone (CORT) secretion in the OVX rat. Given the potential relationship between diurnal rhythms of clock gene expression in SCN neurons and rhythms of DA turnover in NDNs, it is reasonable to assume that a molecular lesion of the SCN may lead to transient disruption of CCG in SCN neurons. Disruption of clock-controlled genes, such as AVP and VIP, in SCN neurons would disrupt the ability of the SCN to communicate photoperiodic cues to target regions, such as the mediobasal hypothalamus. We believe that our results will add to our general understanding of the functional role for clock genes in target oscillators within the brain that may or may not depend on the SCN for their synchronization.

MATERIALS AND METHODS

Animals

Adult female Sprague-Dawley rats (>60 days of age) weighing 250–300 g (Charles River Labs, Wilmington, MA) were housed under a standard 12:12-h light-dark cycle with lights on at 0600 h with constant temperature (25°C) and humidity. Standard rat chow and water were available ad libitum. The room was illuminated with four 40-W fluorescent bulbs, producing a minimum illumination of 100 lux at cage level. During the 12-h dark phase, animals were handled with the aid of infrared goggles (Unitec; Night Vision Optics, Huntington Beach, CA) and killed in dim red light (<1 lux). All experimental protocols were approved by the Florida State University Animal Care and Use Committee.

Bilateral Ovariectomy and Analysis of Drinking Rhythms

Animals were anesthetized with halothane and OVX bilaterally. All animals were placed under a standard 12:12-h light-dark cycle (lights on 0600 h–1800 h) for 2–3 days for habituation to the home cage. Feeding and drinking patterns are established methods for determining circadian time (CT), a subjective measure of time based on the activity of the animal, independent of the 12:12-h light-dark cycle (14, 45, 68). Drinking was measured over the 24-h day with an automated device (Lickometer; Dilog Instruments, Tallahassee, FL) counting individual licks in 30-s bins over 24 h and the onset of the subjective activity period (CT12) was calculated. CT12 was used as a reference time for tissue collection under a standard 12:12-h light-dark cycle. The use of the CT scale allowed us to make direct comparisons between data collected in the current experiments and analyses of DA turnover performed under constant conditions in previously published reports (57, 58). Double-plotted actograms of drinking activity (12-h moving average of drinking activity around a central peak of activity) were produced with Circadia software (version 2.1.16; Behavioral Cybernetics, Tallahassee, FL).

Tissue Preparation for Immunocytochemistry and Digital Microscopy

Animals were deeply anesthetized with halothane and transcardially perfused through the descending aorta with 60 ml of prewash (0.1 M PBS pH 7.35 containing 0.5% sodium nitrite and 10,000 IU/l of heparin), followed by 200 ml of ice-cold 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO) in 0.1 M PBS, pH 7.35. Following perfusion, brains were removed and placed in 4% paraformaldehyde to postfix at 4°C overnight. The following morning, brains were blocked immediately anterior to the optic chiasm, just posterior of the mammillary bodies, and then cryoprotected in a 20% sucrose solution for 36–48 h. Brains were sectioned on a sliding microtome (Richard-Allan Scientific, Kalamazoo, MI) at 40-μm thickness and collected in four adjacent series. Sections were collected in 12-well plates containing cryoprotectant solution (71) and stored at −20°C.

Sections were processed for PER1/2, CLOCK, and TH immunoreactivity (IR). Each series of sections was rinsed three times for 15 min in 0.1 M PBS containing 0.1% Triton X-100 and 0.1% sodium azide (PBX) to remove cryoprotectant (Triton X-100 and PBX; Sigma-Aldrich). Nonspecific binding was blocked in 10% normal goat serum (Chemicon, Temecula, CA) in PBX for 1 h. Polyclonal primary antibodies against mouse PER1 (rabbit host, polyclonal used at 1:10,000; kind gift of Dr. David Weaver), mouse PER2 (rabbit host, polyclonal used at 1:1,000; Alpha Diagnostics, San Antonio, TX), or mouse CLOCK (goat host, polyclonal used at 1:10,000, Santa Cruz Biotechnology, Santa Cruz, CA) were incubated with monoclonal anti-mouse TH (1:10,000; Chemicon, Temecula, CA) for 48 h at 4°C on a rotating bench-top shaker. Mouse antigens were chosen due to the >90% homology in amino acid sequence between mouse, rat, and human circadian clock proteins (NCBI GeneBank) (3, 16, 37, 55). All primary and secondary antibodies were diluted in PBX. Sections were washed three times for 10 min with PBX between each step. Donkey anti-rabbit or anti-goat CY3 (excitation = 550 nm, emission = 570 nm) conjugated and anti-mouse CY2 (excitation = 492 nm, emission = 510 nm) conjugated secondary antibodies were added (1:600 in PBX; Jackson Immunochemicals, West Grove, PA), and sections were again incubated at 4°C for 12–18 h. Sections were then rinsed with PBX three times for 15 min, mounted, and coverslips were applied with diluted aqua-polymount (Polysciences, Warrington, PA). After several hours, the edges of the coverslips were sealed with nail polish. Controls included sections wherein primary antibody was excluded or PER1/2 and CLOCK primary antibodies were preabsorbed with a substantial amount (10- to 100-fold higher concentration) of the peptide fragments they were raised against (data not shown).

Neurons in the RARN (THDA), DARN (TIDA), and PeVN (PHDA) were identified as DAergic based on the presence of TH-IR as previously described (11, 3840). PER1/2 or CLOCK and TH double-labeled neurons were identified and counted within the ARN (DMARN and RARN) and PeVN. Although clock proteins translocate back to the nucleus, like most proteins, they are formed in the cytoplasm and require phosphorylation and dimerization to reach the nucleus (37). However, their transcriptional activation/repression function occurs within the nucleus. Therefore, we counted PER1/2 and CLOCK-IR nuclei within TH-IR neurons. Given the absence of nuclear staining for TH, we were able to clearly and efficiently identify clock protein-IR nuclei in TH-IR neurons (Fig. 1). Images were taken with a Leica DMLB compound stereomicroscope fitted with short-pass filters (CY2, 488 nm; CY3, 596 nm) and a SPOT-RT-cooled charge-coupled device camera (Diagnostic Instruments, Sterling Heights, MI) attached to a microcomputer. Image acquisition and analysis was conducted using Metamorph software (Universal Imaging, Downingtown, PA). Grayscale images of circadian clock gene and TH-IR were overlaid and pseudocolored in Metamorph. For display purposes, representative videomicrographs were taken of each region for each clock protein with a Zeiss 510 confocal laser scanning microscope at 1-μm thick sections.

Fig. 1.

Clock gene expression within the nucleus of neuroendocrine dopaminergic neurons (NDNs). CLOCK immunoreactivity (IR; red stain; white arrowheads) within the nucleus of NDNs [tyrosine hydroxylase IR (TH-IR) neuron; green stain]. An absence of granular staining within the nucleolus is a clear indicator of nuclear staining. Several non-TH-IR neurons within the hypothalamus were also identified as CLOCK-IR.

