|Home | About | Journals | Submit | Contact Us | Français|
Short interfering RNAs (siRNA) targeting prepro-orexin mRNA were microinjected into the rat perifornical hypothalamus. Prepro-orexin siRNA-treated rats had a significant (59%) reduction in prepro-orexin mRNA compared to scrambled siRNA-treated rats 2 days postinjection, whereas prodynorphin mRNA was unaffected. The number of orexin-A-positive neurons on the siRNA-treated side decreased significantly (23%) as compared to the contralateral control (scrambled siRNA-treated) side. Neither the colocalized dynorphin nor the neighbouring melanin-concentrating hormone neurons were affected. The number of orexin-A-positive neurons on the siRNA-treated side did not differ from the number on the control side 4 or 6 days postinjection. Behaviourally, there was a persistent (~ 60%) increase in the amount of time spent in rapid eye movement (REM) sleep during the dark (active) period for 4 nights postinjection, in rats treated with prepro-orexin siRNA bilaterally. This increase occurred mainly because of an increased number of REM episodes and decrease in REM-to-REM interval. Cataplexy-like episodes were also observed in some of these animals. Wakefulness and NREM sleep were unaffected. The siRNA-induced increase in REM sleep during the dark cycle reverted to control values on the 5th day postinjection. In contrast, the scrambled siRNA-treated animals only had a transient increase in REM sleep for the first postinjection night. Our results indicate that siRNA can be usefully employed in behavioural studies to complement other loss-of-function approaches. Moreover, these data suggest that the orexin system plays a role in the diurnal gating of REM sleep.
Orexins (also known as hypocretins; de Lecea et al., 1998; Sakurai et al., 1998) are peptides that are produced mainly in the perifornical region of the hypothalamus (PFH). Two isoforms of orexins (orexin-A and -B) are derived from proteolytic cleavage of a precursor peptide (prepro-orexin) and they act on two G-protein-coupled receptors.
Although orexins are involved in multiple functions, the most studied one is their role in the regulation of sleep–wakefulness and in the aetiology of sleep-related disorders, especially narcolepsy (Taheri et al., 2002). Increased rapid eye movement (REM) sleep in the dark period and cataplexy is generally observed in orexin- or orexin receptor-knockout mice, as well as in mice or rats in which a truncated Machado–Joseph disease gene product, ataxin-3, has been used to destroy the orexin neurons (Chemelli et al., 1999; Hara et al., 2001; Willie et al., 2003; Beuckmann et al., 2004). Chemical destruction of neurons containing orexin receptors in the PFH had similar effects (Gerashchenko et al., 2001). Although these experiments provide much insight into the roles of orexins in sleep, they also have potential confounding factors. Developmental compensation is always a concern in constitutive gene-knockout studies. Chemical lesions destroy other cells, including melanin-concentrating hormone (MCH) neurons, which also participate in REM sleep regulation (Verret et al., 2003). While ataxin-3-induced cell death is highly specific, it destroys other neuromodulators present in orexinergic neurons such as dynorphin (Chou et al., 2001), which works with orexins to indirectly excite the histaminergic system (Eriksson et al., 2004).
RNA-mediated interference is a sequence-specific gene suppression induced by double-stranded RNA (Fire et al., 1998). Small double-stranded RNAs having 21–23 nucleotides with a symmetric overhang of two or three nucleotides are capable of inducing RNA interference. These double-stranded RNAs are often termed short interfering RNAs (siRNAs; Elbashir et al., 2001).
The finding that siRNA mediates RNA interference forms the basis of a powerful new technique to complement the traditional arsenal of loss-of-function techniques. siRNA avoids the laborious creation and maintenance of genetically modified animals and the potential pitfall of developmental compensation in constitutive knockout animals. Unlike inducible knockouts, siRNA effects are reversible, with a duration usually starting 1 day after administration and lasting for several days, providing a good window for studying the function of the targeted gene (Dillon et al., 2004; Ryther et al., 2005). Unlike traditional lesion techniques, siRNA does not affect neighbouring cells and, unlike techniques causing cellular degeneration, such as ataxin-3, siRNA is more specific, not affecting all genes. Finally, compared with antisense, a technology we have previously used, siRNA appears to be more effective and specific and does not require potentially toxic chemical modifications, and its effects last longer (Hough et al., 2003; Miyagishi et al., 2003)
Although widely used in vitro, siRNA applications in in vivo mammalian brain are relatively rare and, to our knowledge, the use of siRNA is novel in sleep research and in studies of the orexin system. In this study, we examined the effect of siRNA targeting the prepro-orexin gene on orexin-A immunoreactivity and sleep–wakefulness in rats.
