Novel biochemical manipulation of brain serotonin reveals a role of serotonin in the circadian rhythm of sleep–wake cycles


Eiko Nakamaru-Ogiso, as above.


Serotonin (5-HT) neurons have been implicated in the modulation of many physiological functions, including mood regulation, feeding, and sleep. Impaired or altered 5-HT neurotransmission appears to be involved in depression and anxiety symptoms, as well as in sleep disorders. To investigate brain 5-HT functions in sleep, we induced 5-HT deficiency through acute tryptophan depletion in rats by intraperitoneally injecting a tryptophan-degrading enzyme called tryptophan side chain oxidase I (TSOI). After the administration of TSOI (20 units), plasma tryptophan levels selectively decreased to 1–2% of those of controls within 2 h, remained under 1% for 12–24 h, and then recovered between 72 and 96 h. Following plasma tryptophan levels, brain 5-HT levels decreased to ∼30% of the control level after 6 h, remained at this low level for 20–30 h, and returned to normal after 72 h. In contrast, brain norepinephreine and dopamine levels remained unchanged. After TSOI injection, the circadian rhythms of the sleep–wake cycle and locomotive activity were lost and broken into minute(s) ultradian alternations. The hourly slow-wave sleep (SWS) time significantly increased at night, but decreased during the day, whereas rapid eye movement sleep was significantly reduced during the day. However, daily total (cumulative) SWS time was retained at the normal level. As brain 5-HT levels gradually recovered 48 h after TSOI injection, the circadian rhythms of sleep–wake cycles and locomotive activity returned to normal. Our results suggest that 5-HT with a rapid turnover rate plays an important role in the circadian rhythm of sleep–wake cycles.


Serotonin (5-hydroxytryptamine, 5-HT) is an important neurotransmitter in the modulation of several essential behavioral and physiological functions, such as mood regulation, sleep, wakefulness, cognition, and appetite (Jimerson et al., 1990; Canli & Lesch, 2007; Di Giovanni et al., 2008; Berger et al., 2009). Impaired or altered 5-HT neurotransmission appears to be involved in depression and anxiety symptoms, as well as in sleep disorders (Ressler & Nemeroff, 2000).

Brain 5-HT synthesis is restricted to a very limited number of cells in the brainstem raphe nuclei. The 5-HT neurons repeatedly extend collaterals and densely spread their nerve terminals throughout the forebrain (Steinbusch, 1981). Serotonergic receptors are diverse and widely distributed in the central nervous system. There are 14–15 subtypes, with distinct molecular structures, pharmacological properties, and regional distributions (Barnes & Sharp, 1999; Raymond et al., 2001; Hoyer et al., 2002). Serotonergic neurotransmission is predominantly of a paracrine nature, such that 5-HT escapes readily from the synaptic cleft and can act on receptors distant from the release site (Bunin & Wightman, 1999). These features have hampered attempts to understand the physiological functions of 5-HT and its role as a modulator in the central nervous system.

Brain 5-HT is synthesized in two steps from the amino acid l-tryptophan (Trp), with tryptophan hydroxylase (TPH)2 (Canli & Lesch, 2007) being the rate-limiting enzyme. As TPH2 has a Km of 22–125 μm for Trp (Mockus & Vrana, 1998), which is higher than the tissue Trp levels, the brain 5-HT synthesis rate is physiologically regulated by the substrate Trp levels in situ (Fernstrom, 1983). In addition, brain Trp levels are known to be rapidly equilibrated with plasma Trp levels (Wurtman et al., 1980). Therefore, experimental reduction of brain 5-HT levels can be obtained by depleting Trp in plasma. In fact, Trp depletion by ingestion of Trp-free amino acid mixtures, which is effective in temporarily lowering the levels of Trp in plasma, has been widely used to study the behavioral effects of transiently reduced synthesis of brain 5-HT (Young et al., 1985; Delgado et al., 1990; Reilly et al., 1997; Fadda et al., 2000).

In this study, we pursued a different approach to achieve more extensive, sustained Trp depletion in vivo by administering a novel Trp-degrading enzyme, called tryptophan side chain oxidase I (TSOI), isolated from Pseudomonas (Takai & Hayaishi, 1987). TSOI has a lower Km (1 μm) for Trp, and there is no product inhibition and no regeneration of Trp from products (Takai & Hayaishi, 1987). TSOI does not require exogenous cofactors. These features promoted the feasibility of extensive Trp depletion in vivo. We first studied the biochemical effects of 5-HT depletion by TSOI on brain amino acid and neurotransmitter metabolism. Second, we applied this new 5-HT depletion method to investigate the circadian rhythm of sleep–wake cycles, in which 5-HT plays an important role.