Stereotaxic Implantation of Bilateral SCN Cannulae and Intra-SCN Injection of Deoxyoligonucleotides

A minimum of 5 days after OVX, rats were anesthetized (100 μl/100 g body wt) with a mixture of ketamine (49 mg/ml)/xylazine (1.8 mg/ml) and implanted stereotaxically with bilateral stainless steel guide tubes (1.5 mm apart, 9.5 mm in length, 27 gauge) whose tips were placed at the dorsal border of the SCN (0.8 mm posterior to bregma, 7.9 mm ventral to the dorsal surface of the dura mater). Bilateral 33-gauge solid steel mandrils were placed inside the guide tubes and dust caps were used to secure the apparatus. Animals were allowed to recover on a heated pad and returned to their home cage for 2–3 days. After thorough recovery, animals were transferred, during the light portion of the 12:12-h light-dark cycle between 1000 and 1400 h, to the Lickometer device and allowed to habituate for a minimum of 24 h. Animals were placed in the Lickometer device in a 12:12-h light-dark cycle with lights on from 0600 h to 1800 h for 5 days. On the fifth day under a 12:12-h light-dark cycle animals were injected intra-SCN with clock gene antisense cocktail. Briefly, animals were anesthetized with halothane and 33-gauge bilateral mandrils were removed. Bilateral internal cannulae (33 gauge, 10.5 mm length, 1 mm extension) were inserted into the guide tubes, and 800 nl of antisense deoxyoligonucleotides (AS-ODN) or random-sequence oligos (RS-ODN) were injected at 200 nl/min with two 1-μl Hamilton syringes attached to an automated syringe pump (KD Scientific, Fisher Scientific, Fair Lawn, NJ). AS-ODNs were generated against the 5′ transcription start site (5′INI) and 3′ cap site of per1, per2, and clock mRNA. Sequences for mper1 (GenBank accession no.: NM 011065) and mper2 (GenBank accession no.: NM 011066) AS-ODN were slightly modified from those used by Akiyama and colleagues (2, 70) to slightly increase the GC content of the oligonucleotide. Clock AS-ODNs were developed independently in our laboratory using the known sequence of clock mRNA (GenBank accession no.: NM 007715). Per1, per2, and clock AS-ODN and RS-ODN sequences are listed in Table 1. According to the protocol developed by Akiyama and colleagues (2, 70), we verified both time and dose dependent effects of AS-ODN on PER1/2 and CLOCK expression in the SCN with immunoblots for each clock protein (data not shown).

View this table:
Table 1.

Antisense and random sequence deoxyoligonucleotide sequences

Preliminary experiments verified a significant knockdown of PER1/2 and CLOCK expression in the SCN by >60% within 6 h of injection with 3 nmol of AS-ODN (2.5 μg/μl, 800 nl injection, data not shown). Per1, per2, and clock mRNA levels recovered to control values by 12 h postinjection, and remained at normal levels at 36 and 48 h after infusion of AS-ODN (data not shown). Therefore, to verify the acute effects of clock gene mRNA knockdown, all experimental animals were injected 6 h before death. Control animals were injected with RD-ODNs with the same nucleotide content (%AGCT) as the AS-ODN but are not complementary to clock gene mRNA sequences (verified with Primer 3.1, MIT, AS-ODN, and RS-ODN sequences listed in Table 1). Following death, small tissue blocks containing the bilateral SCN were collected, and immunoblotting for PER1/2 and CLOCK was used to quantify the effects of AS-ODN and RS-ODN in the SCN. Optical density (OD) for each gene was normalized to β-actin loading control for each sample and densitometry was carried out on a gel documentation system (Bio-Rad, Hercules, CA). OD (U/mm2) of PER1/2 and CLOCK, and actin loading controls were determined and used to calculate relative protein abundance. Data represent the mean relative protein abundance of all AS-ODN and RS-ODN animals used for DA turnover analysis.

RT-PCR Analysis of Clock Gene Expression

RT-PCR was used to determine the tissue distribution of several clock gene mRNAs in OVX female rat brain and pituitary gland. Animals kept under a standard 12:12-h light-dark cycle were briefly anesthetized under hypercapnic (50% CO2) conditions and killed by decapitation ∼8 zeitgeiber time, the approximate peak time of per1 and per2 mRNA expression in the SCN (55). Brains were rapidly removed and placed in a coronal brain matrix on ice (ASI Instruments, Warren, MI). Individual 2-mm thick sections containing the SCN or the medial basal hypothalamus (including the ARN and ME) were placed on a diethylpyrocarbonate-treated sterile glass slide. A single, 4-mm2 cube including the entire SCN or ARN was excised from the coronal brain slices and placed in 1 ml of Trizol reagent (Invitrogen, Carlsbad, CA), followed by homogenization on ice. After homogenization, 100 μl of chloroform was added, and each tube was briefly vortexed and then placed in a centrifuge for 15 min at 10,000 rpm (12,000 g). The aqueous phase was removed from each tube and pooled according to sample. An equal volume of isopropanol was added to each tube, and all tubes were placed in a −80°C freezer. After 6 h at −80°C, precipitated RNA was pelleted by centrifugation at 13,000 rpm for 20 min at 4°C. The remaining isopropanol was removed, and the pellets were washed twice with 70% ethanol, allowed to air dry, and then resuspended in diethylpyrocarbonate-treated water. Total RNA concentration was analyzed by UV/vis spectrophotometry and diluted to normalize the sample to 500 ng/μl of which 5 μg was used for RT. Messenger RNA was reverse transcribed to cDNA using the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Following RT, clock genes were amplified using 2 μl of RT reaction with the Platinum PCR SuperMix (Invitrogen) and gene-specific primers for clock genes from mouse sequences including mper1, mper2, mclock, and mbmal1 (10 pmol/reaction, 1 μl final volume of 50 μl). PCR was performed in a PTC-200 thermocycler (MJ Research, Watertown, MA) at 94°C for 1 min followed by 35 cycles of amplification (94°C for 1 min, 58°C for 1 min, 72°C for 4 min) and a final extension at 74°C for 15 min. Samples were stored at −20°C until use. Reaction products were separated with 2% agarose (analytical grade; Invitrogen) gels in Tris/acetate/EDTA (TAE; Fisher Scientific, Suwanee, GA) buffer. Gels were stained by immersion in ethidium bromide solution (0.1 μg/ml in TAE) for 10 min followed by two 20- to 30-min TAE rinses. Gels were photographed and analyzed using a Kodak Gel Logic 100 Imaging and Analysis System (Eastman Kodak, Rochester, NY). Remaining PCR products were purified with a Qiagen PCR purification Kit (Qiagen, Valencia, CA) and sequenced at the Florida State University core molecular analysis facility. NCBI blast sequence analysis confirmed amplification of rat clock gene products with mouse primers and supported high homology (>90% in all cases) between mouse, rat, and human clock genes.

Western Blot Analysis and Immunodetection of Clock Gene Products in Neuroendocrine Tissues

Extraction of clock proteins.

Proteins were extracted from neuroendocrine tissues according to the method of Lee et al. (37). Briefly, animals maintained under a standard 12:12-h light-dark cycle were killed at CT12 under hypercapnic conditions and decapitated (CT12 represents the approximate peak of PER1 and PER2-IR in the SCN; Refs. 3, 16, 55). The brain and pituitary gland were rapidly removed, and the brain was placed in a chilled brain-sectioning matrix (ASI Instruments) on ice, whereas the pituitary gland was rapidly frozen in liquid nitrogen. Two-millimeter-thick frontal sections of the brain, including the SCN and the ARN, were made with a sterile razor blade and placed on a chilled glass slide. Using a scalpel, a 3 × 3-mm cube, including the SCN and a 3 × 4-mm angled dissection of the ARN were removed, as well as the cerebellum and a majority of the piriform cortex from both hemispheres. After tissues were dissected, they were rapidly frozen in liquid nitrogen and stored at −80°C until protein extraction. Tissues were thawed on ice in chilled homogenization buffer containing (in mM) 20 HEPES, 100 NaCl, 1 DTT, 5 sodium-β-glycerophosphate, 1 Na orthovanadate, 1 EDTA, 0.5 PMSF, plus 0.05% Triton X-100, and a cocktail of protease inhibitors including aprotonin (10 μg/ml), leupeptin (5 μg/ml), and pepstatin A (2 μg/ml) (37). Tissue samples were sonicated on ice with a minihomogenizer and disposable pestle (Kimble-Kontes, Vineland, NJ) using 10–15 strokes (3–4 s/stroke). After centrifugation at 12,000 g for 15 min, supernatant was transferred to a fresh tube and recentrifuged. Supernatant was removed, and extracted protein concentration was determined by the micromodified bichonchoninic acid protein detection system (Pierce, Rockford IL).