Our first two sets of experiments were designed to verify the specificity of prepro-orexin siRNA treatment and to examine the degree of orexin peptide and mRNA knockdown, respectively, following prepro-orexin siRNA administration. Our third experiment was designed to evaluate the effect of prepro-orexin siRNA treatment on the behavioural states of the animals.
Adult Sprague–Dawley male rats (300–350 g; Charles River Laboratories, Wilmington, MA, USA) were housed under 12 : 12 h light : dark cycle (lights on at 07.00, off at 19.00 h) at 22 ± 1 °C. Food and water were provided ad libitum. All animals were treated in accordance with the Association for Assessment of Laboratory Animal Care policy on care and use of laboratory animals and all efforts were made to minimise the number of animals used. All experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Animal Research Committee of the Boston Veterans Administration Healthcare System.
For experiments 1 and 3, rats were anesthetized with an i.p. cocktail of ketamine (90 mg/kg) and xylazine (10 mg/kg). They were then killed with an overdose of pentobarbital (100mg/kg) followed by transcardiac perfusion. For experiment 2, rats were anesthetized with isoflurane and decapitated for tissue collection.
A pool of three siRNAs (equal concentration) was used (Table 1). All siRNAs were designed individually to target prepro-orexin (prepro-orexin siRNA). A corresponding control pool of three siRNAs with scrambled sequences and no homology to known rat genes (scrambled siRNA) was used. All siRNAs were annealed and HPLC-purified (Ambion, Austin, TX, USA). A pool of siRNAs was used because it largely circumvents the task of testing each siRNA individually, reduces possible nonspecific effects (Dharmacon Co., private communication) and may increase target knockdown (Akaneya et al., 2005).
We first verified the knockdown of prepro-orexin peptide by performing orexin-A immunohistochemistry. Next, we verified the specificity of prepro-orexin siRNA effect by double-labelling of orexin-A and MCH (localized to the same region, but not colocalized) and orexin-A and dynorphin B (colocalized in orexin neurons). Finally, we quantified the degree of orexin peptide knockdown by counting orexin-A-positive neurons.
One of the major limitations of using immunohistochemical techniques to count the number of orexin-positive neurons from sampled sections is the potential variability due to differences in the site of injection, differences in the number of orexin-positive neurons in different sections and differences in the immunohistochemical staining. Therefore, for immunohistochemistry, we performed both the siRNA and the scrambled treatment in the same animal on homotopic contralateral PFH sites. In each animal, we counted orexin neurons in two sections both from the siRNA-treated ipsilateral side and the scrambled siRNA-treated contralateral side and compared them with each other. We believe that this experimental design controls for any variability in the number of orexin neurons between animals in sampled sections.
Under general anaesthesia, prepro-orexin siRNAs were unilaterally injected (0.03 nmol in 0.3 μL water) into one side of the orexinergic PFH using a Hamilton syringe. Scrambled siRNA was injected on the contralateral side. Coordinates (Paxinos & Watson, 1998) for the PFH target site were: AP, −3.3; ML, 1.5; and DV, 8.3 mm. Based on our initial, preliminary dose-finding immunohistochemistry and behavioural state data, the maximal prepro-orexin siRNA effect was found to be on day 2 postinjection but returned to normal values on day 5 postinjection. Therefore, we killed the animals at three different time points, namely days 2, 4 and 6, and verified the effect and specificity of prepro-orexin siRNA. For histology, rats were deeply anaesthetized with sodium pentobarbital and transcardially perfused with saline followed by 4% formaldehyde. The brain was isolated and placed overnight in 4% formaldehyde then transferred to 20% sucrose for cryoprotection, and coronal sections (30 or 40 μm) were cut on a freezing microtome.