Materials and methods


All procedures for animal care and use were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by the Institutional Animal Care and Use Committees of Tokyo University Graduate School of Medicine. Male Sprague-Dawley rats weighing 240–300 g were housed five per cage (26 × 38 × 21 cm), maintained at 25 °C and 55% humidity under 12 h of light (08:00–20:00)/12 h of dark with ad libitum access to laboratory chow and tap water for at least 1 week. They were then deprived of food but not water overnight before and during experiments. A total of 170 rats were used in this study: 18 rats for analysis of the time course of TSOI enzyme activities in plasma; 97 rats for study of the metabolic effects of TSOI injection on brain Trp and 5-HT levels, as well the levels of other amino acids and catecholamines; six or seven rats for analysis of the effect of TSOI injection on regional amine contents; 27 rats for in vivo microdialysis experiments; and 21 rats for physiological assessments. In principle, we used at least three rats per group for biochemical experiments, and seven rats per group for physiological experiments.

Enzyme purification and application in vivo

TSOI was purified from Pseudomonas (ATCC 29574) essentially as previously described (Takai & Hayaishi, 1987), with modifications to obtain a lipopolysaccharide-free (as assessed by the Limulus amebocyte lysate system), gram-scale preparation of the enzyme. Upon in vivo administration, the preparations with specific activity of 2–4 units (μmol/min)/mg protein and more than 95% purity (as judged by native polyacrylamide gel electrophoresis) were diluted to 3 mL in saline, and injected intraperitoneally under light anesthesia at 12.00 h or 14:00 h. Control subjects received saline.

Tissue preparation

For continuous monitoring of TSOI activity and amino acid levels in the circulation, the blood samples were withdrawn at scheduled intervals from the tail vein of a rat with heparinized capillaries (Drummond Scientific Company, Broomall, PA, USA). The plasma was obtained by centrifugation, either in the presence of 5 mm KCN, an inhibitor of TSOI, to prevent further Trp depletion in vitro, or in the absence of KCN for the TSOI activity assay. Upon decapitation, cervical blood was collected in the presence of 100 μL of 30 mg/mL heparin, after which the plasma was obtained by centrifugation. The brain was quickly removed and homogenized in four volumes of ice-cold 0.4 m perchloric acid containing 0.15% sodium metabisulfite and 0.05% disodium EDTA (PCA solution) with a Polytron homogenizer (Kinematica, Lucerne, Switzerland), and deproteinized by centrifugation at 20 000 g for 20 min. The blood and plasma were similarly deproteinized. To isolate specific brain areas, the brain was placed in the mold (RBM 4000C; Activational Systems, Atlanta, GA, USA), frozen on dry ice, and cut into 1–2-mm slices with a razor blade. Appropriate regions were microdissected according to Paxinos & Watson (1986) and Palkovits & Brownstein (1988), and subjected to deproteinization with PCA solution. All samples were stored at −80 °C. PCA precipitates were solubilized with 1 m NaOH overnight, and used for protein determination according to Lowry et al., with bovine serum albumin as a standard (Lowry et al., 1951).

High-performance liquid chromatography (HPLC) analyses

Aromatic amines and their metabolites were separated by using ion-paired reversed phase HPLC coupled with simultaneous fluorometric, electrochemical detection, according to Wagner et al. (1982), with minor modifications. Samples (15–60 μL) kept at 4 °C were injected through an autosampler [CMA200/240 (Carnegie Medicine, Stockholm, Sweden) or model 234 (Guilson, Villiers-le-bel, France)] onto an A-314 C18 column (5 mm, 300 × 6.0 mm; YMC, Kyoto, Japan). The flow rate was 0.45–55 mL/min through a pump (LC9A/10AD; Shimadzu, Kyoto, Japan) at a pressure of 80–110 kg/cm2 at 25–32.5 °C. The mobile phase was composed of 75 mm sodium phosphate, 2.78 mm sodium octyl sulfate, 0.33 mm triethylamine, and 0.1 mm EDTA. The final pH was adjusted to 3.43 before 25% (v/v) methanol was added. Dual detections were performed: the first with a fluorescence detector with a 12-μL flow cell (RF550/RF10A; Shimadzu), with excitation at 280 nm and emission at 330 nm; and the second by amperometry [ECD-100 or ECD-300 (Eicom, Kyoto, Japan) or LC4B (Bioanalytical Systems, West Lafayette, IN, USA)] on a graphite carbon working electrode set at +0.6 or 0.75 V (vs. Ag/AgCl). We measured the levels of 5-HT, norepinephrine (NE), dopamine (DA), 5-hydroxyindoleacetic acid (5-HIAA), 3,4-dihydroxyphenylacetic acid, homovanilic acid (HVA), Trp and tyrosine in various tissues. Quantifications were performed by comparison with the peak heights or areas of the standards placed every 10 samples. Linearity was checked on the basis of standards between 10 fmol and 100 pmol. Free amino acid contents were determined with an amino acid analyzer (L-8500; Hitachi, Tokyo, Japan) with o-phthalaldehyde post-column derivatization using fluorescence detection, and comparison with the standards. Data were expressed as mean ± standard error of the mean (SEM) values in nmol/g wet tissue for whole brain, μmol/L for plasma samples, or in pmol/mg protein for microdissected brain tissues.