SDS-PAGE and electroblotting.

Extracted proteins were resolved by SDS-PAGE electrophoresis according to the established protocol of Lee et al. (37). Because clock proteins vary significantly in molecular weight (i.e., ∼85–100 kDa CLOCK; ∼180 kDa PER1 and PER2) we used separate gel concentrations for each to maximize resolution (8% CLOCK, 7% PER1/2). Samples were mixed with equal volume of 2× sample buffer containing: 100 mM Tris·HCl (pH 6.8), 4% SDS, 20% glycerol, 5% β-mercaptoethanol (β-ME), 2 mM EDTA, and 0.1 mg/ml bromophenol blue and heated at 95°C for 5 min (37). Samples were briefly centrifuged at 12,000 g and cooled to room temperature. Samples and molecular weight markers were loaded into 10-well preset polyacrylamide gels using the Miniprotean II System (Bio-Rad) and electrophoresis was carried out using constant 150 V for 45 min to 1 h in Tris-glycine-SDS buffer (8 mM Tris-base, 40 mM glycine, 0.1% SDS). Electroblotting was carried out with standard buffers. Briefly, gels were placed in transfer buffer (Tris-glycine-SDS-methanol, 8 mM Tris-base, 40 mM glycine, 0.1% SDS, and 20% methanol) for 1 h to overnight before electroblotting with constant 100 V for 90 min. Transfer was verified by visual inspection of Pagemarker Prestained molecular weight marker and following brief staining with fast-green protein stain.

Immunodetection.

Following verification of successful transfer, blots were placed in 5% nonfat milk for 30 min at RT, followed by primary antibody (1:1,000 anti-mouse CLOCK guinea pig, 1:1,000 anti-mouse PER1/2 guinea pig; kind gift of Dr. Choogon Lee, The Florida State University College of Medicine, Tallahassee, FL) for 12–16 h at 4°C. As described above, mouse antigens were chosen due to the >90% homology in amino acid sequence between mouse, rat, and human circadian clock proteins (NCBI GeneBank) (3, 16, 37, 55). Blots were rinsed three times for 10 min with Tris-buffered saline with 0.1% Tween-20 (TTBS) and placed in secondary antibody (anti-guinea pig IgG conjugated to horseradish peroxidase at 1:5,000; Jackson Immunolabs) for 1 h at RT. Following 6 to 8 10-min washes in TTBS, CLOCK, or PER1/2, protein levels were visualized using enhanced chemiluminescence (Amersham, Piscataway NJ) according to the manufacturer's specifications. Verification of clock gene staining within tissue extracts was verified by comparison with protein extracts from liver of wild-type (WT) and PER1/2 double-knockout mice or kidney proteins from WT and CLOCK knockout mice (kindly provided by Dr. Choogon Lee, data not shown). Blots were stripped of primary/secondary antibody complex according to the protocol provided by Pierce using their stripping and reprobing buffer and reprobed with anti-actin mouse monoclonal primary antibody (1:1,000; Sigma-Aldrich, St. Louis, MO).

Tissue Preparation and Measurement of DA and Dihydroxyphenylacetate by HPLC with Electrochemical Detection

Six hours after AS- or RS-ODN injection, animals were briefly sedated by inducing hypercapnia (50% CO2/O2) and then rapidly decapitated. Trunk blood was collected, and serum samples were frozen at −20°C until assayed for PRL and CORT concentrations by RIA. The brain and pituitary gland were quickly removed and placed on a chilled glass slide. The ME, NL, IL, and ALs of the pituitary gland were carefully dissected, placed in homogenization buffer (0.2 N perchlorate with 50 μM EGTA) and rapidly (∼30 s) frozen in an ArticIce tube transport block (USA Scientific, Ocala FL). Tissue samples were stored at −80°C until assayed for DA and dihydroxyphenylacetate (DOPAC). On the day of analysis for catecholamines, tissue samples were thawed and processed for HPLC-electrochemical detection (HPLC-EC) analysis as previously described (58). The HPLC-EC technique has been well established in our laboratory (12). The concentrations of DA and DOPAC, a primary metabolite of DA, were measured in tissue extracts from the pituitary gland and mediobasal hypothalamus as previously described (58). The amount of catecholamine in each sample was estimated by direct comparison to the area under each peak for known amounts of catecholamine. The amount of 3,4-dihydroxybenzylamine (RT = 6.5 min) recovered was compared with the amount of 3,4-dihydroxybenzylamine added as internal standard and corrected for sample loss (usually <5%). Assay sensitivities are 30 pg for DA and 15 pg for DOPAC. DA turnover is defined as the exocytotic release of DA from neuroendocrine DAergic nerve terminals, DA reuptake, and the degradation of DA to DOPAC by monoamine oxidase in the presynaptic terminal (41).

Protein Assay

The amount of protein in samples for HPLC-EC and Western blot analysis were measured using a micromodified form of the bichonchoninic acid Protein Assay Kit (Pierce, Rockford, IL) as previously described (58). Assay sensitivity was 1 μg protein and the intra-assay coefficient of variation was 5–10%.

RIA

The concentration of PRL in serum was determined by RIA using National Institute of Diabetes and Digestive and Kidney Diseases materials supplied through the National Pituitary Hormone Distribution Program as previously described (11). Serum concentrations of PRL are expressed as ng/ml in terms of the rat PRL Reference Preparation-3 standard. Serum CORT concentration was determined using the commercially available Coat-a-Count rat CORT RIA kit (Diagnostic Products, Los Angeles, CA) according to the manufacturer's specifications. For both PRL and CORT, assay sensitivity was 1 ng/ml, and the interassay and intra-assay coefficients of variation were 10 and 5%, respectively.

Experimental Design

To establish the relationship between diurnal rhythms of DA turnover in NDN neurons and rhythmic clock gene expression, we have identified the pattern of PER1/2 and CLOCK-IR in DAergic neurons of OVX animals under a standard 12:12-h light-dark cycle. We used RT-PCR for clock gene products to verify their expression within tissues of the HPG axis, including the ARN and pituitary gland. Tissues were obtained from OVX animals perfused at CT0, -6, -12 or -18. Brain sections were stained for PER1/2, CLOCK, and TH-IR. The percentage of PER1/2 or CLOCK-TH double-labeled cells was determined in the PeVN, RARN, and DMARN. For antisense knockdown experiments, four adult female OVX Sprague-Dawley rats were housed individually in cages attached to the automated drinking device under light-dark conditions for 5 days. On the fifth day under a 12:12-h light-dark cycle, animals were injected with antisense ODN (AS-ODN) or RS-ODN. Four animals were each killed 6 h after ODN-injection at CT0, -6, -9, -12, -15, or -18 and tissue was collected for HPLC-EC and Western blot analysis. Therefore, animals injected with AS- or RS-ODN at CT18 were killed at CT0, and animals injected at CT0 were killed at CT6. Serum was collected to determine serum PRL and CORT by RIA.