Orexin-A immunochemistry was performed in four animals, killed 2 days postinjection. One-third of the sections were used for orexin immunohistochemistry and another one-third for orexin-A and MCH double labelling. Free-floating sections were successively incubated in 0.3% Triton X-100 (2 h), 3% normal donkey serum in PBS (1 h), and overnight at 4 °C with rabbit antiorexin-A primary IgG antibody (1 : 2000; Peninsula Laboratories, San Carlos, CA, USA) and 0.3% Triton X-100. After incubation in secondary donkey antirabbit IgG antibody (1 : 400; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1 h, sections were stained using the Vectastain ABC kit and the DAB substrate kit (Vector Laboratories, Burlingame, CA, USA) following the manufacturer's instructions.
Sections were incubated in 0.25% Triton X-100 and 5% normal donkey serum in PBS for 1 h followed by streptavidin/biotin blocking 15 min each (Streptavidin/Biotin Blocking Kit (Vector). A mixture of rabbit antiorexin-A IgG antibody (1 : 300, Peninsula Laboratories, San Carlos, CA, USA) and chicken anti-MCH IgY antibody (1 : 300, Chemicon International, Temecula, CA, USA) were used to incubate sections overnight. The sections were then incubated with biotinylated donkey antirabbit IgG antibody (1 : 300, Jackson ImmunoResearch) for 1 h, Cy3 conjugated streptavidin (1 : 3000, Jackson ImmunoResearch) for 30 min and FITC conjugated goat antichicken IgY (1 : 300, Jackson ImmunoResearch) antibody for 1 h. Sections were mounted, dried, and coverslipped with Vectorshield (Vector). Orexinergic neurons were identified with Cy3 (red fluorescence) labelling whereas MCH neurons were identified with FITC (green fluorescence) labelling.
In a separate group of animals (n = 9), we performed orexin-A and dynorphin B double labelling on day 2, day 4 and day 6 postinjection. The protocol for double labelling of orexin-A and dynorphin B was similar to the protocol described above except that monoclonal mouse antiorexin-A IgG antibody (1 : 300; R & D systems, Minneapolis, MN, USA) and rabbit antidynorphin B IgG antibody (1 : 2400; EMD Biosciences, San Diego, CA, USA) were used as primary antibodies, and biotinylated donkey antirabbit IgG antibody (1 : 300; Jackson ImmunoResearch) and donkey antimouse IgG antibody conjugated to FITC (1 : 300; Jackson ImmunoResearch) were used as secondary antibodies. The specificity of the rabbit antidynorphin B IgG antibody had been verified by preabsorption assay (EMD Biosciences, personal communication) and has been used previously (Shirayama et al., 2004). Dynorphin B-immunoreactive neurons were identified with Cy3 (red fluorescence) labelling whereas orexin-A-containing neurons were identified with FITC (green fluorescence) labelling.
Fluorescence microscopy was carried out with an Axioplan 2 microscope and Slidebook software (Slidebook 4.1; Intelligent Imaging Innovations, Denver, CO, USA). In each animal, every third section was double-labelled for orexin-A and dynorphin B; of these, the two sections with the highest quantity of orexin-A immunoreactivity were selected for bilateral counting. We used the profile of the neuronal cell body and the presence of the fluorescence in the cytoplasm to count neurons and counted all the orexin-A- and dynorphin B-positive neurons in the selected sections (Thakkar et al., 2002).
We next verified the effectiveness and selectivity of siRNA-induced knockdown of the prepro-orexin message by performing real-time PCR for prepro-orexin and dynorphin.