In vivo microdialysis

In vivo microdialysis for a freely moving rat was performed as described elsewhere, with two types of dialysis probe (I-shaped, 0.2 mm in diameter, 1–2 mm in length; U-shaped, 0.2 × 0.45 mm, 1–2 mm) stereotaxically installed according to the coordinates of Paxinos & Watson (1986) with a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA). The coordinates were as follows: corpus striatum, bregma +0.5 mm, lateral 3.6 mm, and vertical 7.2 mm; dorsal raphe nucleus (DRN), bregma −7.8 mm, lateral 2.0 mm, vertical 7.0 mm, and angle 18°; prefrontal cortex, bregma 3.2 mm, lateral 0.8 mm, and vertical 4.2 mm; preoptic area (POA), bregma –0.3 mm, lateral 0.6 mm, and vertical 9.4 mm; and anteroventral thalamic nucleus, bregma −1.8 mm, lateral 1.6 mm, and ventral 5.5 mm. The guide cannula was then fixed to the skull of the rat with dental cement and two anchor screws (1.4 × 5 mm; MTGiken, Tokyo, Japan), and the rats were given prophylactic intramuscular injections of potassium penicillin G (200 000 IU/kg; Meijiseika, Tokyo, Japan). After surgery, rats were housed individually in a wooden chamber (450 × 450 × 600 mm) with a fan (2.8–3.1 m3/min, 48 decibels) for at least 1 week with a 12-h light/dark cycle (08:00 to 20:00, 250–350 lux) at 22–25 °C before the experiments. A two-channel liquid swivel [CMA122p (Carnegie Medicine) or TCS2-23 (Tsumura, Tokyo, Japan)] was used in some experiments. A Ringer solution (147 mm Na+, 4 mm K+, 2.3 mm Ca2+, 155.6 mm Cl, pH 5.5–6.0) was perfused by use of a micro-infusion pump [EP-800 (Eicom) or CMA100 (Carnegie Medicine)] with gas-tight syringes at rates of 1–3 μL/min. Approximately 2 h after the start of perfusion, the dialysates were injected into the HPLC system (described above) at 20-min intervals on-line via an autoinjector (AS-10; Eicom) with a modified short program. Local extracellular 5-HIAA and HVA levels in dialysates were monitored. All experiments were started in the light period of the day, to avoid light disturbance in the dark period. At the end of each experiment, the brains were sectioned coronally into 30–50-mm slices on a cryostat (HM505E; Microm, Waldorf, Germany) and stained with cresyl violet to verify the site of the probe under a microscope.

Physiological assessment

Sleep–wake stages were analyzed in freely moving rats by polygraphic recordings, essentially as previously described (Timo-Iaria et al., 1970). Electroencephalograms (EEGs) from the frontal cortex (bregma 3.0 mm, lateral 1.5 mm), electromyograms (EMGs) from the cervical muscle and electro-oculograms (EOGs) were derived from chronically implanted electrodes via a slip ring [Aeroflex Airflyte (Naples, FL, USA) or MTGiken]. Usually, rats were entrained to the light–dark cycle within 10 days after the surgery. They were then deprived of food but not water overnight before TSOI injection and during experiments. Polygraphs were recorded on the thermal analog recorder, monitored with an oscilloscope, and simultaneously saved onto digital audiotapes for further power spectra analyses with bimutus (Kissei, Matsumoto, Japan). Behaviors were monitored with infrared and CCD cameras, and recorded on videotape. The amount of activity was measured at 1-min intervals by counting the number of times that the EMG output signals exceed a threshold.

Thermistors (internal diameter, 1 mm; Takara, Tokyo, Japan) were implanted into the frontal cortex, thalamus and/or POA to measure brain temperatures. The plugs used to connect the thermistors to the recording apparatus were embedded in acrylic dental resin anchored to the skull with small stainless steel screws. The rats were allowed to recover from surgery and to adapt to the recording apparatus for at least 1 week.

Statistical analyses

All data are shown as means ± SEM or means ± standard deviation. The control and TSOI-treated groups were compared by use of an unpaired two-tailed Student’s t-test. For the data that contain repeated measurements over time, Proc GLM in sas/jmp 9.0 (SAS institute Inc., Carey, NC) was also used to account for variability between subjects and within subjects.


Plasma Trp depletion by TSOI

When TSOI was injected into rats, the TSOI activity became detectable in the blood within 15 min and reached a maximal level between 3 and 6 h. The activity then gradually decreased with a half-time of 15 h, and became undetectable after 48 h (Fig. 1A). Although individual variations were apparent, the magnitude of the increase in TSOI activity in the blood was essentially dose-dependent (Fig. 1A). During the first 2 h after TSOI administration (20 units), plasma Trp levels rapidly decreased to below 2% of the control levels, and remained low for the next 10 h (Fig. 1B). However, recovery to the normal Trp levels took much longer, usually between 72 and 96 h after TSOI administration, although TSOI activity completely disappeared in the bloodstream after 48 h. As rats were under fasting conditions, this slow recovery rate might be a consequence of slow Trp mobilization from tissues. The results suggest that TSOI is highly efficient in depleting Trp in vivo.