Data Analysis

The percentage of PER1/2 or CLOCK/TH double-labeled cells and the number of single-labeled cells represent the means ± SE of five images/region/animal and a total of three animals per time point (15 total images/region/time point), as a function of CT. Although they exhibit a distinct rhythm, all of our data do not conform to a sine/cosine wave function, which prohibits a nonlinear regression analysis to present the data as a function of time. Moreover, as samples were obtained from each animal following decapitation or perfusion with fixative at individual time points over a 24-h period, it is difficult to extrapolate accurate phase and period measures. Although it is preferable when performing circadian studies to collect serial samples of individual animals, analyses of recovered tissue preclude such an approach in our experiments. To facilitate direct comparisons between current and previous experiments, all data points, regardless of treatment condition, were aligned by CT. The percentage of PER1/2 and CLOCK/TH double-labeled neurons were analyzed with ANOVA for time-of-day effects, followed by Bonferroni post hoc statistical tests. Serum PRL, serum CORT, DA turnover in the ME, NL, and IL, and DA concentration in the AL are expressed as mean (ng/ml, ng/ml, and DOPAC-to-DA ratio, respectively) ± SE of four animals and were analyzed with two-way ANOVA for 1) time of day effects, 2) ODN effects, and 3) the interaction between time and ODN treatment, followed by Bonferroni paired post hoc statistical tests. Relative protein abundance from AS- and RS-ODN animals were analyzed with ANOVA for ODN effects, followed by Bonferroni post hoc tests. In all experiments, significant differences were considered at P < 0.05. ANOVA tests were performed, and graphs were created with GraphPad Prism software (GraphPad Software, San Diego, CA).

RESULTS

Diurnal Rhythms of PER1-IR in TIDA and PHDA Neurons, but not THDA Neurons

Circadian clock protein IR within NDNs is characterized by dense granular nuclear staining and an absence of staining within the nucleolus (see Fig. 1). PER1-IR nuclei were clearly labeled within TIDA, THDA, and PHDA neurons located within the DMARN, RARN, and PeVN, respectively (Fig. 2, A, C, and E, arrowhead). More than 50% of NDNs within the DMARN, RARN, and PeVN expressed nuclear PER1-IR during the light period. We observed several PER1-IR nuclei that were not double-labeled with TH (Fig. 2, A, C, and E, arrow), as well as several TH-IR neurons that did not express PER1 (Fig. 2, A, C, and E, asterisk). Analysis of PER1-IR within the DMARN of the hypothalamus as a function of CT revealed a main effect of time (F = 4.85, P < 0.05, Fig. 2B). Pairwise comparisons as a function of CT revealed a significant diurnal rhythm of nuclear PER1 expression in TIDA neurons with a peak at CT18 (P < 0.05) compared with a nadir at CT6 (Fig. 2B). Therefore, we observed a significant diurnal rhythm of PER1-IR within TIDA neurons. We have previously determined that TIDA neurons display a diurnal rhythm of DA release with a significant acrophase between CT0 and CT6 (57, 58). Analysis of PER1-IR within the RARN of the hypothalamus did not reveal a significant effect of CT (F = 0.52, P > 0.05, Fig. 2D), suggesting that THDA neurons do not display diurnal rhythms of nuclear PER1 expression (Fig. 2D). Analysis of PER1-IR within the PeVN of the hypothalamus revealed a significant main effect of time of day (F = 4.91, P < 0.05, Fig. 2F). Individual comparisons revealed a significant diurnal rhythm of PER1-IR within PHDA neurons characterized by an acrophase at CT12, compared with basal levels at CT0 (P < 0.05), CT6 (P < 0.01), and CT18 (P < 0.01, Fig. 2F). Together, these data suggest that diurnal rhythms of DA synthesis and release from TIDA and PHDA neurons may be regulated by both SCN afferents and local regulation by nuclear PER expression.

Fig. 2.

PER1 expression in NDNs under a standard 12:12-h light-dark cycle. We observed strong nuclear PER1 expression (red; arrowhead) within tuberoinfundibular dopaminergic (TIDA; green) (A), tuberhypophyseal neuroendocrine dopaminergic (THDA; C), and periventricular hypophyseal dopaminergic (PHDA) neurons (E). Several neurons within these regions expressed TH-IR but failed to show strong PER1 staining (*) or exhibited strong PER1 staining without detectable TH-IR (arrow). PER1 expression peaked in TIDA neurons (B) at CT18 under a standard light-dark cycle. We failed to detect a significant rhythm of PER1 expression within THDA neurons (D). F: PER1 expression peaked at CT12 under a standard 12:12-h light-dark cycle in PHDA neurons. B, D, and F: differing letters indicate significance across circadian time (CT). #Significant peak above basal level regardless of adjacent differences (P < 0.05; see results). Scale bar in A = 50 μm.

Diurnal Rhythms of PER2-IR in TIDA, THDA, and PHDA Neurons

PER2-IR nuclei were clearly labeled within TIDA, THDA, and PHDA neurons located within the DMARN, RARN, and PeVN, respectively (Fig. 3, A, C, E, arrowhead). More than 50% of NDNs within the DMARN, RARN, and PeVN expressed nuclear PER2-IR during the subjective day. We observed several PER2-IR nuclei that were not double-labeled with TH (arrow), as well as several TH-IR neurons that did not express PER2 (Fig. 3, A, C, E, asterisk). Analysis of PER2-IR within TIDA revealed a main effect of time (F = 5.10, P < 0.05, Fig. 3B). Pairwise comparisons exposed a significant diurnal rhythm of PER2-IR within TIDA neurons characterized by peak levels of PER2 expression at CT6 (P < 0.05) and CT12 (P < 0.05), compared with a nadir at CT0 (Fig. 3B). PER2-IR levels returned to basal levels at CT18 (P < 0.05, compared with peak levels at CT12). In contrast with PER1-IR, which peaked at CT18 under a standard light-dark cycle in TIDA neurons, PER2-IR peaked between CT6 and CT12 in animals housed under a standard light-dark cycle (Fig. 2B vs. Fig. 3B). Analysis of PER2-IR within the RARN of the hypothalamus revealed a main effect of time (F = 3.70, P < 0.05, Fig. 3D). Comparisons within animals under a light-dark cycle as a function of time delineate a significant diurnal rhythm of PER2 expression within THDA neurons defined by a significant increase from CT0 to CT6 (P < 0.05), followed by a sustained level throughout the remainder of the subjective day (Fig. 3D). Analysis of PER2-IR within the PeVN of the hypothalamus revealed a significant main effect of time (F = 6.93, P < 0.01, Fig. 3F). Pairwise comparisons within animals housed under a standard 12:12-h light-dark cycle as a function of time revealed a significant diurnal rhythm of PER2 expression with a significant peak at CT6 (P < 0.05, compared with a nadir at CT0), followed by a sustained level of PER2 expression throughout the remainder of the subjective day (Fig. 3F). These data suggest a potential role for PER2 within all three populations of NDNs in the regulation of rhythmic DA release and metabolism.

Fig. 3.

PER2 expression in NDNs under a standard 12:12-h light-dark cycle. We observed strong nuclear PER2 expression (red; arrowhead) within TIDA (green) (A), THDA (C), and PHDA neurons (E). Several neurons within these regions expressed TH-IR but failed to show strong PER2 staining (*) or exhibited strong PER2 staining without detectable TH-IR (arrow). B: PER2 expression peaked in TIDA neurons at CT6 and CT12 under light-dark conditions. D: PER2 expression peaked at CT6 within THDA neurons under light-dark conditions. F: PER2 expression peaked at CT6 under light-dark in PHDA neurons. B, D, and F: differing letters indicate significance across CT. Scale bar in A = 50 μm.