Although immunohistochemical detection of remaining protein expression is currently widely used as an effective assay in gene-knockdown experiments (Turchi & Sarter, 2001; Thakkar et al., 2003; Bhargava et al., 2004) as well as in ataxin-3-induced degeneration (Beuckmann et al., 2004), the immunological technique does not allow the quantification of the remaining protein expression even with the use of stereology. For example, a neuron with partial reduction in prepro-orexin peptide would still be detected and counted as an orexin-A-immunoreactive neuron although the intensity might vary. To overcome these limitations, we used real-time PCR to confirm and quantify the effectiveness of siRNA-induced down-regulation of the prepro-orexin message.
Under general anaesthesia, in a group of nine rats prepro-orexin siRNAs were bilaterally injected (0.03 nmol in 0.3 μL water) into the orexinergic PFH. A separate group of nine rats were bilaterally injected with scrambled siRNA into the PFH. Forty-eight hours after treatment the animals were killed by decapitation using a guillotine. The hypothalamus was quickly dissected out and placed on dry ice and subsequently stored at −80 °C until processed for RNA extraction and real-time PCR.
The hypothalamic tissue samples were first homogenized (Dounce homogenizer) in TRIzol® reagent (0.5 mL per sample) at 4 °C and incubated for 5 min at room temperature (RT). Subsequently, 100 μL of chloroform was added and the mixture was vortexed for 15 s and then left for 2 min at RT. Next, the samples were centrifuged at 12 000 g for 15 min at 4 °C and the supernatant of each sample was transferred to a new tube containing 1 μL of glycogen. Isopropyl alcohol (250 μL) was added to precipitate the RNA. The samples were subsequently centrifuged at 12 000 g for 15 min at 4 °C. The supernatant was discarded and the pellets were washed twice with 75% ethanol, air-dried and resuspended in 50 μL RNase-free water and stored at −80 °C until further analysis.
Reverse transcription for cDNA preparation was carried out with Superscript II (Invitrogen) enzyme. RNA (2 μL) was heated with 1 μL random hexamers, 2 μL 10 mm dNTP mix and 6 μL DEPC-treated water at 65 °C for 5 min and then put on ice. A mixture of 2 μL 10 × reverse transcription buffer, 4 μL 25 mm MgCl2, 2 μL 0.1 m DTT and 1 μL RNase OUT™. was subsequently added and the samples were incubated at 25 °C for 2 min. Next, 1 μL (50 units) of Superscript II was added to each sample and incubated at 42 °C for 50 min followed by 75 °C for 15 min. Next, 1 μL of ribonuclease H was added and each sample was incubated for 15 min at 37 °C.
Each sample was run in duplicate in the Applied Biosystems real-time PCR machine (Model 7700; Applied Biosystems, Foster City, CA, USA). The primer and probe sets from the TaqMan Gene Expression Assays, with efficiencies of 100% for rat prepro-orexin, dynorphin and endogenous control 18S RNA, were used (prepro-orexin: Rn00565995_m1; prodynorphin: Rn00571351_m1; Eukaryotic 18S rRNA Endogenous Control: 4333760T). The PCR reaction mixture contained 2 × TaqMan® Universal PCR Master Mix (25 μL), 20 × TaqMan® Gene Expression Assay (2.5 μL), cDNA (4 or 8 μL) and water with a total volume at 50 μL. The amplification was run by 40 cycles of denaturation at 95 °C followed by annealing and extending at 60 °C. The relative quantification was performed using the comparative threshold cycle method (ΔΔCt method; Livak & Schmittgen, 2001). The ΔCt values were determined by subtracting the reference 18S RNA values from the target gene Ct values. The change between the mRNA expression levels of the experimental (prepro-orexin siRNA-treated) and the control (scrambled siRNA-treated) samples was expressed as 2−(ΔΔCt), where ΔΔCt equals the difference in ΔCt between the control and the experimental sample.
To determine the effects of prepro-orexin siRNA on behavioural states of the animals, we performed bilateral injections of prepro-orexin siRNA into the PFH and monitored sleep and wakefulness. In a separate group, bilateral injections of scrambled siRNA (control) was performed and sleep–wakefulness was monitored.
Under sterile surgical conditions, each rat was implanted with standard EEG and EMG recording electrodes. Bilateral guide cannulae (Plastics One, Roanoke, VA, USA) were implanted 1 mm above the target sites for later microinjections.