Figure 1.

 Effects of TSOI injection in vivo. (A) Time courses of changes in TSOI activity in the tail vein after intraperitoneal injection of TSOI [20 units (•), 10 units (○), and five units (Δ)]. TSOI activities were measured essentially as described in Takai & Hayaishi (1987). Results from different rats receiving 20 units are expressed as means ± SEM (n = 4–10 rats). (B) Time course of cervical plasma Trp levels after TSOI injection [20 units (—•—), 10 units (inline image), and 5 units (inline image)], or saline injection as a control (inline image). Results are expressed as means ± SEM (n = 3–9 rats, mostly n = 3). Controls vs. rats treated with 5, 10 and 20 units of TSOI (at 6, 12, 24, 36 and 48 h after TSOI injection), t4 = 2.776, P < 0.005.

5-HT depletion in the whole brain

After TSOI administration, Trp levels in the brain tissue declined, following changes in blood Trp levels (Fig. 2). They fell to 15% of the control levels 2 h after the injection of 20 units of TSOI, remained at a minimum (10–15%) for the next 3–24 h, and then gradually recovered to normal. Brain 5-HT levels gradually decreased to 30% of the control levels 6 h after TSOI administration, remained at a minimum (30% of the control level) for 20–30 h, and recovered to the control levels 72 h after TSOI administration (Fig. 2). A greater decrease was observed for 5-HIAA, a major 5-HT metabolite (Fig. 2). This may have resulted from rapid metabolism of brain 5-HT and/or a heavy dependence of brain 5-HT synthesis on the availability of precursors from the circulation. In contrast to such massive 5-HT depletion, NE and DA (Fig. 2), and the DA metabolites 3,4-dihydroxyphenylacetic acid and HVA generally remained at control levels (data not shown), indicating the metabolic specificity of this perturbation. We also examined whether the marked decrease in Trp concentration after TSOI administration affected other amino acid levels in the brain tissue. The levels of most amino acids, including those of glutamate, taurine, GABA, and tyrosine, were unchanged, whereas those of branched amino acids, such as valine, showed a significant increase (Fig. S1). These results are consistent with the report that branched amino acids compete with Trp in crossing the blood–brain barrier (Pardridge, 1979).

Figure 2.

 Time courses of changes in the levels of Trp, 5-HT, 5-HIAA, DA and NE in the whole brain, after TSOI injection [20 units (—•—), 10 units (inline image), and 5 units (inline image)] or saline injection as a control (inline image). Results are expressed as means ± SEM (n = 3–9 rats, mostly n = 3). For Trp, 5-HT and 5-HIAA levels, controls vs. rats treated with 20 units of TSOI (at 2, 6, 12, 24, 36 and 48 h after TSOI injection), t4 = 2.776, < 0.005, except for 5-HT and 5-HIAA levels at 48 h, t4 = 2.776, P < 0.05. The decrease in DA levels at 6 h was not statistically significant (t4 = 2.776, P = 0.1265).

Regional differences in 5-HT depletion by TSOI

To investigate regional differences in the effect of TSOI administration, the concentrations of 5-HT, NE and DA in representative brain regions were measured. As shown in Table 1, 7 h after TSOI injections, regional 5-HT contents decreased to different extents: 9.8% ± 1.5% of the control level in the pineal body; 23.4% ± 2.3% in the DRN; 32.2% ± 2.6% in the frontal cortex; and 43.5% ± 4.6% in the olfactory bulb. In contrast, the local levels of NE and DA remained unchanged, except in the pineal gland and the accumbens. This suggests that catecholamine metabolism was mostly unaffected by the TSOI-induced 5-HT depletion. The significant increases in DA and NE levels in the pineal gland, and the increase in NE level in the accumbens, probably resulted from the loss of strong inhibitory control by 5-HT over the NE–DA system in these regions. We also monitored local extracellular 5-HT and 5-HIAA levels by using intracerebral microdialysis to assess 5-HT turnover rates (Fig. S2). TSOI administration induced rapid and extensive decreases in dialyzed 5-HIAA in all regions tested. The highest rate (t1/2 = ∼1 h) was measured in the DRN, and the average rate was t1/2 = ∼3 h in most regions, such as the thalamus (Fig. S2A) and the POA (Fig. S2B). The slowest rate (t1/2 = ∼5 h) was seen in the frontal cortex and hippocampus (data not shown). We then investigated the reversibility of local 5-HT synthesis by Trp loading (intraperitoneal) or intracerebroventricular application of tetrahydrobiopterin (BH4), a physiologically unsaturated cofactor for TPH (Thony et al., 2000). By monitoring 5-HIAA levels in the POA, we clearly observed a transient recovery of 5-HIAA levels after Trp loading or intracerebroventricular injection of BH4 (Fig. S2B). Although the exact mechanism underlying the effect of BH4 is not entirely clear, our results support the notion that the 5-HT depletion caused by TSOI was directly attributable to Trp depletion, and that the 5-HT metabolic pathway was not affected by TSOI.