Constitutive Expression of CLOCK Protein in TIDA, THDA, and PHDA Neurons

CLOCK-IR nuclei were intensely labeled within TIDA, THDA, and PHDA neurons located within the DMARN, RARN, and PeVN, respectively (Fig. 4A, C, and E, arrowhead). More than 50% of NDNs within the DMARN, RARN, and PeVN expressed nuclear CLOCK-IR during the subjective day. We observed several CLOCK-IR nuclei that were not double-labeled with TH (arrow), as well as several TH-IR neurons that did not express CLOCK (Figs. 4A, C, and E, asterisk). One-factor analysis of CLOCK-IR within the DMARN (F = 2.03, P > 0.05), RARN (F = 0.05, P > 0.05), and PeVN (F = 3.07, P > 0.05) failed to reveal a significant main effect of time in any regions. Therefore, we did not observe a significant rhythm of CLOCK expression within TIDA neurons (Fig. 4B), THDA neurons (Fig. 4D), or PHDA neurons (Fig. 4F) under a standard 12:12-h light-dark cycle. Thus data from these experiments suggest that CLOCK protein is constitutively expressed within all three populations of NDNs (Figs. 4, B, D, and F). Given the established interactions between the PER family of proteins and the CLOCK protein in SCN neurons, our data suggest a potential role in the regulation of circadian rhythms of gene expression and physiological activity within NDNs. However, without functional evidence, we cannot assume that the circadian clock genes drive circadian rhythms in NDNs in the absence of signals from the SCN.

Fig. 4.

CLOCK expression in NDNs under a standard 12:12-h light-dark cycle. We observed strong nuclear CLOCK expression (red; arrowhead) within TIDA neurons (green) (A), THDA neurons (C), and PHDA neurons (E). Several neurons within these regions expressed TH-IR but failed to show strong CLOCK staining (*) or exhibited strong CLOCK staining without detectable TH-IR (arrow). Constitutive CLOCK expression was observed in TIDA (B), THDA (D), and PHDA (F) neurons under a standard 12:12-h light-dark cycle. Similar letters indicate no significant differences across CT. Scale bar in A = 50 μm.

Acute Knockdown of per1, per2, and clock mRNA Expression Disrupts Diurnal Rhythms of Drinking Behavior

Before antisense knockdown of circadian clock gene expression in NDN, we attempted to verify the expression of circadian clock gene mRNA within the ARN and pituitary gland of OVX rats. As shown in Fig. 5, per1, per2, bmal1, and clock gene mRNA are expressed within the SCN, ARN, and pituitary of OVX rats. Gene products were detected with RT-PCR as described in materials and methods and were verified by DNA-sequencing at our core facility. To verify a functional circadian clock and to determine a reference point for AS-ODN injection and tissue collection, animals were placed in our Lickometer device following stereotaxic surgery. The beginning of the 12-h activity period, identified as CT12, was determined 2 days before tissue collection for each animal and averaged to predict the onset of activity on the following day. We predicted CT12 under entrained conditions with an assumed error of 10–15 min, given a variance in activity onset among our rats (generally 10–15 min from cycle to cycle), which we consider acceptable with a sampling frequency of 2–4 h. To determine the effects of AS-ODN on the rhythm of drinking activity, a pair of animals were placed in a standard 12:12-h light-dark cycle for a single day, followed by injection of AS-ODN or RS-ODN on the second day under either condition (Fig. 6, grey arrowhead). Animals were allowed to remain in the Lickometer device for several days following AS-ODN or RS-ODN injection. In both AS-ODN-and RS-ODN-injected animals under a standard 12:12-h light-dark cycle, CT12 was ∼1730 + 0.25 h (Fig. 6). AS-ODN-treatment eliminated the diurnal rhythm of drinking activity when compared with a RS-ODN controls, which recovered after ∼72 h (Fig. 6). Drinking behavior dropped to a minimum during this period, marked by an average number of water bottle licks <10 licks/10 min bin. These data suggest that AS-ODN cocktail transiently disrupted the function of the central molecular oscillator within the SCN.

Fig. 5.

RT-PCR amplification of per1, per2, clock, and bmal1 mRNA from suprachiasmatic nucleus (SCN), arcuate nucleus (ARN), and pituitary from overiectomized rats. Animals were killed at ∼8 zeitgeiber time under hypercapnic conditions. Individual 2-mm thick tissue punches containing the SCN or the medial basal hypothalamus (including the ARN and median eminence) and the entire anterior lobe (AL) of the pituitary were dissected and used for total RNA extraction. RT using oligo(dT) primers was carried out to amplify cDNA. Gene-specific primers for per1, per2, clock, and bmal were used to amplify cDNA. PCR products were separated with agarose gel electrophoresis and stained with ethidium bromide. Ld, nucleic acid markers.

Fig. 6.

Injection of clock gene antisense deoxyoligonucleotides (AS-ODN) disrupts rhythms of drinking behavior. Injection of random-sequence oligos ODN (RS-ODN; gray arrowheads) into the SCN failed to disrupt diurnal rhythms of drinking activity. Following AS-ODN injection into the SCN, diurnal rhythms of drinking activity were abolished for up to 72 h, marked by a constitutively low amount of drinking. Dark arrowheads indicate approximately CT12. Data shown are individual animals representative of each treatment group. L:D, light-dark.

AS-ODN cocktail against per1, per2, and clock mRNA Reduces PER1/2 and CLOCK Protein Expression in the SCN

Following injection with AS- or RS-ODN, animals were killed as described above, SCN tissue blocks were collected, and clock protein levels were analyzed with Western blot. Given the necessity for acute death of animals for DA turnover analysis, we could not perfuse the animals and, therefore, could not verify the physical placement of cannulae in the SCN. However, we verified the effects of AS-ODN on the level of clock proteins in the SCN with Western blot analysis. Thus animals that received AS-ODN but did not show a significant (>50%) knockdown of circadian clock proteins were not included in the experiment. Moreover, we did perfuse animals that were treated with AS- or RS-ODN and were analyzed for drinking behavior (as described above) and stained them for PER1/2 and CLOCK to verify successful cannula placement in these representative animals (data not shown). One-factor analysis of PER1/2 and CLOCK expression within SCN tissue extracts revealed a significant main effect of ODN treatment (F = 57.19, P < 0.01). Densitometry of Western blots containing SCN samples from animals injected with AS- or RS-ODN revealed significant decreases in PER1/2 and CLOCK protein levels within the SCN following AS-ODN injection (see Fig. 7A). As shown in Fig. 7, CE, SCN tissue samples from AS-ODN-injected rats contained significantly less PER1 (P < 0.01, Fig. 7C), PER2 (P < 0.01, Fig. 7D), and CLOCK (P < 0.01, Fig. 7E) protein than RS-ODN-injected controls. Tissue samples from the piriform cortex, a brain region known to express PER1/2 and CLOCK but located well outside of the hypothalamus, were collected and analyzed for CLOCK protein expression as a positive control (Fig. 7B). Analysis of CLOCK expression as a function of ODN treatment failed to reveal a significant effect of AS-ODN injection in the SCN on CLOCK expression in the piriform cortex (Fig. 7F, F = 6.037, P > 0.05). Similar observations were found for PER1 and PER2 expression within the piriform cortex (data not shown). These data support a localized and specific reduction of PER1/2 and CLOCK expression within the SCN following precise AS-ODN injection.

Fig. 7.

Injection of clock gene AS-ODN reduces PER1/2 and CLOCK expression within the SCN. Injection of AS-ODN significantly reduced PER1 (A and C), PER2 (A and D), and CLOCK (A and E) expression within the SCN compared with RS-ODN-treated controls. B and F: SCN injection of AS-ODN failed to reduce CLOCK gene expression within the piriform cortex. Similar results were observed for PER1/2 (data not shown). A and B: blots are single tissues from representative animals of each group. C and F: bars represent pooled values or relative (r) protein abundance units for all animals used in each group (AS- and RS-ODN treated, n = 24). In A and B, arrows indicate nonspecific binding of CLOCK primary antibody, whereas arrowheads indicate CLOCK specific staining. *Significant effects of AS- vs. RS-ODN treatment (P < 0.01).