Recording cables were connected 1 day postsurgery. After at least a week of recovery and habituation to the recording cage, a 24-h baseline sleep recording beginning at dark onset was obtained. The next day, 4 h before dark onset (15.00 h), the rats were removed from the recording cage and gently swaddled in a towel during bilateral injection (~ 2 min) with either prepro-orexin siRNA or scrambled siRNA (0.03 nmol per side). Twenty-four-hour sleep recordings lasted a minimum of 6 days (Icelus Software, Mark Opp, University of Michigan, Ann Arbor, MI, USA; and Gamma Software, Astromed-Grass, Warwick, RI, USA); animal behaviour was video-recorded using an infrared camera for 8 h per dark period.
At the end of experiments, animals were killed and their orexinergic zones were identified by orexin-A immunohistochemistry (described above). Animals with injection sites > 1 mm AP and/or DV outside the zone of orexin-A-immunoreactive neurons were considered ‘missed hits’. Data were obtained from 12 rats; six were bilaterally treated with prepro-orexin siRNA and six with bilateral scrambled siRNA. Rats (n = 2) with injections outside the target zone were analysed separately.
The sleep data were recorded in 12-h blocks (light and dark periods) and were scored visually off-line in 10-s epochs of Wakefulness, non-REM (NREM) and REM using standard criteria (Thakkar et al., 2003). The amount of time spent in each state was calculated along with number of episodes and average duration of each behavioural state for each 12-h block. Additionally, we calculated average REM sleep latency (defined as time from onset of REM sleep to the preceding wakefulness) and average REM-to-REM interval (defined as time between two REM episodes). Cataplexy or sleep onset REM (SOREM)-like episodes were identified by the concomitant presence of desynchronized, low-amplitude EEG with theta activity and a sudden reduction in muscle tone that was preceded by > 30 s of wakefulness. The average duration and average number of cataplexy or SOREM-like episodes was also calculated.
While sleep data were analysed by one-way repeated-measures (RM) anova followed by post hoc Bonferroni-corrected t-tests, neuronal counts were analysed using paired t-test and PCR data were analysed with Student's t-test.
The prepro-orexin siRNA-injected (unilateral) side demonstrated a marked reduction in orexin-A immunoreactivity (Fig. 1; n = 4) on day 2 as compared to the side that received scrambled siRNA (control). Double immunofluorescence labelling of MCH and orexin-A revealed that there was a marked reduction in orexin-A immunoreactivity on the prepro-orexin siRNA-injected side (Fig. 2) as compared with the side treated with scrambled siRNA; in contrast, there was no reduction in MCH immunoreactivity on the prepro-orexin siRNA-injected side as compared with scrambled siRNA side.
Double immunofluorescence labelling of dynorphin B and orexin-A revealed that all orexin-A-immunoreactive neurons were double-labelled and showed dynorphin B immunoreactivity. Dynorphin B immunoreactivity was also found in many other nonorexin-positive neurons (Fig. 3). Similar findings were also reported by two previous studies (Bayer et al., 2002 and Harthoorn et al., 2005). On the prepro-orexin siRNA-injected (unilateral) side, the number of orexin-A-labelled neurons was significantly reduced 2 days after receiving the prepro-orexin siRNA treatment, as compared with the scrambled siRNA side (23% reduction; n = 5, mean ± SEM 219 ± 29 vs. 285 ± 21; P < 0.05, t = 3.3, paired t-test; Fig. 3, Table 2). In contrast, there was no reduction in dynorphin B neurons on the prepro-orexin siRNA side as compared with the scrambled siRNA side (n = 3, mean ± SEM 1283 ± 143 vs. 1331 ± 153; Table 3). The prepro-orexin siRNA-treated side did not show any significant change in orexin-A-immunoreactive neurons on days 4 (n = 3, mean ± SEM 145 ± 21 vs. 144 ± 41) and 6 (n = 3, mean ± SEM 213 ± 47 vs. 175 ± 42) postinjection as compared with scrambled siRNA (control) side.
Real-time PCR data showed that there was a 58.5% down-regulation of prepro-orexin mRNA (Fig. 4; n = 9, P < 0.05, Student's t-test) in the siRNA-treated (bilateral injections) group: mean, 41.5% of control; range, 29.8–57.9%; controls being the scrambled siRNA-treated (bilateral injections) group (mean, 100%; range, 81.4–122.8%). Although there was still a slight decrease (16.7%) in prodynorphin in the siRNA-treated group, the change was not statistically significant and its range of variation overlapped with that of the control group.
The prepro-orexin siRNA-treated animals (bilateral injections) showed a significant increase (n = 6, F6,30 = 6.1, P < 0.01, one-way RM anova; see Fig. 5, Table 4) in the amount of time spent in REM sleep exclusively during the dark (active) period after prepro-orexin siRNA administration (bilateral) as compared to the baseline (preinjection night). The increase persisted over the first four postinjection nights (all P < 0.05, post hoc test). In contrast, the scrambled siRNA-treated (bilateral injections) rats (controls) had only a transient increase in REM sleep during the first postinjection night (n = 6, F3,15 = 6.9, P < 0.01, one-way RM anova; Fig. 5 for the first night postinjection (P < 0.05, post hoc test) as compared to the baseline (preinjection night), perhaps due to cellular trauma from the microinjection and a consequent decrease in orexinergic neurotransmission.
Further analysis of REM sleep characteristics revealed that the increase in the amount of time spent in REM sleep after prepro-orexin siRNA injection was mainly due to an increase in the number of REM sleep episodes (n = 6, F6,30 = 3.5, P < 0.05, one-way RM anova; Table 5) and rapid REM-to-REM sleep cycling (F6,30 = 4.0, P < 0.01, one-way RM anova; Table 5). There was no significant change in average REM duration or latency to REM sleep.
In scrambled siRNA-treated animals, the transient REM sleep increase was also due to an increased number of episodes and a shortened REM sleep interval (P < 0.05 for the first night postinjection; Table 5). The transient increase in REM sleep was only observed with injections in the PFH; there were no REM sleep changes in rats (n = 2) where the injections were in nonorexinergic zone (see Fig. 6).
In contrast to the REM sleep enhancement during the dark period, during the light (inactive) period neither the prepro-orexin siRNA- or scrambled siRNA-treated rats showed any change in REM sleep compared to the baseline during the first three postinjection days (Fig. 7). In addition, neither group of animals showed any statistically significant changes in wakefulness or NREM sleep during the dark or the light period (Table 4).
Cataplexy or SOREM-like events were observed in three of the six prepro-orexin siRNA-treated rats and occurred almost exclusively during the dark period (only one cataplexy or SOREM-like episode was observed during the light period in one rat on postinjection day 2). The cataplexy or SOREM behaviour was always followed by wakefulness. Cataplexy or SOREM-like episodes were not present in: (i) rats before prepro-orexin siRNA injection; (ii) scrambled siRNA-treated animals, and (iii) in animals with prepro-orexin siRNA injections in nonorexinergic zones. The average percentage of time spent in cataplexy or SOREM-like episodes in the prepro-orexin siRNA-treated rats in the dark period was very short: baseline, 0%; night 1, 0.07%; night 2, 0.34%; night 3, 0.03%; night 4, 0.01%; night 5, 0.06%; and night 6, 0%.
The results of our study demonstrate the following. (i) PFH administration of prepro-orexin siRNA against the prepro-orexin gene caused a persistent and significant increase in REM sleep exclusively during the dark period without significantly affecting any other behavioural state. The effect on REM sleep was mainly due to a significant increase in the number of REM sleep episodes and a reduction in the REM sleep to REM sleep interval. There was no significant change in average REM duration or latency to REM sleep. Furthermore, there was no change in any of the behavioural states during the light period. (ii) In contrast, in scrambled siRNA-treated rats, except for a transient increase in REM sleep during first postinjection night, there was no change in any behavioural state during either the light or the dark period.