Table 1.   Regional 5-HT, NE and DA contents at 7 h after TSOI administration
  1. Acb, accumbens; FC, frontal cortex; HF, hippocampal formation; OB, olfactory bulb; PIN, pineal body. Data (pmol/mg protein) except for PIN (pmol/whole gland) are presented as mean ± SEM (control, n = 3–4; TSOI, n = 3). For 5-HT, control rats vs. rats treated with 20 units of TSOI, t4 = 2.776 or t5 = 2.5706, P < 0.005. For NE in PIN, t4 = 2.776, P = 0.0180, and for DA in PIN, P = 0.003; for NE in Acb, t4 = 2.776, P = 0.0066. *P < 0.05 and ***P < 0.005, as compared with control values. TSOI/saline was injected at 12:00 h, and all rats were killed in the light phase to avoid light–dark cycles.

PIN545.8 ± 76.453.4 ± 8.0***21.5 ± 4.739.2 ± 3.2*2.3 ± 0.410.1 ± 0.7***
DRN122.0 ± 9.828.6 ± 2.8***79.8 ± 7.385.7 ± 11.310.3 ± 1.212.7 ± 1.5
MRN80.2 ± 10.617.2 ± 1.2***37.7 ± 3.441.8 ± 1.14.6 ± 0.74.7 ± 0.1
HF42.3 ± 3.316.9 ± 2.1***32.0 ± 1.829.3 ± 5.81.1 ± 0.11.6 ± 0.5
Acb27.9 ± 1.614.6 ± 1.5***9.1 ± 0.812.9 ± 0.5*646.3 ± 17.4648.3 ± 107.7
FC18.9 ± 0.76.1 ± 0.5***18.2 ± 1.216.1 ± 0.71.3 ± 0.11.8 ± 0.3
OB10.8 ± 0.24.7 ± 0.5***15.8 ± 0.515.5 ± 0.85.3 ± 0.44.6 ± 0.6

Effects of 5-HT depletion on the circadian sleep–wake rhythm

After TSOI administration, as 5-HT depletion reached a maximum, rats became quiescent, and gross body movements and voluntary activities were markedly reduced. At this stage, we observed that control rats immediately started eating if food pellets were given, whereas TSO-treated rats did not show any appetite, even though they were also under fasting conditions. The EEG showed that the slow-wave sleep (SWS), which is characterized by large-amplitude and slow waves (Fig. 3A) and is normally observed early in the light phase, became short, owing to interruption by waking, which is characterized by a low-amplitude desynchronized EEG (Fig. 3A). Thus, the circadian sleep–wake rhythm was dramatically disintegrated into fragmental patterns (Fig. 3C). Apart from these changes, however, there was no difference in Fourier analyses of the SWS and rapid eye movement (REM) sleep between TSOI-injected and control rats, and no abnormal wave was observed (data not shown). When we applied tactile or air puff stimuli, SWS waves became transiently desynchronized but adapted during stimulation, suggesting that the sensory processing ability remained intact in TSOI-injected rats. In TSOI-induced 5-HT-depleted rats, however, regular short sleep–wake cycles were prominent (Fig. 3B). We measured the switching events between two sleep–wake phases. Under normal conditions, we counted the switching event as ∼300 times a day (∼200 times during the day and ∼100 times during the night), whereas after TSOI injection the number was doubled to ∼550 times a day (∼300 times during the day and ∼250 times during the night). It is known that sleep–wake consolidation is circadian-dependent, and, in fact, suprachiasmatic nucleus (SCN) ablation also increases sleep fragmentation (Ibuka & Kawamura, 1975). Thus, our results suggest that 5-HT is involved in the coupling mechanism between the SCN circadian oscillator and sleep–wake cycles.

Figure 3.

 Effects of TSOI on sleep–wake cycles. (A) Polygraphic recordings of EOGs, EEGs and EMGs from the cervical muscle of rats, and assignment of sleep–wake stages. W, waking. (B) Predominant patterns of EEGs from multiple cortices at the stage of maximal 5-HT depletion. (C) Representative somnograms of control (left) and TSOI-injected (right) rats. The arrow denotes the timing of TSOI (20 units) injection at around 12:00 h in the light period.

Figure 4 shows circadian patterns of SWS and locomotive activity recorded from the same rat after saline and TSOI injection. The hourly SWS time significantly increased at night, but decreased in the day after TSOI injection (D3 and L3 in Fig. 4B). As a result, the amounts of SWS during the day became almost the same as those during the night (D3 and L3 in Fig. 5A), and the total SWS time in L3 significantly decreased (Fig. 5A). However, the cumulative SWS times on day 3 were 597.8 min (203.8 + 394.0 min) for the control and 557.8 min (266.6 + 291.2 min) for TSOI (Table S1), and there was no significant difference between the control and TSO-injected rats. This suggests that brain 5-HT plays a critical role in the regulation of the diurnal rhythm of sleep–wake cycles. As the brain 5-HT level gradually recovered 48 h after TSOI injection (dotted line in Fig. 4B), the circadian rhythms of sleep–wake cycles and locomotive activity returned to the normal pattern (Figs 4B and 5). Interestingly, we also found that the REM sleep (Fig. 3A) was significantly reduced during the day in TSO-injected rats (L2 and L3 in Fig. 5 and Table S1).