Acute Knockdown of PER1/2 and CLOCK Protein Expression in the SCN Differentially Affects Diurnal Rhythms of PRL and CORT Secretion

Two-factor analysis of serum PRL levels in AS-ODN- and RS-ODN-treated animals maintained under a standard light-dark cycle as a function of treatment and CT failed to reveal a significant main effect of time (F = 0.84, P > 0.05) or treatment (F = 1.949, P > 0.05) or an interaction between time × treatment (F = 0.71, P > 0.05). In agreement with previous reports from our laboratory and others (12, 38), we failed to detect a significant diurnal rhythm of serum PRL in control RS-ODN-treated OVX rats. Moreover, treatment with AS-ODN failed to affect the timing or magnitude of PRL secretion in OVX animals under a standard light-dark cycle (Fig. 8A). Thus we cannot conclude from these data that AS-ODN-induced disruption of the molecular oscillator exerted a significant effect on PRL secretion in the OVX rat. Two-factor analysis of serum CORT levels in AS-ODN and RS-ODN-treated animals maintained under a standard light-dark cycle as a function of time and treatment revealed a significant effect of time (F = 3.04, P < 0.05) and treatment (F = 7.48, P < 0.01), but not an interaction between time × treatment (F = 1.12, P > 0.10). Individual comparisons within RS-ODN-treated animals under a 12:12-h light-dark cycle established a significant rhythm of CORT secretion with a rise to peak level above baseline at CT24 (P < 0.05 when compared with basal levels at CT6 and CT9, Fig. 8B). Treatment with AS-ODN eliminated this diurnal rhythm of CORT secretion (P > 0.05 across time within AS-ODN-treated animals under a standard 12:12-h light-dark cycle). Thus AS-ODN-treatment disrupted the transcriptional feedback loop in SCN neurons, leading to disruption of the diurnal rhythm of CORT secretion in the OVX rat.

Fig. 8.

Clock gene knockdown disrupts diurnal rhythms of corticosterone (CORT) secretion, but not prolactin (PRL) secretion. A: serum PRL did not display a significant diurnal rhythm in RS-ODN animals maintained under a 12:12-h light-dark cycle. Treatment with AS-ODN failed to affect PRLS secretion in these animals, as we did not observe a significant difference at any time between AS- or RS-injected animals. B: serum CORT exhibited a significant diurnal rhythm in RS-ODN-treated rats with a rise to peak at CT0 that was abolished by AS-ODN treatment. Data from RS (dashed line)-treated animals are double plotted, whereas data from AS animals (solid line) are single plotted (right). Data are mean values ± SE for n = 4 rats. Differing letters within treatment (a, RS-ODN treated; A, AS-ODN treated) indicate significant effects of time (P < 0.05). *Significant peaks above basal level regardless of adjacent differences (P < 0.05).

Acute Knockdown of PER1/2 and CLOCK Protein Expression in the SCN Differentially Affects Diurnal Rhythms of DA Turnover in NDNs and DA Concentration Within the AL

Two-way ANOVA of DA turnover within the ME of animals maintained under a 12:12-h light-dark cycle as a function of time and ODN treatment revealed a significant effect of time (F = 12.97, P < 0.001) and treatment (F = 20.25, P < 0.001), but not the interaction between time × treatment (F = 0.88, P > 0.05). Individual comparisons within RS-ODN-treated rats under a standard light-dark cycle revealed a significant biphasic diurnal rhythm of DA turnover defined by peaks at CT0 (P < 0.05) and CT15 (P < 0.01) above basal levels at CT9 and CT18 (Fig. 9A). Treatment with AS-ODN abolished the peak at CT15 (P < 0.05 when compared with RS-ODN rats), resulting in a U-shaped rhythm with a single peak above baseline at CT0 (P < 0.05, Fig. 9A). Thus AS-ODN modified the magnitude of the diurnal rhythm of DA turnover in the ME in the early night when compared with RS-ODN controls.

Fig. 9.

Clock gene knockdown differentially affects diurnal rhythms of dopamine (DA) turnover in the median eminence (ME), neuronal lobe (NL), and intermediate lobe (IL), and DA concentration in the anterior lobe (AL). A: DA turnover within the ME displayed a significant diurnal rhythm in RS-ODN-injected controls that was influenced by AS-ODN injection into the SCN. B: DA turnover within the IL displayed a significant diurnal rhythm in both RS-ODN and AS-ODN injected rats. C: DA turnover within the NL failed to display a rhythm in either RS-ODN-treated controls or AS-ODN-injected rats. D: DA concentration in the AL of RS-ODN-treated animals failed to exhibit a significant diurnal rhythm. Treatment with AS-ODN induced a diurnal rhythm of DA in the AL under a 12:12-h light-dark cycle. In AD, data from RS-ODN-treated animals are double plotted (dashed line), while data from AS-ODN-treated animals are single plotted on the right (solid line). For AC, DA turnover is expressed as the DOPAC-to-DA ratio within DA synaptic terminals. D: shows total DA in the AL of the pituitary, which contains few, if any, DAergic synapses. Data are means ± SE for 4 rats. Differing letters within treatment (a, b: RS-ODN treated; A, B: AS-ODN treated) indicate significant effects of time (P < 0.05). *Significant peaks above baseline within treatment regardless of adjacent differences (P < 0.05); #significant effects of ODN-treatment (P < 0.05).

Two-way ANOVA of DA turnover within the IL of animals maintained under a 12:12-h light-dark cycle as a function of time and ODN treatment revealed a significant effect of treatment (F = 3.90, P < 0.05) and an interaction between time and treatment (F = 4.10, P < 0.001), but no main effect of time (F = 2.15, P = 0.07). Pairwise comparisons within RS-ODN-treated controls under a standard 12:12-h light-dark revealed a significant diurnal rhythm of DA turnover in the IL (Fig. 9B). DA turnover within the IL of RS-ODN controls under a 12:12-h light-dark cycle exhibited an acrophase at CT0 (P < 0.05) compared with a nadir at CT15. AS-ODN-treatment significantly adjusted the shape of this rhythm by increasing DA turnover within the IL at CT0 (P < 0.001, AS-ODN vs. RS-ODN) and decreasing DA turnover at CT6 (P < 0.05, AS-ODN vs. RS-ODN, Fig. 9B). Thus AS-ODN affected the magnitude and timing of the diurnal rhythm of DA turnover within the IL of OVX rats by advancing the peak of DA turnover from CT6 to CT0. From these data, we can conclude that AS-ODN treatment modulates rhythms of DA turnover within the IL. These data agree with the results for DA turnover from the ME presented above. Because these two populations displayed similar rhythms of DA turnover in previous experiments (58) it is not surprising that they exhibit similar responses with respect to AS-ODN treatment (compare A and B of Fig. 9).

Two-factor analysis of DA turnover within the NL of animals maintained under a standard light-dark cycle as a function of time and ODN treatment did not show a significant effect of time (F = 1.76, P > 0.05) and treatment (F = 1.86, P > 0.05), or a significant interaction between time × treatment (F = 1.11, P > 0.05). In contrast with our prior results (58), pairwise comparisons within RS-ODN-treated rats maintained in a standard 12:12-h light-dark cycle failed to reveal significant rhythms of DA turnover in the NL (Fig. 9C). Moreover, animals treated with AS-ODN also failed to display a significant diurnal rhythm of DA turnover in the NL (Fig. 9C). Thus although AS-ODN injection into the SCN failed to significantly affect the timing or magnitude of DA turnover in the NL, our inability to detect a rhythm of DA turnover in the current experiment obviates any conclusions regarding the effects of AS-ODN on the THDA neurons.