The effectiveness and the specificity of prepro-orexin siRNA treatment on the targeted protein and message was verified by the following. (i) Orexin-A immunohistochemistry showed a significant loss of orexin-A immunoreactivity on day 2 postinjection (when the behavioural effects of prepro-orexin siRNA were maximal) in and around the site of prepro-orexin siRNA injection, as compared with that of scrambled siRNA injection (Fig. 1); MCH and dynorphin B immunoreactivity was unaffected (Figs 2 and and3),3), suggesting that there was no global cell loss or protein synthesis down-regulation after siRNA administration. (ii) Real-time PCR showed a marked reduction in the prepro-orexin message in the prepro-orexin siRNA-treated site as compared with that of the scrambled siRNA-treated side (Fig. 4). Furthermore, the expression of prodynorphin was unaffected. (iii) The effect was site-specific and injections outside the orexinergic PFH did not have any effect on REM sleep. (iv) The behavioural effects of bilateral siRNA injections observed on day 4 postinjection were probably due to reduced orexins at the terminals, which may take time to normalise. (v) Our sleep and video recording data are also highly congruent with orexin down-regulation with the time course shown by the immunohistochemical changes, and the recovery of prepro-orexin immunoreactivity after 4–6 days after injection supports the absence of a generalised toxic effect on PFH orexinergic neurons.
Our study is, to our knowledge, the first to employ siRNA to elucidate the role of orexins in the regulation of behavioural state. Accompanied by appropriate controls, siRNA technology appears to have advantages over other techniques as it provides a highly specific and reversible inhibition of expression of a particular gene.
Numerous studies, including ours, indicate that naked siRNA are taken up by neurons in vivo without transfection reagents or viral vectors, and produce target gene knockdown (Makimura et al., 2002; Shishkina et al., 2004; Thakker et al., 2004; Thakker et al., 2005). Furthermore, naked siRNA have the advantage of not eliciting interferon and other nonspecific responses (Heidel et al., 2004; Fedorov et al., 2005).
Both the timing of effects and the dose of siRNA used are consistent with the notion that siRNA can achieve the same gene-suppressive effects in 2–4 days with a dose ≥ 100-fold lower than that required for inhibition by antisense oligonucleotides (Hough et al., 2003; Miyagishi et al., 2003). We observed the effects of siRNA after 1 day, even though the dose (0.03 nmol) in our experiment was > 100-fold lower than the dose used in previous siRNA infusion and antisense studies (Dorn et al., 2003; Thakker et al., 2004, 2005).
Finally, we note that a microinjection approach may further reduce the chance of nonspecific effects by limiting the spread of siRNA to other brain regions. We found a single treatment (bilateral injection) of prepro-orexin siRNA produced a significant knockdown (~ 59%) of the prepro-orexin message; this was similar to other continuous siRNA infusion studies where the knockdown ranged from 30 to 50% (Dorn et al., 2003; Thakker et al., 2004, 2005). Moreover, this down-regulation of orexin expression, in turn, induced 4 days of behavioural state alteration, with a maximal effect at 2 days postinjection. This appears to be in accord with data from most RNA interference studies and a recent extensive literature review, which noted that siRNA effects on mRNA peaked at 36–48 h and started to disappear at ~ 96 h (Ryther et al., 2005).
Our study and others have found that a single microinjection of siRNA into the brain worked well (Makimura et al., 2002; Bhargava et al., 2004). In contrast, chronic infusion either did not work (Isacson et al., 2003) or needed several days and a very high dose to work (Dorn et al., 2003; Thakker et al., 2004, 2005). Local infusion of siRNA may decrease the required dose for effectiveness (present study) while, in the cited infusion studies, intracerebroventricular or intrathecal application of siRNA may have decreased the effectiveness. Additionally, those previous infusion studies have targeted receptors, ion channels or transporters whose turnover rate is probably lower than that of orexin neurotransmitter peptides.