Figure 4.

 Circadian rhythms of SWS and locomotive activity in saline-injected (A) and TSO-injected (B) rats. SWS episodes were summed per 1 h. Data were taken from the same rat. The amount of locomotive activity was integrated at 1-min intervals with the counter IC unit connected to the computer by using signals from EMG output. The arrowhead denotes the timing of TSOI (20 units) administration. The broken line (inline image) in (B) shows the time course of brain 5-HT levels after TSOI injection, taken from Fig. 2.

Figure 5.

 Circadian rhythms of SWS and locomotive activity for the controls (A), and their perturbations upon depletion of active 5-HT pools (B). The comparison between the control and TSOI-treated groups was performed with Proc GLM. Up to 36 h after TSOI injection, the TSOI-treated group showed a significant increase in dark-phase SWS and REM sleep, and in locomotive activities (P < 0.0001), whereas a decrease in light-phase SWS was observed (P < 0.0001). The comparison between the control and TSOI-treated groups for each 12-h period was performed with an unpaired Student’s t-test. SWS: L2, control (n = 7) vs. TSOI (n = 11), t16 = 1.546, P = 0.1417; D3, control (n = 7) vs. TSOI (n = 11), t16 = 2.719, P = 0.0152; L3, control (n = 7) vs. TSOI (n = 11), t16 = 3.542, P = 0.0027. REM: L2, control (n = 7) vs. TSOI (n = 11), t16 = 2.965, P = 0.0091; D3, control (n = 7) vs. TSOI (n = 11), t16 = 0.3139, P = 0.7577; L3, control (n = 7) vs. TSOI (n = 11), t16 = 2.453, P = 0.026. Locomotive activity (arbitrary units, as 100% in D2): control (n = 6) vs. TSOI [n = 13, except for D5 (n = 11) and L5 (n = 9)]. L2, t17 = 1.464, P = 0.1616; D3, t17 = 13.69, P < 0.0001; L3, t17 = 3.328, P = 0.0040; D4, t17 = 4.180, P = 0.0006; L4, t17 = 1.707, P = 0.1059; D5, t15 = 2.848, = 0.0122; L5, t13 = 2.598, = 0.0221. *< 0.05, **P < 0.01, and ***P < 0.005, as compared with control values.


In the present study, rapid and specific depletion of brain 5-HT in rats by the Trp-degrading enzyme TSOI has been demonstrated. Our results from 5-HT depleted rats obtained with this new method strongly suggest that 5-HT is involved in maintaining the circadian rhythm of sleep–wake cycles by increasing the continuity of both sleep and wake episodes, rather than simply by promoting arousal or sleep.

In earlier sleep studies, depletion of brain 5-HT in animal experiments has been mostly carried out with p-chlorophenylalanine (PCPA) (Koe & Weissman, 1966), an efficient inhibitor of TPH. However, maximal depletion requires high doses and 1–2 days (Jouvet, 1969). PCPA is known for its lack of specificity, as it also inhibits the activity of tyrosine hydroxylase, the rate-limiting enzyme for catecholamine synthesis. The inhibition mechanism of PCPA is complicated, and it has been thought that PCPA is incorporated into proteins, thus causing the inhibitory effect of PCPA to last for more than a week (Gal & Whitacre, 1982). Nevertheless, these studies have provided most of the biochemical bases for brain 5-HT function. A new potent inhibitor of TPH, p-ethynylphenylalanine, has been developed, but it still inhibits tyrosine hydroxylase activity (Stokes et al., 2000). Specific serotonergic neurotoxins, such as 5,6-dihydroxytryptamine (Baumgarten et al., 1971), and 5-HT uptake inhibitors, including substituted amphetamine derivatives such as 3,4-methylenedioxymethamphetamine and p-chloroamphetamine (Gurtman et al., 2002; Harkin et al., 2003), have also been used in animal experiments to assess the functional roles of 5-HT in the brain. All known 5-HT-depleting agents have been found to be limited in specificity, reversibility, and effectiveness, which may cause complications, molecular compensation or adaptive changes of 5-HT neurons. These issues have often made it harder to interpret the consequences of 5-HT depletion by these agents. In fact, there are fundamental discrepancies among the behavioral manifestations induced by different 5-HT depletors (Jouvet, 1969; Sugden & Fletcher, 1981; Gurtman et al., 2002).