Two-factor analysis of DA concentration within the AL of animals maintained under a standard 12:12-h light-dark cycle as a function of time and ODN treatment revealed a significant effect of time (F = 2.80, P < 0.05) and treatment (F = 8.84, P < 0.001), but not an interaction between time × treatment (F = 1.18, P > 0.05). Comparisons as a function of time within RS-ODN-treated rats under a standard light-dark cycle failed to reveal a significant diurnal rhythm of DA concentration within the AL (Fig. 9D). AS-ODN injection into the SCN unmasked a significant diurnal rhythm of DA concentration in the AL with a significant acrophase between CT6 and CT9 (P < 0.05), when compared with a nadir at CT0 (Fig. 9D). The diurnal rhythm of DA concentration in the AL revealed by AS-ODN-treatment represents a significant increase above RS-ODN controls at both CT6 (P < 0.05) and CT9 (P < 0.01).

DISCUSSION

The purpose of these experiments was to determine the pattern of PER1/2 and CLOCK expression within NDNs and ascertain the functional relationship between rhythms of clock gene expression within NDNs and previously determined rhythms of DA turnover within TIDA, THDA, and PHDA neurons. Previously, we reported significant circadian rhythms of DA turnover in the OVX rat (58). We hypothesized that these circadian rhythms of DA release, which dictate the timing of the ovarian steroid-induced PRL surge on the afternoon of proestrus, are facilitated by autonomous rhythms of clock gene expression within NDNs. We attempted to validate this hypothesis using antisense knockdown to disrupt the transcriptional feedback loops regulating circadian rhythms of activity in SCN neurons. Given the established relationship between the SCN and DAergic target neurons in the ARN, we hypothesized that disruption of the circadian clock in the SCN would abolish diurnal rhythms of DA release from NDN synaptic terminals.

PER1/2 and CLOCK-IR nuclei are clearly labeled within TH-IR neurons in the DMARN, RARN, and PeVN. Generally, circadian clock protein-IR was limited to the nucleus and is readily distinguishable from TH-IR cytoplasmic staining. Only those neurons expressing a clear clock protein-IR nucleus were considered clock protein-IR cells. While only TH-IR and clock protein/TH-double-labeled cells were counted in the current experiment, we observed numerous PER1/2 and CLOCK-IR nuclei within the analyzed regions that were not TH-IR. Under a standard light-dark cycle, PER1-IR within TIDA neurons displayed a diurnal rhythm with a peak late in the subjective day. In addition, PER2 expression displayed a diurnal rhythm of expression with significant peaks at CT6 and CT12. We have previously reported a diurnal rhythm of DA turnover within TIDA neurons with a biphasic pattern, defined by peaks at CT6 and CT12 (57, 58). Thus PER2 expression, but not PER1 expression, peaked in parallel with the time of peak DA turnover within these neurons. Like the SCN, TIDA neurons express CLOCK-IR in a constitutive manner under a standard 12:12-h light-dark cycle. Based on these cumulative data, we can assume that TIDA neurons may act as dampened oscillators, with light-entrained or light-activated rhythms of clock gene expression. Recent findings agree with this conclusion, suggesting that neurons within the ARN are unable to express free-running rhythms of PER expression in isolated cell culture for more than 2–3 cycles (75, 76).

Unlike TIDA neurons, THDA neurons failed to exhibit a diurnal rhythm of PER1 expression, but did display a diurnal rhythm of PER2 expression with a significant peak at CT6. Like TIDA neurons, THDA exhibit a diurnal rhythm of DA turnover with a biphasic pattern defined by peaks at CT6 and CT12. Therefore, the rhythm of PER2 expression, but not PER1 expression, within THDA neurons corresponds to the rhythm of DA turnover observed for these neurons under a standard light-dark cycle. These data are surprising, given our previous report that THDA neurons exhibit clear diurnal rhythms with significant peaks near CT6 and CT12 (57, 58). It remains to be seen whether PER1 and PER2 have redundant roles in the regulation of molecular feedback loops of circadian clock genes and/or DA synthetic enzyme expression in NDNs. Additional evidence regarding the molecular interactions of PER proteins in NDNs is required to either confirm or refute this hypothesis. Like TIDA neurons, THDA neurons express CLOCK-IR in a constitutive manner under light-dark conditions. In PHDA neurons, both PER1 and PER2 expression displayed diurnal rhythms with significant peaks at CT12 and CT6, respectively. We have shown that DA turnover within the IL peaks between CT10 and CT14, followed by a trough at CT18 (57, 58). In agreement with data from both TIDA and THDA neurons, PER expression patterns within PHDA neurons correspond to the timing of DA turnover within the IL, both of which occur primarily during the light phase. Like both TIDA and THDA neurons, PHDA neurons express CLOCK in a constitutive manner. Several experiments have concluded that CLOCK protein is also constitutively expressed within the SCN (7, 26, 28, 44, 54, 60). Therefore, it would appear that constitutive CLOCK expression is a defining feature of the molecular clock found within both the central circadian oscillator in the SCN and its primary central targets, including the NDNs of the hypothalamus. Thus our results suggest a potential role for the core molecular feedback loops driving the core circadian clock in the timing of molecular and physiological events in NDNs.

Recently, Kriegsfeld et al. (33) reported a diurnal rhythm of PER1 expression in neuroendocrine cells of the ARN in the female mouse, using the same primary antiserum, with a significantly greater number of PER1::GFP/TH double-labeled cells at CT10 than CT22. These data suggest that PER1 expression peaks in the latter portion of the subjective day and reaches a nadir at or near the middle of the subjective night (CT18–22, Ref. 33). Furthermore, the results of Bae and colleagues (3, 16, 25) revealed that PER1-IR peaked within the SCN at or near CT12. These data would suggest that the peak in PER1-IR in NDNs parallels the peak of PER1 expression within the SCN. However, experiments indicate that PER1 expression in peripheral tissues peaks ∼6–12 h after PER expression within the SCN. Our data show that PER1-IR peaks at CT18, ∼6–12 h after the peak of PER1 expression we and others have observed within the SCN (3, 16, 25).

To our best knowledge, we are the first to examine the rhythmic expression of these proteins in the NDNs of steroid-depleted OVX rats. Several experiments suggest that ovarian steroid receptors are expressed in SCN neurons and that steroids exert direct effects on gene expression and physiological activity within the SCN (34, 46, 47, 49, 63, 64). Therefore, it is difficult to predict what effect removal of endogenous ovarian steroids would have on clock gene expression within the SCN and NDNs. We can assume that light-entrained rhythms of PER expression, which are highly dependent on neural input from the retina, are likely less dependent on the influence of ovarian steroids. Of course, we cannot rule out the dramatic effects of ovarian steroid hormone withdrawal on steroid-sensitive neuronal oscillators within the SCN and/or NDNs. Future experiments using steroid hormone replacement models will address this fundamental issue.

Several hypotheses could be offered to explain the function of the diurnal rhythms of clock protein expression we have observed. We have determined in previous experiments that NDNs express VPAC2 receptors that are affected by the level of circulating ovarian steroid hormones (19). Furthermore, we have shown that disruption of VIP peptide expression within the SCN affects the activity of NDNs under a standard 12:12-h light-dark cycle (20). Numerous studies have shown that VIP peptide displays a diurnal rhythm of expression within the SCN characterized by a significant increase in VIP expression in the late subjective night between CT18 and CT22 (4, 27, 51, 53, 62, 6567, 77). We can conclude from these studies that VIP release from SCN neurons entrains the activity of NDNs in the late subjective night. Moreover, additional evidence suggests VIP induces PER1/2 expression in the SCN during the late subjective night (50). Therefore, we can assume that VIP, released from SCN afferents within the DMARN, RARN, and PeVN binds to VIP type-2 receptors and differentially activates PER1/2 expression through increased intracellular cAMP- and CREB-mediated signaling (36). However, this hypothesis cannot explain the phase relationships between PER1 and PER2 expression in TIDA neurons or the absence of PER1 expression in THDA neurons. Additional experiments are necessary to verify the functional redundancy of PER1/2 within NDNs. Although we lack the evidence, we cannot rule out a role for arginine vasopressin of SCN origin in our model. Further experiments are necessary to delineate the precise role of both VIP and AVP in the activation and maintenance of clock gene expression within NDNs.