Our major finding is that REM sleep increases after prepro-orexin siRNA administration occurred exclusively during the dark period. This strongly suggests a diurnal regulatory role of orexin. There is anatomical evidence of a suprachiasmatic nucleus-to-PFH projection (Deurveilher & Semba, 2005). In the rat, orexin release is higher during the dark (active) phase than the light (inactive) phase and this consolidation is disrupted by lesions of the circadian pacemaker in the suprachiasmatic nucleus (Deboer et al., 2004). In the squirrel monkey, which has consolidated periods of sleep and waking similar to humans, orexin release is higher during the latter third of the day (active period), indicating that in primates the orexin system may act to oppose accumulating sleep drive during the day (Zeitzer et al., 2003). Furthermore, orexin neurons are most active during wakefulness and are relatively quiet during NREM and REM sleep (Alam et al., 2002; Lee et al., 2005; Mileykovskiy et al., 2005).
Our behavioural results of altered diurnal distribution of REM-like events are strikingly similar to data obtained from rats in which orexin-containing neurons are destroyed postnatally by progressive degeneration (Beuckmann et al., 2004). In this study, the most dramatic change in REM-like events was the difference in diurnal distribution; REM-like events (including REM sleep and cataplexy or SOREM) were increased approximately two-fold over the wild type in the normally REM-poor dark period. However, there are two important interpretative differences between prepro-orexin siRNA-induced orexin knockdown and ataxin-3-induced cell death. First, although postnatal neurodegeneration avoids the problem of developmental compensation the degenerative process can take several weeks, which may give other sleep regulatory systems enough time to compensate. Second, orexin neurons contain other neuromodulators, such as dynorphin, which are involved in sleep regulation (Eriksson et al., 2004) and whose effects cannot be disambiguated from those of orexin in the degeneration methodology.
Although Mochizuki et al. (2004) found only a slight increase in REM percentage in the dark period, cataplexy-like events occurred frequently during the dark period in their constitutive orexin-knockout mice. This study raised an important theoretical consideration (Mochizuki et al., 2004). Do increased REM-like events (cataplexy) caused by the reduction or absence of orexins during the dark cycle represent an alteration in the diurnal distribution of REM components or, alternatively, an abnormal dark period expression of cataplexy, without affecting the diurnal distribution of REM sleep? Based on their findings, Mochizuki et al. (2004) favour the latter interpretation, but their data do not rule out an alteration in the diurnal distribution of all REM components (including cataplexy) in orexin-knockout animals. Additionally, their use of constitutive orexin-knockout animals had the potential confound of developmental compensation. In contrast, in the present study a selective knockdown of orexin neurotransmission showed that the diurnal distribution of full-criteria REM sleep, as well as REM components (cataplexy or SOREM) was altered. It is as if the removal of orexin (in prepro-orexin siRNA-treated animals) greatly increases the probability of the REM oscillator being turned on, allowing it to run during its normal suppression during the active period (see review of REM oscillator mathematical modelling by Massaquoi, McCarley and colleagues in Steriade & McCarley, 2005). Our finding that the increase in the amount of REM sleep was due to the increase in REM sleep bouts and decrease in REM-to-REM sleep interval further support this conjecture.
Thus, taken together, previous and current data support the hypothesis that the diurnal control of the distribution of REM sleep and REM-component phenomena and their consolidation into the light (inactive) phase is under orexinergic control.
Indeed, it has been shown that a failure to consolidate REM sleep into the sleep (inactive) circadian phase is highly characteristic of human narcolepsy (Passouant et al., 1969). These data also argue against the hypothesis that orexin acts as a necessary factor controlling the intracycle events of the wake–NREM–REM sleep cycle, as this cycle occurs in orexin-deficient animals and humans. Rather, it seems that orexin is best viewed as a modulatory, consolidating factor shaping the diurnal (circadian) phase of occurrence of REM sleep.
A relatively low dose of prepro-orexin siRNA induced a visible knockdown of orexin peptide immunoreactivity and robust REM sleep changes during the inactive period, suggesting that orexins may play a role in diurnal gating of REM sleep. Our approach indicates that simple and straightforward siRNA administration can be used as a convenient yet powerful tool with broad behavioural study applications.
Department of Veterans Affairs Medical Research Service (R.B.) and NIMH MH62522, R37MH39683 and MH01798. We thank John Franco for animal care and Dr Ritchie Brown for providing helpful comments on the manuscript.