Therefore, we pursued the development of a different approach to achieve rapid and specific 5-HT depletion in vivo, which allows us to directly verify metabolic/physiological responses in real time. Our study showed that the behavioral and physiological consequences of 5-HT depletion by TSOI did not include insomnia and hyperlocomotion, although these were major physiological responses induced by PCPA (Jouvet, 1969) and 3,4-methylenedioxymethamphetamine (Gurtman et al., 2002). Interestingly, a recent study involving TPH2 gene knockout mice showed that daytime sleep was significantly increased in mice lacking brain 5-HT (Alenina et al., 2009). Although according to this study, sleep promotion was the net effect of a lack of 5-HT in the central nervous system, this might be indirectly caused by the lack of 5-HT during development (Alenina et al., 2009). In that sense, at least, our rapid and selective arrest of active 5-HT pool(s) with TSOI will provide an alternative acute 5-HT depletion model with which to investigate brain 5-HT functions with fewer complications.

The most crucial finding in this study is that 5-HT depletion by TSOI elicited the loss of the sleep–wake circadian rhythm without affecting daily total sleep time (Fig. 5B; Table S1). Current ideas of how 5-HT modulates sleep–wake mechanisms are multi-directional (Monti & Jantos, 2008), and include suppression of vigilance/reactivity, deactivation of wakefulness by antagonizing the catecholamine arousal mechanisms, induction of SWS, and indirect release of hypnogenic factors (Dugovic, 2001). Previously, the effect of chronic Trp depletion by use of a Trp-free diet on the circadian rhythm of wheel-running activity in rats was investigated (Kawai et al., 1994). The authors found that the long-term Trp-free diet disordered the circadian patterns by obscuring the activity onset, but did not affect the primary circadian pacemaker. They suggested that the strength of coupling between the primary and secondary pacemakers is weakened by chronic 5-HT depletion. These results are in line with our finding that 5-HT depletion elicited loss of the sleep–wake circadian rhythm. It is tempting to speculate that the primary role of the active 5-HT pools is to couple the sleep–wake center(s) with the SCN, a primary circadian oscillator in the brain, by modulating the switch for alternations of sleep–wake stages.

The involvement of 5-HT in the circadian clock and/or modulation of neuronal activities of the SCN has been suggested (Barassin et al., 2002). It is known that the serotonergic input into the circadian system has a major influence on the regulation of the circadian system response to light (Pickard & Rea, 1997; Muscat et al., 2005; Knoch et al., 2006). However, in this study, a loss of rhythmicity in the sleep–wake rhythm is sufficient to suggest a similar loss of rhythmicity in a host of other rhythms. Thus, we investigated the effect of TSOI on the circadian rhythm of brain temperatures. As seen in Fig. S3, although small temperature fluctuations with minute intervals disappeared, the circadian rhythms of brain temperatures were maintained after TSOI injection. This suggests that the primary circadian oscillator remained intact after 5-HT depletion by TSOI. This result is in line with findings from SCN brain slices, which maintained their circadian firing patterns in the absence of 5-HT innervation (Barnes et al., 2003; Yamaguchi et al., 2003; Hamada & Shibata, 2010). Therefore, at present, we hypothesize that the sleep fragmentation was caused by the decoupling between the SCN circadian oscillator and sleep–wake cycles as a result of 5-HT depletion. However, the possibility cannot be ruled out that the localized part of the SCN, which functions to regulate the sleep–wake or locomotive circadian rhythms (LeSauter & Silver, 1999; Lee et al., 2009), lost its control over the circadian rhythms due to 5-HT depletion with TSOI. It will be critical to know whether the circadian rhythms of neuronal activities (multi-unit activity) and/or clock gene expression patterns are intact in the SCN and/or other brain regions that are important for sleep–wake control in 5-HT-depleted rats, such as the basal forebrain/POA (Jones, 2005).

Here, a critical question can be raised: why is the circadian sleep–wake rhythmicity abolished in the present study by TSOI [which causes only transient (up to 36 h) 5-HT depletion], but not in a study that used the selective neurotoxin 5,7-dihydroxytryptamine (5,7-DHT) (which causes irreversible, massive damage of serotonergic neurons, followed by exhaustive 5-HT depletion)? In contrast to the effect of TSOI, the general circadian pattern of the wheel-running rhythm remained unchanged by 5,7-DHT, although 5,7-DHT induced the rapid appearance of advanced activity onset, delayed offset, and longer duration of the nocturnal activity phase (Morin & Blanchard, 1991). Although this apparent discrepancy cannot be easily explained, we have considered possible explanations for the contradiction. First, we speculate that the remaining 5-HT after TSOI injection in the SCN and other related regions actively influences neurons that are involved in the regulation of the sleep–wake cycles. As a result, the effects of 5-HT depletion may have been dramatically intensified, as compared with those after total elimination of 5-HT. In fact, the sleep–wake circadian rhythm remained in TPH2 kockout mice, which have almost no detectable 5-HT in their brains (Alenina et al., 2009), whereas the sleep–wake rhythm was completely destroyed in the rats that received the TPH inhibitor PCPA, which depletes 5-HT, but not completely (Jouvet, 1969). Second, TSOI depletes 5-HT on the basis of local turnover rates; thus, the extent and rate of 5-HT depletion were actually very different in the different brain regions, as described in Results (Table 1; Fig. S2). In other words, brain regions where 5-HT plays a major role (or is consumed actively), such as the pineal body, DRN, and median raphe nucleus (MRN), are more severely affected by TSOI injection. Therefore, there is a clear distinction in the process and time course of 5-HT depletion between TSOI and 5,7-DHT. TSOI is more likely to be less affected by various compensation mechanisms (e.g. upregulation of various 5-HT receptors and adaptive changes, such as plasticity occurring in other neurons interacting with 5-HT neurons), owing to the faster process of 5-HT depletion, than 5,7-DHT. In addition, it has also been reported that normal circadian wheel-running rhythmicity of animals that received 5,7-DHT in the MRN was rapidly disrupted during exposure to constant light (Meyer-Bernstein & Morin, 1996). This suggests that the circadian wheel-running rhythm could also be weakened by 5-HT depletion. It will be very interesting to compare the effects of TSOI and 5,7-DHT on other circadian rhythms, such as clock gene expression. We think that the combination of neurophysiological and pharmacological studies will eventually unravel the complicated role of 5-HT in the sleep–wake mechanism and circadian timing.

As 5-HT depletion by TSOI occurs through Trp depletion, it might be argued that the effects of TSOI injection may arise from the inhibition of protein synthesis. Protein synthesis inhibitors reportedly decrease REM sleep, but have no effect on SWS and animal behavior (Pegram et al., 1973). Given that the minimal level of Trp after TSOI injection was approximately 5 μm in the brain tissue (Nakamaru-Ogioso E. & Takai, K., unpublished results), and that the Km of Trp for aminoacyl-tRNA synthase is 1 μm (Xu et al., 2001), it is unlikely that protein synthesis is inhibited by TSOI injection. Another possibility is that the effects of TSOI injection may arise from the product(s) of the TSOI reaction. Although the major in vitro products of Trp catalized by TSOI were reported to be 3-indoleglycolaldehyde and 3-indoleglyoxal (Takai & Hayaishi, 1987), no 3-indoleglycolaldehyde and only trace amounts of 3-indoleglyoxal were detected in the blood and tissues from rats receiving TSOI. Instead, we found that indoleacetic acid (IAA) was the major in vivo product, and showed significantly increased levels in the plasma and tissues, up to ∼2 μm at the maximum stage of Trp depletion with TSOI. IAA is biologically inactive in animals, and has been detected in urine as a normal metabolite (Green et al., 1980). Application of sub-maximal doses of TSOI followed by massive Trp loading or injection of authentic IAA (200 mg/kg) had no marked effects. These results support the notion that the behavioral and EEG changes observed in the present study resulted directly from 5-HT depletion.

The other important finding in this study is that 5-HT depletion elicited hypoactive (depressive) episodes. Considerable evidence has accrued in the last two decades to support the hypothesis that alterations in 5-HT function in the central nervous system occur in patients with major depression. Changes in sleep EEG is suggested to be the most robust marker of major affective illness, and difficulty in sleep maintenance (increased awakenings) is a well-known symptom of depression (Riemann et al., 2001). A main obstacle in depression research is the fact that this condition affects higher cognitive human processes such as motivation and self-esteem, which cannot be easily modeled in animals. However, 5-HT depletion with TSOI, as demonstrated in this study, may at least provide an experimental model for hypoactive (depressive) stages caused by impaired Trp/5-HT metabolism, and perhaps help to establish at least whether primary insomnia and depression share some biological similarities (Riemann et al., 2001). Needless to say, our 5-HT depletion animal model with TSOI may also help in understanding the clinical results of depletion through the ingestion of Trp-free amino acid mixtures (Delgado et al., 1990; Blokland et al., 2002), as brain 5-HT depletion is achieved through Trp depletion in both methods. Combined with other approaches involving molecular knockout of 5-HT receptors, transporters, or monoamine oxidase A/B (Murphy et al., 1999; Gingrich & Hen, 2001), 5-HT depletion by TSOI will be very useful in the study of unanswered questions relating to 5-HT metabolism/functions.


We would like to thank Drs Hiroshi Tanno and Takafumi Futamura for their generous bulk supply of Pseudomonas cells. We are grateful to Drs Yoshiki Kamiyama and Nobukazu Kakui for their assistance with the purification of TSOI. This work was supported by a Grant-in-Aid for Scientific Research from the Minister of Education, Culture, Sports, Science and Technology in Japan (to K. Takai and E. Nakamaru-Ogiso).








dorsal raphe nucleus








5-hydroxyindoleacetic acid


high-performance liquid chromatography




homovanilic acid


indoleacetic acid


median raphe nucleus






preoptic area


rapid eye movement


suprachiasmatic nucleus


standard error of the mean


slow-wave sleep


tryptophan hydroxylase




tryptophan side chain oxidase I