In the present study, we have determined the effects of transient per1, per2, and clock mRNA knockdown on the light-entrained rhythms of PRL secretion, CORT secretion, DA concentration within the AL of the pituitary, and DA turnover within NDN synaptic terminals. We have attempted to ascertain the influence of clock gene-controlled physiological activity and gene expression within the SCN on the rhythm of DA release from NDNs. Using a cocktail of per1, per2, and clock AS-ODN, we have generated a temporary “molecular” lesion of the central circadian oscillator. Based on data from our previous experiments (57, 58), we hypothesized that NDNs may continue to oscillate with a diurnal rhythm in the absence of photoperiodic cues from the SCN. Given that we have already determined that both PER1 and PER2 display diurnal rhythms of expression within TIDA, THDA, and PHDA neurons, we hypothesized that clock gene expression within NDNs may act to facilitate diurnal rhythms of DA release from NDNs in the absence of afferent input from the SCN. A negative effect of clock gene AS-ODN injection in the SCN on diurnal rhythms of DA turnover in NDNs would indicate the capacity of the NDN to behave as a semiautonomous or damped circadian oscillator.

Injection of AS-ODN into the SCN successfully disrupted the diurnal rhythm of drinking activity in the OVX rat. Interestingly, treatment with AS-ODN appeared to induce a long-lasting arrhythmia in drinking behavior for up to 72 h. Although we observed a long duration effect on drinking behavior, we failed to see a similar decrease in PER1/2 and CLOCK expression in SCN tissue extracts beyond 12–18 h (data not shown). Therefore, we can conclude that our transient molecular knockdown exerted long-lasting effects (up to 72 h) on downstream targets of the SCN, although the overall level of clock protein expression returned to normal within 24 h. Because we did not measure the rhythm of PER1/2 or CLOCK expression in the SCN after AS or RS-ODN injection, we cannot rule out significant long-lasting effects of our ODNs on the phase and/or period of the oscillations.

As in previous experiments, we failed to detect a rhythm of PRL secretion in the OVX rat (12, 38). Our inability to detect a significant PRL secretory rhythm precludes determination of AS-ODN treatment affects on PRL secretion. Additional experiments, using both steroid-primed and cycling rats, could provide additional insight into the role of clock genes within the SCN in the control of PRL secretion. Unlike PRL, serum CORT exhibited a significant diurnal rhythm in RS-ODN-treated controls that was disrupted by AS-ODN treatment. Data from numerous experiments suggest that AVPergic afferents of SCN origin terminate on corticotrophin-releasing hormone neurons within the medial parvicellular PVN (6, 3032, 56, 72, 73). Further evidence suggests that AVP and corticotrophin-releasing hormone mRNA are synthesized in PVN neurons with a distinct circadian rhythm, possibly under the direct control of SCN afferents (74). Thus our ability to disrupt the diurnal rhythm of CORT secretion in OVX rats, like the effects we observed on drinking activity, support successful molecular lesion of the SCN with our AS-ODN cocktail.

In agreement with our previous experiments (57, 58), TIDA and PHDA neurons displayed diurnal rhythms of DA turnover with significant peaks in the early subjective day (CT0 in TIDA and PHDA neurons) and early subjective night (CT15 in TIDA neurons). Treatment with AS-ODN against per1/2 and clock mRNA significantly adjusted the magnitude of DA turnover, leading to a significant change in the shape of the rhythm. Treatment with AS-ODN eliminated the second peak of DA turnover within TIDA neurons that occurred at CT15 (∼2030 h) but failed to affect the peak of DA turnover at CT0. AS-ODN treatment advanced the acrophase of DA turnover in the IL from CT6 to CT0. Moreover, AS-ODN treatment increased DA turnover at CT15, such that it no longer represents the absolute nadir of DA turnover, as it does in RS-ODN controls. These data suggest that AS-ODN adjusted the magnitude of DA release but failed to completely eliminate the diurnal rhythm. Furthermore, these results suggest an overall increase in the level of DA release, indicating disruption of a potential DA release-inhibiting factor of SCN or hypothalamic origin. Nonetheless, this effect correlates with our previous experiments, suggesting that treatment with AS-ODN against VIP affected the pattern of immediate early gene expression within NDNs (20). Several studies suggest that AVP expression within neurons of the SCN shell or dorsomedial SCN displays a free-running endogenous rhythm under the direct control of CLOCK/BMAL1 enhancers, while VIP expression exhibits a light-entrained, but not free-running, circadian rhythm (29, 53). However, recent studies suggest that VIP mRNA synthesis is under the direct control of the molecular oscillator in the mouse (9). Although we cannot rule out the potential influence of AVP from the SCN shell, we can assume that, if VIP is the primary neurotransmitter of the SCN-NDN tract, that light-induced or endogenous rhythms of VIP release in the mediobasal hypothalamus are responsible for similar rhythms of DA release from NDNs. Within the AL, we failed to detect a significant rhythm of DA concentration within RS-ODN control animals. However, AS-ODN injection induced a significant diurnal rhythm with peaks between CT6 and CT9. As mentioned above, this response is not expected, given the generally variable, basal level of DA release from all three populations in the OVX rat observed in previous experiments (12). Variation in the rhythmic release of DA from each individual population, as a result of ovarian steroid hormone withdrawal and the absence of a significant PRL rhythm may result in the distinct, albeit dissociated rhythm we have detected here. Overall, these data suggest that AS-ODN treatment may reveal a significant diurnal rhythm of DA turnover and total DA concentration by removing rhythmic inhibition of DA release from a presently unknown inhibitory pathway.

We can conclude, therefore, from these studies and others that NDNs may function as damped oscillators, expressing clock genes in a light-dark cycle under the direct influence of input from the SCN. Additional experiments are needed to gain a more thorough understanding of the role for each clock gene product in NDNs regarding the synthesis and release of DA. Furthermore, future experiments will attempt to garner a better understanding of the effects of ovarian steroids on the molecular clock in both the SCN and NDNs. Our current report indicated the presence of circadian clock gene products in numerous non-DAergic neurons within the arcuate and periventricular nuclei. Future experiments will provide us with a better understanding of the form and function of these neurons, as well as their role as potential target oscillators in the hypothalamus. For example, we cannot make a definitive claim regarding the function of NDNs as damped oscillators without evidence that they display free-running and cell-autonomous rhythms of gene expression and electrical activity. Our data show that AS-ODN treatment significantly reduced PER1/2 and CLOCK expression within NDNs and affected the diurnal rhythms of drinking behavior, CORT secretion, and NDN activity in the OVX rat. Therefore, our current data and previous evidence suggest that the diurnal rhythms of DA release in TIDA, THDA, and PHDA neurons are semidependent on rhythmic output from SCN neurons. Evidence suggests that disruption of clock-controlled genes, such as VIP, either directly via VIP AS-OSN as in previous experiments (20) or indirectly through disruption of per1, per2, and clock transcription (present study), may modulate, but not abolish, rhythms of DA release. This hypothesis relies on information regarding the relationship between clock gene expression in SCN neurons and VIP mRNA synthesis (20). Further experiments are needed to strengthen the potential relationship between CCG in NDNs, VIP synthesis and release from SCN afferents and DA release from NDN synaptic terminals.

GRANTS

This work was supported by National Institutes of Health Grants DK-43200 and DA-19356.

Acknowledgments

Present addresses: M. T. Sellix, Biology Department, PO Box 400328, University of Virginia, Charlottesville, VA 22904-4328; M. Egli, Space Biology Group, Swiss Federal Institute of Technology Zurich (ETHZ), Technoparkstrasse 1, CH-8005 Zurich, Switzerland.

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract