EFFECTS OF LATERAL HYPOTHALAMIC LESION WITH THENEUROTOXIN HYPOCRETIN-2–SAPORIN ON SLEEP IN LONG–EVANSRATS

D. GERASHCHENKO,a C. BLANCO-CENTURION,a

M. A. GRECOa,b AND P. J. SHIROMANIa*aWest Roxbury VA Medical Center and Harvard Medical School, 1400VFW Parkway, West Roxbury, MA 02132, USAbSRI, International, 333 Ravenswood Avenue, Menlo Park, CA 94061,USA

Abstract—Narcolepsy, a disabling neurological disordercharacterized by excessive daytime sleepiness, sleep at-tacks, sleep fragmentation, cataplexy, sleep-onset rapid eyemovement sleep periods and hypnagogic hallucinations wasrecently linked to a loss of neurons containing the neuropep-tide hypocretin. There is considerable variability in the sever-ity of symptoms between narcoleptic patients, which couldbe related to the extent of neuronal loss in the lateral hypo-thalamus. To investigate this possibility, we administeredtwo concentrations (90 ng or 490 ng in a volume of 0.5 �l) ofthe neurotoxin hypocretin-2–saporin, unconjugated saporinor saline directly to the lateral hypothalamus and monitoredsleep, the entrained and free-running rhythm of core bodytemperature and activity. Neurons stained for hypocretin orfor the neuronal specific marker were counted in the perifor-nical area, dorsomedial and ventromedial nucleus of the hy-pothalamus. More neuronal nuclei (NeuN) cells were de-stroyed by the higher concentration of hypocretin-2–saporin(�55%) compared with the lower concentration (�34%) in theperifornical area, although both concentrations lesioned thehypocretin neurons almost equally well (high concentra-tion�91%; low concentration�88%). The high concentrationof hypocretin-2–saporin also lesioned neurons in the dorso-medial nucleus of the hypothalamus and ventromedial nu-cleus of the hypothalamus. Narcoleptic-like sleep behaviorwas produced by both concentrations of the hypocretin-2-saporin. The high concentration produced a larger increasein non-rapid eye movement sleep amounts during the nor-mally active night cycle than low concentration. Neither con-centration of hypocretin-2–saporin disrupted the phase orperiod of the core temperature or activity rhythms. The lowconcentration of unconjugated saporin did not significantlylesion hypocretin or neurons and did not alter sleep. The highconcentration of unconjugated saporin produced some lossof neuronal nuclei-immunoreactive (NeuN-ir) neurons andhypocretin immunoreactive neurons, but only a transient in-crease in non-rapid eye movement sleep. These results ledus to conclude that the extent of hypocretin neuronal losstogether with an accompanying loss of cells in the lateral

hypothalamus may explain the differences in severity ofsymptoms seen in human narcolepsy. © 2003 IBRO. Pub-lished by Elsevier Science Ltd. All rights reserved.

Key words: hypothalamus, hypocretin–saporin, lesion,NREM sleep, REM sleep, NeuN.

The hypocretins, also known as orexins, are recently dis-covered peptides with a discrete localization in the lateralhypothalamus (LH) (De Lecea et al., 1998; Peyron et al.,1998; Sakurai et al., 1998). A single gene encodes hypo-cretin (HCRT), which is cleaved by proteolytic processinginto two smaller peptides, HCRT1 (orexin A) and HCRT2(orexin B) (De Lecea et al., 1998; Sakurai et al., 1998).HCRT-containing neurons project to the entire brain andspinal cord, providing especially heavy innervation to fore-brain and brainstem neuronal populations implicated inwakefulness (Peyron et al., 1998; Trivedi et al., 1998;Greco and Shiromani, 2001). Recently, this peptide wasimplicated in the human sleep disorder narcolepsy basedon the findings that canines with narcolepsy possess amutation in the HCRT2 receptor (Lin et al., 1999). Trans-genic mice with a deletion of the HCRT gene (Chemelli etal., 1999) or mice with a gene-specific ablation of theHCRT neurons (Hara et al., 2001) exhibit symptoms ofnarcolepsy. In human narcolepsy there is a massive loss ofHCRT neurons (Peyron et al., 2000; Thannickal et al.,2000), and consistent with such a neuronal loss, levels ofHCRT1 are undetectable in the cerebrospinal fluid (CSF)of human narcoleptic patients (Nishino et al., 2000b). Apolymorphism in the HCRT gene associated with narco-lepsy has been found (Gencik et al., 2001).

The posterior hypothalamic region where the HCRTneurons are located was identified to be a “wake-center”by von Economo because he observed that patients suf-fering from the viral encephalitic epidemic of 1918 wereexcessively sleepy and postmortem analysis revealeddamage to this region (von Economo, 1930). Since then afew studies have examined changes in sleep after electro-lytic (McGinty, 1969; Nauta, 1946; Ranson, 1939; Shohamand Teitelbaum, 1982; Swett and Hobson, 1968) or exci-totoxic (Denoyer et al., 1991; Sallanon et al., 1988) lesionsof the posterior hypothalamus but the results have beeninconsistent. In these lesion studies, no attempts weremade to specifically identify the phenotype of the neuronsthat were lesioned, and the inconsistent effects on sleepmight have occurred because the lesion methods did notdestroy the appropriate neurons.

*Corresponding author. Tel: �-617-323-7700; fax: �-617-363-5717.E-mail address: pshiromani@hms.harvard.edu (P. Shiromani).Abbreviations: CSF, cerebrospinal fluid; DMH, dorsomedial nucleus ofthe hypothalamus; EEG, electroencephalogram; EMG, electro-myogram; HCRT, hypocretin; ir; immunoreactive; LH, lateral hypo-thalamus; MCH, melanin concentrating hormone; NeuN, neuronalnuclei; NREM, non-REM; REM, rapid eye movement; SAP, saporin;SCN, suprachiasmatic nucleus; SOREMPs, sleep-onset REM sleepperiods; VMH, ventromedial nucleus of the hypothalamus.

Neuroscience 116 (2003) 223–235

0306-4522/03$30.00�0.00 © 2003 IBRO. Published by Elsevier Science Ltd. All rights reserved.PII: S 0 3 0 6 - 4 5 2 2 ( 0 2 ) 0 0 5 7 5 - 4

223

The HCRT-containing neurons possess an autorecep-tor (Horvath et al., 1999), and to lesion such hypocretin–receptor bearing neurons, we have created a neurotoxinby conjugating the ribosomal inactivating protein saporin(SAP) (Gerashchenko et al., 2001) to the HCRT/orexinreceptor-binding ligand HCRT2/orexin-B. The HCRT–sa-porin (HCRT2–SAP) specifically binds to HCRT receptorsand lesions HCRT neurons (Gerashchenko et al., 2001).Microinjection of HCRT2–SAP into the LH produces nar-coleptic behavior that is directly correlated with the loss ofHCRT neurons (Gerashchenko et al., 2001). The presentstudy was done to further characterize the sleep abnor-mality following LH lesions produced by the HCRT2–SAP.Using two concentrations of the HCRT2–SAP versus un-conjugated SAP and saline, we found a persistent hyper-somnia in response to a greater loss of HCRT and adja-cent LH neurons. We also found that animals with loss ofHCRT and adjacent LH neurons continued to wake up atthe same time of day, indicating that the circadian systemwas intact in these animals. This suggests that HCRT andadjacent LH neurons are not responsible for arousal at aspecific time of day, and that the variability in the severityof symptoms seen in human narcolepsy may depend onthe extent of posterior hypothalamic damage.

EXPERIMENTAL PROCEDURES

Animals and surgical preparation

The studies were conducted in accordance with the principles andprocedures described in the National Institutes of Health Guide forthe Care and Use of Laboratory Animals. Twenty-three maleLong–Evans rats (400–550 g) were housed singly in Plexiglascages with wood shavings, and with food and water available adlibitum. The rats were housed in a room where the temperature(21°C) and the lights were controlled (7:00 a.m. to 7:00 p.m. lightson; 100 lux). The rats were implanted under anesthesia (i.m.injection of a cocktail of acepromazine, [0.75 mg/kg], xylazine[20.5 mg/kg] and ketamine [22 mg/kg]) with electrodes to recordthe electroencephalogram (EEG) and electromyogram (EMG).Four miniature stainless steel screw electrodes were positioned inthe skull to sit on the surface of the cortex and were used to recordthe EEG. Two miniature screws were inserted 2 mm on either sideof the midline and 3 mm anterior to bregma (frontal cortex). Theother two screws were located 2 mm on either side of the midlineand 6 mm behind bregma (occipital cortex). The cortical EEG wasrecorded from two contralateral screws (frontal-occipital). Torecord muscle activity (EMG), two flexible multistranded wireswere inserted in the nuchal muscles. The electrodes were placedin a plastic plug and secured onto the skull by using dentalcement. At the time the sleep recording electrodes were im-planted, bilateral microinjections of HCRT2–SAP, unconjugatedSAP or pyrogen-free saline were made to the LH. After the sur-gery, the rats were returned to their home cages and EEG andEMG recordings were collected continuously for at least threeweeks. The rats were then placed in total darkness for an addi-tional three weeks and the rhythm of core body temperatureduring free-run conditions was recorded. After the experiment, therats were perfused (after overdose of Nembutal) and formalin-fixed brains were used for histological analysis. All rats wereweighed on the day of the surgery and two months after thesurgery.

Drug groups and microinjection method

The following five groups were used: (1) saline (n�5); (2) uncon-jugated SAP (90 ng/0.5 �l, n�3); (3) unconjugated SAP (490ng/0.5 �l, n�4); (4) HCRT2–SAP (90 ng/0.5 �l, n�5); (5) HCRT2–SAP (490 ng/0.5 �l, n�6). The HCRT2–SAP conjugate (Ad-vanced Targeting Systems, San Diego, CA, USA), SAP (Sigma,Saint Louis, MO, USA), or pyrogen-free saline was delivered via aglass micropipette with a tip diameter of 20 �m by using a Pico-spritzer. All test substances were injected bilaterally in a volume of0.5 �l. After injection, the pipette was left in place for 5 min andthen withdrawn slowly. All injections were made in the LH at thefollowing coordinates relative to bregma: A��3.3 mm; L�1.6 to1.8 mm; V�8.2 mm below the dura (Paxinos and Watson, 1986).The concentration of HCRT2–SAP used was based on our study(unpublished observations) on the effects of various concentra-tions of the HCRT2–SAP (45, 90, 245, 490 ng/0.5 �l) in lesioningthe HCRT–receptor containing histaminergic cells in the tube-romammillary nucleus. We found that clearly discernible cell losswas first evident with 90 ng/0.5 �l. The 490-ng/0.5-�l concentra-tion represents the full strength of the neurotoxin.

Analysis of sleep–wake states

EEG and EMG signals were recorded on a Grass polygraph andonto a Jaz disk using an analog–digital board (National Instru-ments, Austin, TX, USA). The EEG data were filtered at 70 Hz and0.3 Hz by using a Grass electroencephalograph and continuouslysampled at 128 Hz. The 24-h EEG and EMG recordings obtainedon the 2nd, 6th, 14th and 21st day postinjection were scoredmanually on a computer (Icelus software, Mark Opp, Ann Arbor,MI, USA) in 12-s epochs for awake, rapid eye movement (REM)sleep and non-REM (NREM) sleep by staff blind to the type ofdrug administered to the rats. Wakefulness was identified by thepresence of desynchronized EEG and high EMG activity. NREMsleep consisted of high-amplitude slow waves together with a lowEMG tone relative to waking. REM sleep was identified by thepresence of desynchronized EEG and/or theta activity coupledwith absence of EMG activity. The amount of time spent in wake-fulness, NREM and REM sleep was determined for each hour.After the EEG data were scored, the code was broken to revealthe identity of each rat.

Sleep-onset REM sleep periods (SOREMPs) were identifiedas REM sleep episodes in the day or night that occurred after 2min or more of wakefulness with less than 2 min of an interveningepisode of NREM sleep. Each SOREMP was identified by acomputer program and then substantiated by visual inspection ofthe corresponding EEG and EMG recordings. Details of the crite-ria used to identify SOREMPs in rats have been described previ-ously (Gerashchenko et al., 2001).

Core body temperature recording

At the time the sleep recording electrodes were implanted, atransmitter to record core temperature (model XM-FH, Mini Mitter,Bend, OR, USA) was also implanted in the abdominal cavity. Thistransmitter also recorded gross motor activity. The signal from thetransmitter was collected every 10 min by using telemetry equip-ment (Mini Mitter, Bend, OR, USA) over a six-week period. Thetemperature and activity signals were recorded during entrained(12-:12-h light on and light off) conditions. Three weeks after theinjections, the lights were turned off and temperature recordingswere made during constant darkness conditions (free-run) forthree weeks. The free-run periods (Tau) of temperature and ac-tivity rhythms were determined for the final week in the free-runconditions.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235224

Immunohistochemistry

Two months after the injections of test substances, the animalswere deeply anesthetized with pentobarbital (150 mg/kg i.p.) andperfused transcardially with 0.9% saline (50 ml) followed by 500ml of phosphate-buffered 4% paraformaldehyde (pH 7.0). Thebrains were postfixed overnight, equilibrated in 30% sucrose, andstored at 4°C. One-in-five series of coronal sections were cut at 30�m on a sliding microtome. Each set of coronal brain sections wasincubated overnight at room temperature in the primary antibody.After washing, the sections were incubated with the secondary anti-body for 1 h (Chemicon, Temecula, CA, USA; 1:250 dilution) andthen reacted with avidin–biotin complex for 1 h (Vector Laboratories,Burlingame, CA, USA). The 3,3�-diaminobenzidine (DAB) methodwas used to visualize the reaction product. Omission of the primaryantiserum resulted in no specific staining.

Antibodies

Rabbit anti–HCRT1 (1:70,000, Peninsula Laboratories, Inc., SanCarlos, CA, USA) and mouse anti-NeuN monoclonal (1:1000;Chemicon) antibodies were used. The secondary antibodies, bi-otinylated anti-rabbit immunoglobulin G (IgG) and biotinylatedF(ab')2 fragments of anti-mouse IgG (both from donkey), wereobtained from Chemicon.

In situ hybridization

In situ hybridization was performed as previously described(Greco and Shiromani, 2001). The HCRT-receptor 1 probe wasgenerously provided by Drs. Nahid Waleh and Thomas S. Kilduff(Stanford Research Institute, Palo Alto, CA, USA) The chromo-somal RNA was transcribed from a linearized plasmid using T7(antisense) or T3 (sense) RNA polymerase and S-UTP with ariboprobe kit (Promega, Madison, WI, USA). Acetylated tissuewas incubated overnight at 55°C in hybridization buffer containingprobe (10 c.p.m./ml); washed successively in 2� (standard salinecitrate; SSC)–1-mM (dithiothreitol; DTT) (50°C, 1 h), 0.2� SSC–1-mM DTT (55°C, 1 h), 0.2� SSC–1-mM DTT (60°C, 1 h); dehy-drated; exposed to film; and developed after 48 h.

Cell counts

A person who did not know the lesion status of the rats counted allclearly stained HCRT-immunoreactive (ir) somata in 11 sections(one in five series) that encompassed the full extent of HCRTdistribution (between –10.6 and –30.8 mm from bregma [Paxinosand Watson, 1986]). To identify whether other neurons weredestroyed, we counted in 11 sections NeuN-positive neuronswithin a 0.2-mm rectangular grid positioned above the fornix (seebox in Fig. 2). In a previous report (Gerashchenko et al., 2001), wehad counted neurons containing melanin concentrating hormone(MCH) and adenosine deaminase, and found that HCRT2-SAP(450 ng) lesioned these neurons. In this study, we used NeuN, aprotein specific to neurons, to provide another index of neuronalloss. To identify neuronal loss in adjacent neuronal populations,NeuN-ir nuclei were counted bilaterally within a 0.2-mm rectangu-lar grid in five sections in the dorsomedial nucleus of the hypo-thalamus (DMH), and seven sections in the ventromedial nucleusof the hypothalamus (VMH). All NeuN-ir cells were identified andcounted using a computer program similar to the NIH-Image pro-gram (3D-DOCTOR, Version 3.0.5, Able Software Corporation,1998–2000). Camera lucida drawings (Nikon microscope, EclipseE400 Model, Melville, NY, USA) were made to identify the le-sioned area.

Statistical analysis

Analysis of variance and t-tests with Bonferroni correction (whereappropriate) were used to compare changes in sleep parameters

(SYSTAT, Version 8.0, SPSS Inc., 1998). The same statisticalmethods were also used to compare counts of HCRT-ir andNeuN-ir cells in each anatomic region. Statistical significance wasevaluated at the P�0.05 level.

RESULTS

Neuronal loss

Fig. 1 summarizes the lesioned area in rats treated withHCRT2–SAP (450 ng), HCRT2–SAP (90 ng) and SAP(450 ng). Little or no neuronal loss was apparent with thelow concentration of unconjugated SAP. The higher con-centration of unconjugated SAP produced a lesion but thiswas much smaller compared with the conjugated SAP (seeFig. 1). Loss of HCRT neurons and NeuN-positive neuronsin the DMH, VMH, and perifornical area was assessedquantitatively. The HCRT neurons located laterally wereprimarily lesioned (Fig. 2A, D), whereas a few HCRT-irneurons located medially remained. Compared with thesaline-treated rats (Fig. 2A and Fig. 3B), both concentra-tions of the HCRT2–SAP produced a significant reductionin the numbers of HCRT-positive cells in the LH (88%neuronal loss for the concentration of 90 ng (t�11.506,P�0.001) and 91% for the concentration of 490 ng(t�12.320, P�0.001); between groups F (4, 22)�51.781,P�0.001). Consistent with the neuronal loss, expression ofHCRT receptor mRNA was also reduced (Fig. 2B, E). Thehigher concentration of HCRT2–SAP destroyed moreNeuN-positive cells in the perifornical area (34% neuronalloss for the concentration of 90 ng (t�3.060, P�0.028) and55% for the concentration of 490 ng, t�5.190, P�0.001);between groups F (4, 21)�7.878, P�0.001; Fig. 2C, F andFig. 3A). As shown in Fig. 1, the extent of the lesionproduced by the higher concentration of HCRT2–SAP wasalso greater, with the lesion extending dorsally into themidline thalamus, medially into the DMH and ventrally intothe VMH. In the DMH and VMH, the higher concentrationof HCRT2–SAP produced a significant reduction ofNeuN-ir neurons (DMH, t�3.384, P�0.014, betweengroups F (4, 21)�40.811, P�0.009; VMH, t�2.816,P�0.048, between groups F (4, 21)�2.970, P�0.009; Fig.4A and B). The lower concentration of the HCRT2–SAP(90 ng) or SAP (90 and 490 ng) did not cause a significantloss of NeuN neurons in the DMH or VMH (Fig. 4A and B).

Analysis of sleep data

In the present study, sleep recordings obtained on days 2,6, 14, and 21 after HCRT2–SAP injection were analyzed.In the saline-treated rats, there were no significant differ-ences in sleep between days 6, 14 and 21. Therefore, thedata from these days were combined (Fig. 5).

The diurnal distribution of sleep–wake states across24 h in the saline- and HCRT2–SAP–treated rats is shownin Fig. 5. The percentage of each sleep–wake state during12 h of day and night is presented in Table 1. Saline-treated rats (n�5) were awake more at night and asleepduring the day, as is typical of nocturnal rodents (Fig. 5 andTable 1). On postinjection night 6, the rats treated withHCRT2–SAP (90 or 490 ng) exhibited more sleep time

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 225

Fig. 1. Camera lucida drawings of lesion in rats treated with 90 ng of hypocretin-2–saporin (A), 450 ng of hypocretin-2–saporin (B), and 450 ng ofsaporin (C). Microinjection of these agents was made to the lateral hypothalamus and loss of NeuN labeled neurons was used to demarcate the lesionarea. Three schematic sections from the rat atlas of Paxinos and Watson (Paxinos and Watson, 1986) are shown at the top of the figure to identifythe location of the drawings relative to bregma. B, bregma.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235226

than saline-treated rats (Table 1). These rats had signifi-cant increases in both NREM and REM sleep at night(Table 1). The significant increase in NREM and REMsleep was also present on night 14. On night 21 postinjec-tion, rats treated with 90 ng of HCRT2–SAP continued tohave significantly elevated amounts of REM sleep duringthe night (Table 1), whereas the rats given the higherconcentration of HCRT2–SAP (490 ng) continued to havesignificantly higher amounts of both NREM and REM sleepat night compared with saline-treated rats (Fig. 5, Table 1).The increase in sleep at night in rats lesioned withHCRT2–SAP occurred because of an increase in the du-ration of bouts of both REM sleep and NREM sleep (Table2). During the night, these rats had significantly morefragmented sleep, as indicated by more transitions be-tween wake–NREM sleep (Table 2).

During the day cycle of the 6th day postinjection,NREM sleep levels were significantly higher in rats givenHCRT2–SAP (90 or 490 ng) but then on succeeding daycycles they returned to normal levels. During the day, theHCRT2–SAP– (90 or 490 ng) treated rats had a significantdecline in REM sleep, with the higher concentration pro-ducing a greater reduction in REM sleep (Fig. 5, Table 1).Previously, we found that albino rats also had a similarincrease in REM sleep at night and a decrease during theday cycle (Gerashchenko et al., 2001).

Unconjugated SAP (490 ng) increased NREM sleepduring both lights-on and -off periods on the 6th postinjec-tion day (Table 1), but by day 14 and 21, NREM sleepamounts returned to control levels (Table 1). Low concen-tration of SAP (90 ng) produced only transient changes inthe amounts of REM sleep but no change in NREM sleep.All tested parameters of sleep/wakefulness (Fig. 5, Tables

1, 2, 3) were not different from control values in rats treatedwith 90 ng of SAP on day 21 postinjection.

Sleep-onset REM sleep periods

Both concentrations of HCRT2–SAP induced SOREMPs.As in the previous study (Gerashchenko et al., 2001),these were evident primarily during the night cycle (Fig. 6).The higher concentration of HCRT2–SAP induced moreSOREMPs at night compared with the values of thesaline group (90 ng, day 6: average�S.E.M.�1.8�0.7,t��2.571, df�8, P�0.033; 90 ng, day 21: average�S.E.M.�1.0�0.4, t��2.500, df�8, P�0.037; 490 ng, day6: average�S.E.M.�3.2�1.1, t��2.631, df�9, P�0.027;490 ng, day 21: average�S.E.M.�3.0�1.2; t��2.261,df�9, P�0.050). SOREMPs were rarely seen in rats in-jected with saline or unconjugated SAP (Fig. 6).

Core temperature

Fig. 7 summarizes the entrained diurnal rhythm of coretemperature in saline versus HCRT2–SAP–treated rats21 days after injection. In all three groups, the temper-ature peak occurred soon after lights off and the tem-perature nadir occurred soon after lights on. This indi-cates that the HCRT2–SAP lesions did not disrupt theentrained rhythm of core temperature by either advanc-ing or delaying the phase position of the temperaturerhythm. Saline-treated rats demonstrated a robust diur-nal rhythm in core temperature, with average core tem-perature at night being significantly greater than duringthe day (average day�S.E.M.�37.12�0.28; averagenight�S.E.M.�37.73�0.34; t�3.771, df�4, P�0.02). Thisdiurnal distribution of core temperature is consistent with the

Fig. 2. Effects of hypocretin-2–saporin on hypocretin-containing neurons and expression of hypocretin mRNA receptor in the lateral hypothalamus.A, B and C represent the lateral hypothalamic sections of saline-treated rats. D, E and F represent the sections of the rats treated with 90 ng ofhypocretin-2–saporin. Hypocretin-containing neurons were identified by immunohistochemistry (A, D), and expression of hypocretin mRNA receptorwas identified by in situ hybridization (B, E). NeuN immunostaining and the rectangular boundaries used to count stained cells are shown in C andF. DMH, dorsomedial nucleus of the hypothalamus; VMH, ventromedial nucleus of the hypothalamus; mt, mammillothalamic tract; f, fornix; opt, optictract; 3V, 3rd ventricle. The scale bar in F applies also to photomicrographs A, C and D. The scale bar in E applies to photomicrograph B.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 227

higher amount of wakefulness at night. The mean temper-ature did not differ between day and night in the HCRT2–SAP– (490 ng) treated rats (average day�S.E.M.�36.84�0.18; average night�S.E.M.�36.93�0.15; t�0.48;df�4). This is also consistent with the reduction of a diurnaldifference in NREM and REM sleep in these rats (see Fig.5). During the day, the core temperature in the HCRT2–SAP (490 ng) was not significantly different from that ofsaline rats. However, during the night the core temperatureof these rats was significantly lower (t�2.416; df�8;P�0.04) compared with that of saline rats (Fig. 7). Coretemperature in rats administered the low concentration ofHCRT2–SAP (90 ng) was not different compared with thatof saline-treated rats. The temperature in SAP-treated ratsalso was not significantly different from that of saline-treated rats.

Because endogenous rhythms such as temperatureand activity are very strongly influenced by light, we placedthe rats in constant darkness (free-run condition) and con-tinued to observe the rhythm of core body temperature andactivity for three weeks. In rats, in the absence of anyexternal cues, the period of the temperature and activity

rhythms lengthens gradually and is generally about 24.2 h.In the present study, in the saline-treated rats, the free-runperiod of temperature and activity rhythm (tau) was 24.16(�0.07) and this was not significantly different in theHCRT2–SAP– (24.02�0.03) or SAP-treated rats(24.10�0.03) (Fig. 8). This indicates that the lesions of theLH and HCRT neurons did not affect the phase or period ofthe endogenous rhythm of core temperature or activityrhythm.

Body weight

Two months postinjection the weight of rats given 90 ng ofunconjugated SAP was not significantly different from thatof saline-treated rats (saline�10.45�1.20% increase inweight; 90 ng of SAP�5.96�4.00% increase in weight;t�10.349; df�6). The high concentration of SAP (490 ng)induced a little reduction in body weight (9.4�7.62% re-duction in weight; t�20.916; df�7; P�0.022 vs. salinegroup). Rats given both concentrations of HCRT2–SAPhad a significant reduction in weight compared with saline-treated rats (�13.5�10.31% for the 90-ng concentration(t�2.312; df�8; P�0.050); �23.8�6.28% for the 490-ngconcentration (t�4.872; df�9; P�0.001)).

DISCUSSION

Human narcolepsy has been linked to a loss of HCRTneurons (Peyron et al., 2000; Thannickal et al., 2000) but

Fig. 3. Number of NeuN-immunoreactive (A) neurons in the perifor-nical area and HCRT-immunoreactive (B) neurons in the lateral hypo-thalamus of rats treated with saline, saporin (SAP), or hypocretin-2–saporin (HCRT2–SAP). Both concentrations of HCRT2–SAP pro-duced a significant decline in neurons, with the higher concentration ofHCRT2–SAP producing a greater loss of NeuN-immunoreactive neu-rons. The total number of HCRT- or NeuN-immunoreactive cells wasdetermined in 30-�m-thick sections, as described in the ExperimentalProcedures. Asterisks denote P�0.05 compared with saline.

Fig. 4. Number of NeuN-immunoreactive neurons in the dorsomedialnucleus (A) and the ventromedial nucleus of the hypothalamus (B).Data are presented as mean�S.E.M. in each group of rats. *P�0.05versus saline group.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235228

Fig. 5. Mean (�S.E.M.) percentage of wakefulness, non–rapid eye movement (NREM) and rapid eye movement (REM) sleep during 24 h in ratsadministered hypocretin-2–saporin (HCRT2–SAP) or saline into the lateral hypothalamus. The 24 h are represented in 2-h blocks. The dark barrepresents the 12-h lights-off period. On day 6 postinjection, the rats treated with either concentration of HCRT2–SAP (90 or 490 ng) experiencedsignificantly more NREM and REM sleep at night compared with saline-treated rats. On day 21 postinjection, nighttime amounts of NREM and REMsleep were elevated in the rats treated with 490 ng of HCRT2–SAP, and REM sleep levels were increased in rats treated with 90 ng of HCRT2–SAP.

Table 1. Average percent (�S.E.M.) of wakefulness, NREM, or REM sleep, in rats administered hypocretin2-saporin (HCRT2-SAP), saporin (SAP),or saline into the lateral hypothalamus

Group Day post-injection

Light-off period Light-on period

Wakefulness NREM sleep REM sleep Wakefulness NREM sleep REM sleep

Saline 2 59.3�3.3 34.1�2.7 6.6�0.8 33.7�2.2 50.5�1.7 15.8�0.86 68.1�3.5 26.7�2.8 5.3�0.8 33.8�2.7 51.4�2.1 14.8�0.9

14 70.3�2.9 26.0�2.4 3.7�0.5 32.8�3.0 52.4�2.2 14.8�0.921 70.8�3.1 25.8�2.7 3.5�0.5 31.4�3.2 53.2�2.5 15.4�1.0

Saporin 2 52.2�4.7 39.5�3.9 8.4�1.0 38.7�3.3 48.0�2.5 13.4�1.090 ng 6 54.9�5.1 37.6�4.2 7.5�1.0 41.1�4.3 48.1�3.4 10.8�1.0*

14 57.5�4.9 35.9�4.0 6.6�1.0* 39.0�4.8 49.9�3.9 11.0�1.2*21 65.9�4.2 30.4�3.7 3.7�0.6 34.0�3.8 53.6�3.0 12.4�0.9

Saporin 2 45.6�3.7* 45.0�3.0* 9.4�1.0 28.0�2.6 58.0�2.3* 14.0�1.1490 ng 6 53.0�3.7* 40.0�3.2* 7.0�0.8 28.3�2.4 57.8�2.0* 13.9�0.7

14 62.1�4.1 33.6�3.5 4.4�0.7 34.7�3.2 53.8�2.5 11.5�0.9*21 64.2�3.8 31.2�3.1 4.7�0.8 30.7�3.0 56.2�2.6 13.1�0.7

HCRT2-SAP 2 49.9�3.8 40.4�3.0 9.7�0.9* 36.2�3.2 50.9�2.5 12.9�0.9*90 ng 6 49.5�2.9* 43.1�2.6* 7.4�0.6* 27.3�2.1 60.3�1.9* 12.4�0.5*

14 55.2�3.0* 38.1�2.6* 6.7�0.6* 31.8�3.0 56.8�2.6 11.4�0.7*21 65.2�3.3 29.1�2.8 5.6�0.6* 33.0�2.6 55.7�2.3 11.3�0.6*

HCRT2-SAP 2 58.3�3.4 33.5�2.7 8.2�0.8 36.4�2.4 49.1�1.8 14.5�0.8490 ng 6 40.2�1.7* 51.1�1.4* 8.7�0.5* 28.3�1.6 62.2�1.6* 9.5�0.5*

14 50.5�2.0* 42.7�1.8* 6.9�0.5* 42.9�2.3* 48.7�1.9 8.4�0.6*21 53.2�2.6* 38.9�2.3* 7.9�0.7* 45.0�2.9* 46.4�2.6 8.6�0.6*

The values were calculated during 12 h lights-off or 12 h lights-on period.* P�0.05, significant difference compared with the values of the saline group.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 229

it is not known to what extent loss of other neurons in theLH and surrounding posterior hypothalamus contribute tothe disease and the variation in symptoms. The currentlyavailable models of narcolepsy cannot adequately answerthis question, since in the murine model (Chemelli et al.,1999; Hara et al., 2001), only the HCRT neurons areaffected, whereas in the canine model, the disease is dueto a mutation in the HCRT2 receptor (Lin et al., 1999). Onthe other hand, the HCRT2–SAP (Gerashchenko et al.,2001) could be used, since the number of neurons lostwould depend on the concentration of the neurotoxin. Us-ing two concentrations of HCRT2–SAP, we find that whenHCRT neurons and a greater number of adjacent LH neu-rons are lost, there is a corresponding increase in totalsleep time, in addition to SOREMPs and increased REM

sleep. The extent of HCRT neuronal loss together with anaccompanying loss of cells in the LH may explain thedifferences in severity of symptoms seen in human narco-lepsy.

HCRT2–SAP–induced neuronal loss in the LH

In a previous study, we demonstrated specificity of theHCRT2–SAP (Gerashchenko et al., 2001). Using fluores-cent-activated cell sorting analysis, we found that the con-jugate binds with a high affinity to HCRT receptor–bearingChinese hamster ovary cells, but not to substance-P re-ceptor–bearing cells (Gerashchenko et al., 2001). We alsodemonstrated that HCRT2–SAP binds to the HCRT2 re-ceptor and, to a lesser degree, to the HCRT1 receptor

Table 2. Average number of transitions to NREM, REM sleep, or wakefulness and duration of wakefulness, NREM, or REM sleep in rats administeredhypocretin2-saporin (HCRT2-SAP), saporin (SAP), or saline into the lateral hypothalamus during lights-off period

Group Day post-injection

Average number of transitions Average duration of bouts (min)

Wake-NREM NREM-REM NREM-Wake Wake NREM REM

Saline 2 88.0�7.1 33.2�4.9 55.2�5.4 5.2�0.5 2.8�0.2 1.4�0.16 74.4�8.9 31.6�4.5 44.0�5.0 7.2�1.2 2.6�0.2 1.2�0.1

21 67.3�6.8 22.0�4.4 45.5�5.6 8.3�1.6 2.7�0.3 1.1�0.1Saporin 2 101.0�8.7 35.3�3.9 66.3�6.4 4.0�0.6 2.8�0.2 1.7�0.290 ng 6 90.7�10.8 35.7�5.8 56.0�11.7 4.7�0.8 3.0�0.1 1.5�0.1

21 65.7�8.8 19.3�4.5 47.0�5.2 7.8�1.7 3.4�0.2 1.3�0.2Saporin 2 102.3�15.1 40.5�6.0 62.5�11.3 3.6�0.4 3.4�0.5 1.7�0.1*490 ng 6 77.5�13.1 33.8�5.5 43.8�8.6 5.5�0.9 4.1�0.8 1.5�0.2

21 62.0�8.7 18.8�3.0 44.0�7.4 8.0�1.2 3.8�0.5 1.7�0.2*HCRT2-SAP 2 83.6�6.3 43.6�1.3 40.8�5.9 4.6�0.4 3.6�0.3 1.6�0.0*90 ng 6 96.2�13.4 31.4�4.6 65.2�10.2 4.6�1.3 3.2�0.3 1.7�0.1

21 67.3�8.7 19.0�3.0 48.5�7.3 7.8�1.6 3.2�0.3 2.2�0.3HCRT2-SAP 2 79.0�4.2 33.8�2.0 47.2�4.6 5.5�0.5 3.1�0.3 1.7�0.1490 ng 6 97.0�5.5* 35.3�5.1 61.7�3.3* 3.3�0.4* 3.8�0.3* 1.9�0.2*

21 74.5�5.9 27.3�4.0 47.3�4.1 5.7�0.4 4.1�0.6* 2.1�0.3*

* P�0.05, significant difference compared with the values of the saline group.

Table 3. Average number of transitions to NREM, REM sleep, or wakefulness and duration of wakefulness, NREM, or REM sleep in rats administeredhypocretin2-saporin (HCRT2-SAP), saporin (SAP), or saline into the lateral hypothalamus during lights-on period

Group Day post-injection

Average number of transitions Average duration of bouts (min)

Wake-NREM NREM-REM NREM-Wake

Wake NREM REM

Saline 2 150.4�4.4 77.4�4.6 75.2�5.4 1.8�0.1 2.4�0.1 1.4�0.16 128.4�5.4 62.8�3.6 67.0�4.7 2.1�0.1 2.9�0.1 1.7�0.1

21 106.3�4.0 58.3�3.1 50.3�2.8 2.2�0.1 3.5�0.2 1.8�0.1Saporin 2 134.0�9.0 61.7�4.7* 73.0�4.4 2.3�0.1* 2.6�0.3 1.5�0.190 ng 6 113.3�4.4 48.7�2.7* 65.3�6.9 2.8�0.3 3.1�0.2 1.6�0.2

21 99.0�2.5 53.0�3.2 46.3�5.5 2.7�0.0* 3.9�0.1 1.7�0.0Saporin 2 116.0�9.8* 60.8�11.2 58.0�2.5* 1.9�0.2 3.8�0.7 1.7�0.1490 ng 6 105.5�19.2 50.5�9.6 55.5�11.0 2.6�0.6 4.2�0.8 1.9�0.3

21 90.0�8.2 47.0�6.4 43.8�6.8 2.7�0.2 4.7�0.7 2.0�0.2HCRT2-SAP 2 111.0�7.6* 54.0�7.2* 58.0�3.9* 2.6�0.4 3.4�0.3* 1.7�0.1*90 ng 6 123.4�8.3 50.2�4.4 74.0�8.4 1.8�0.2 3.6�0.2* 1.8�0.2

21 98.0�11.1 39.3�3.5* 59.0�8.0 2.7�0.3 4.3�0.5 2.0�0.1HCRT2-SAP 2 119.3�4.9* 58.3�4.2* 61.2�4.8 2.4�0.1* 3.0�0.1* 1.7�0.1490 ng 6 107.7�3.8* 39.8�1.4* 66.8�3.6 2.1�0.2 4.2�0.2* 1.7�0.2

21 81.6�8.2* 29.2�5.2* 52.8�6.5 4.4�0.8* 4.2�0.4 2.3�0.2

* P�0.05, significant difference compared with the values of the saline group.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235230

(Gerashchenko et al., 2001), which is consistent with theproperties of the ligand alone (Sakurai et al., 1998). Cyto-toxic effects in the brain were demonstrated by applyingthe HCRT2–SAP in the LH, where it destroyed some im-munohistochemically identified cells but not others (Gerash-chenko et al., 2001). HCRT neurons are known to contain theHCRT receptor (Horvath et al., 1999), but to what extentthese receptors are present on non-HCRT neurons in the LHis still unclear. In the present study, the two concentrationsdestroyed an equal number of HCRT neurons but the higherconcentration destroyed more NeuN-labeled neurons. Spar-ing of some cells in this and our previous study (Gerash-chenko et al., 2001) might be related to presence and sub-type of HCRT receptor on the neurons.

In the present study, unconjugated SAP did not lesionas many neurons as the HCRT2-conjugated SAP (Figs. 3and 4), a finding that is consistent with previous reportsthat intracellular entry of the SAP is facilitated by conjugat-ing SAP with a ligand (Wiley, 1992). Unconjugated SAPalso did not produce as severe or long-lasting sleepchanges as HCRT2–SAP. The two concentrations ofHCRT2–SAP produced quite similar loss of HCRT neu-rons, and both concentrations also produced a decrease inNeuN-labeled neurons. However, there were more NeuNneurons lost with the higher concentration of HCRT2–SAP.The higher concentration also produced neuronal loss thatextended into the thalamus and to adjacent nuclei such asthe DMH and VMH. Thalamic lesions cause insomnia (Lu-garesi et al., 1986; Marini et al., 1988), but in our study therats had more sleep, including REM sleep.

On the other hand, damage to the DMH may havecaused the weight loss in rats administered 490 ng ofHCRT2–SAP. DMH lesions have been shown previouslyto reduce body weight (Bernardis and Bellinger, 1998).Rats given 490 ng of unconjugated SAP or 90 ng ofHCRT2–SAP also lost weight (�9.4% and �13.5%, re-spectively). Since these rats did not have a significantreduction in the number of NeuN-positive neurons (Fig. 4)in the DMH, the weight loss may be due to the loss of someLH neurons in these animals. Postoperative aphagia andadipsia associated with dramatic loss of body weight istypically seen in rats with electrolytic lesions of the LH(Bernardis and Bellinger, 1993; Bernardis et al., 1999). It ispossible that loss of MCH neurons also contributes to bodyweight loss following the LH lesion. MCH-deficient miceweigh less because of a significant decrease in food intakeand increased metabolic rate (Shimada et al., 1998) andpreviously (Gerashchenko et al., 2001) we found thatHCRT2–SAP (490 ng) destroyed some MCH neurons.

Effects of HCRT neuronal loss on coretemperature rhythm

The rhythm of core body temperature during entrained andfree-run conditions was monitored to determine whether inthe HCRT2–SAP–lesioned rats the biological clock, thesuprachiasmatic nucleus (SCN), is functioning normally.The rats given the high concentration of HCRT2–SAP (490ng) slept more at night when they should have beenawake, and this could have occurred because these rats

Fig. 6. Number of sleep-onset rapid eye movement sleep periods (SOREMPs) during nighttime in rats 2, 6 and 21 days following injection ofhypocretin-2-saporin, unconjugated saporin, or saline. *P�0.05 versus saline group.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 231

were not receiving an appropriate awakening signal fromthe SCN. Nevertheless, these rats continued to display atemperature peak and nadir at the same time of day assaline-treated rats, indicating that the SCN is sending out asignal at the appropriate time of day. The rats given the lowconcentration of HCRT2–SAP also displayed intact en-trained and free-run temperature and activity rhythms (Fig.8), indicating that the SCN is functioning properly in ratswith a massive loss of HCRT neurons. SCN function hasnot been assessed in the murine model of narcolepsy. Inhuman narcoleptics there is a profound loss of HCRTneurons, but the circadian clock functions normally (Dantzet al., 1994), a finding that is consistent with the presentresults in HCRT2–SAP-treated rats.

Animal models of narcolepsy versushuman narcolepsy

At present, narcoleptic-like behavior associated with thedysfunction of the HCRT system has been described inhumans (Nishino et al., 2000b; Peyron et al., 2000;Thannickal et al., 2000; Overeem et al., 2001; Kales et al.,1982), dogs (Lin et al., 1999), mice (Chemelli et al., 1999;Hara et al., 2001), and rats (Gerashchenko et al., 2001).The existence of several animal models of narcolepsy

allows us to compare symptoms between animal and hu-man narcolepsy.

An important symptom in human narcolepsy is exces-sive daytime sleepiness, which is defined as a continuoussubjective feeling of sleepiness and the presence of irre-sistible sleep attacks (Overeem et al., 2001). As measuredby the multiple sleep latency test, human narcoleptics haveexcessive daytime sleepiness but generally do not have anincreased amount of sleep over a 24-h period. Sleep poly-somnography conducted either at home or in the labora-tory has found only few differences in total sleep measuresbetween patients with narcolepsy and normal humans.Among the differences are higher daytime amounts ofREM sleep and stage 1 sleep (drowsiness) in narcolepticpatients, whereas amounts of NREM sleep (stages 3 and4 of sleep) usually do not differ significantly (Broughton etal., 1988; Nobili et al., 1995). Interestingly, when theamount of sleep in untreated patients with narcolepsy iscompared with that of normal habitual nappers, slow-wavesleep amounts are significantly higher in patients with nar-colepsy (Broughton et al., 1998).

As in human narcolepsy, narcoleptic dogs display frag-mentation of vigilance states characterized by significantlyshorter mean duration of wake, drowsy, and deep sleepepisodes (Nishino et al., 2000a; Kaitin et al., 1986). Day-

Fig. 7. Core temperature rhythm in entrained conditions (12-h lights on/lights off) 21 days after administration of saline or hypocretin-2–saporin(HCRT2–SAP) into the lateral hypothalamus. Values represent average core temperature (°C) at 10-min intervals for each group of rats. Averagetemperature during the lights-off period in rats given the high concentration of HCRT2–SAP was significantly lower compared with that of saline-treatedrats (P�0.04). Nevertheless, these rats showed a temperature peak and nadir at the same phase position as saline-treated rats. When these rats wereplaced in constant dark conditions for three weeks, the period of the temperature rhythm was not different from that of saline-treated rats.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235232

time amounts of drowsy state, light sleep, deep sleep, andREM sleep are not significantly different between narco-leptic and normal dogs (Nishino et al., 2000a; Kaitin et al.,1986). A behavioral phenotype that resembles narcolepsyis also displayed by HCRT knockout mice (Chemelli et al.,1999) and mice with an acquired loss of HCRT neurons(HCRT/ataxin-3 mice) (Hara et al., 2001). These micehave fragmented sleep, little diurnal variation in theamount of REM sleep, and behavioral arrests. HCRTknockout mice have increased NREM sleep and REMsleep time during the normally active lights-off period.Nighttime amounts of NREM sleep in HCRT/ataxin-3 mice,however, are not different from that of wild-type mice (Haraet al., 2001).

Similar to that in the HCRT knockout and HCRT/ataxin-3 mice, lesions of the LH with HCRT2–SAP (90 ng)increase REM sleep during the night, reduce REM sleepduring the day, and produce SOREMPs and sleep frag-mentation. When a higher concentration of HCRT2–SAP(490 ng) is used, other LH neurons in addition to the HCRTneurons are destroyed and these rats have an increase inNREM sleep during the night. On the other hand, nighttimeamounts of NREM sleep were not different from controlvalues in narcoleptic dogs and ataxin-3 mice (Lin et al.,

1999; Hara et al., 2001). Taken together, these data sug-gest that the HCRT neuronal loss produces the REM sleepabnormalities and when adjacent LH neurons are also lost,there is also a hypersomnia during the night in rats.

Interestingly, in human narcolepsy, during the earlystages of the illness, there is a profound hypersomnia,which then resolves gradually to sleep attacks andSOREMPs as the illness progresses (Kales et al., 1982). Inthe present study, both 90 and 490 ng of HCRT2–SAPproduced a significant increase in the total sleep amountscalculated over 24 h on day 6 postinjection (25% and 34%increase, respectively). In the 90-ng HCRT2–SAP rats, thehypersomnia returned to normal levels by day 21, but theREM sleep abnormalities persisted. The transient hyper-somnia in the SAP-treated animals might be due to loss ofsome HCRT and LH neurons. In these rats there was nolong-term hypersomnia or SOREMPs, perhaps becausethe unconjugated SAP did not destroy a sufficient numberof the LH and HCRT neurons. Convincing episodes ofcataplexy alone, without an overlying SOREMP, were notpresent in the HCRT2-SAP–treated rats. In fact, clear,unambiguous incidences of cataplexy have not been ob-served in the HCRT/orexin null mutant mice, either. Incanine narcolepsy, specific incidences of cataplexy are

Fig. 8. Activity rhythms of representative rats administered saline (A) or high concentration of hypocretin-2–saporin (HCRT2-SAP) (B) into the lateralhypothalamus. Each panel depicts activity patterns during a 12-h:12-h light–dark entrainment cycle and during dark–dark (free-run) conditions. Activitywas recorded every 10 min from a transmitter (Mini Mitter, Bend, OR) implanted in the abdominal cavity. The time after administration of saline orHCRT2–SAP is identified to the left of each panel (week 2, week 4, etc). Each line represents a 48-h record of activity pattern, with the second 24-hactivity record being repeated in the first half of the subsequent line. The 12-h:12-h light–dark entrained condition is represented by the dark (lights-off)and white (lights-on) bars at the top of each chart. Midway in each chart, the asterisk depicts the point where the lights were turned off and the animalswere maintained in a dark environment devoid of any external lighting cues. Under such conditions, the rat’s intrinsic clock dictates periods of rest andactivity. Saline-treated rats (representative rat depicted in panel A) were more active and awake in the lights-off period and this pattern was maintainedduring the free-run condition. Rats administered HCRT2–SAP (representative rat depicted in panel B) were most active when the lights were turnedoff (during entrained condition) but then activity levels declined during the rest of the time because the animals had hypersomnia. Most important,during the free-run condition the HCRT2–SAP rats continued to be most active at the same time of day despite of a loss of HCRT and neighboringLH neurons, indicating that the HCRT and surrounding LH neurons are not the primary recipients of a signal to awaken from the circadian oscillator.

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 233

triggered in response to food or play and are short, lastingon the average 23 s (Wu et al., 1999). It is quite possiblethat stimuli that trigger cataplexy need to be identified inrodents. Alternatively, in order for cataplexy to occur, onewould have to lesion the brainstem effector neurons impli-cated in triggering cataplexy.

Nevertheless, both human narcoleptics and the highconcentration of HCRT2–SAP–treated rats are exces-sively sleepy during the normal wake–active period. Wesuggest that the hypersomnia results from the loss ofHCRT- and adjacent non–HCRT-containing LH neurons.We reach this conclusion based on the sleep changes inthe low concentration of HCRT2–SAP treated rats. Theserats, like the HCRT/ataxin knockout mice, have REM sleepabnormalities at night but without a corresponding hyper-somnia. In these rats, the diurnal rhythm of temperature issimilar to that of saline-treated rats. On the other hand, therats given the higher concentration of HCRT2–SAP have agreater loss of adjacent non–HCRTcontaining neurons.These rats also have a hypersomnia at night. Loss ofnon-HCRT wake-related neurons produced by high con-centrations of HCRT2–SAP may have accounted for thehypersomnia seen in the present study.

HCRT neurons are not the only population of wake–active neurons in the perifornical lateral hypothalamicarea. Based on the electrophysiological characteristics ofthe neurons in this area, 53% of recorded neurons wereclassified as wake-/REM-related and 38% as wake-related(Alam et al., 2002). These results suggest that other neu-rons in the LH are also responsible for wakefulness, com-pared with the HCRT neurons. Moreover, recent findingssuggest that in some narcoleptics, CSF levels of HCRT arenormal (Ripley et al., 2001), raising the possibility that inthese patients not all of the HCRT neurons have been lostand/or there is additional loss of other adjacent LH neu-rons. One can rule out neurons containing MCH as beingresponsible for narcolepsy, since they are not lost in hu-man narcoleptics (Thannickal et al., 2000) and MCHknockout mice do not display narcoleptic behavior (Shi-mada et al., 1998). Thus, there appears to be an additionalcell type not measured in human narcoleptics and which isdestroyed in our studies. Destruction of these neurons inaddition to the HCRT neurons might be causing hypersom-nia.

Taken together, the results show that when HCRTneurons are lost, REM sleep is increased, including inad-vertent triggering of REM sleep episodes during purposefulbehavior. Loss of other LH neurons in addition to theHCRT neurons results in a hypersomnia due to an in-crease in both NREM and REM sleep. The extent of LHneuronal loss may explain the symptom variability seen inhuman narcoleptics.

Acknowledgements—We thank Jill Winston for expert technicalassistance and Elizabeth Winston and Samara Shiromani for dataanalysis. Supported by National Institutes of Health grantsNS30140, AG09975, AG15853, and MH55772, and Medical Re-search Service of the Department of Veterans Affairs.

REFERENCES

Alam MN, Gong H, Alam T, Jaganath R, McGinty D, Szymusiak R(2002) Sleep-waking discharge patterns of neurons recorded in therat perifornical lateral hypothalamic area. J Physiol 538:619–631.

Bernardis LL, Bellinger LL (1993) The lateral hypothalamic arearevisited: neuroanatomy, body weight regulation, neuroendocrinol-ogy and metabolism. Neurosci Biobehav Rev 17:141–193.

Bernardis LL, Bellinger LL (1998) The dorsomedial hypothalamic nu-cleus revisited: 1998 update. Proc Soc Exp Biol Med 218:284–306.

Bernardis LL, Medige J, Gillespie R, Wu H, Zivt I (1999) The lateralhypothalamic syndrome in the weanling rat: bone geometry andbiomechanics. Physiol Behav 67:799–802.

Broughton R, Dunham W, Newman J, Lutley K, Duschesne P, RiversM (1988) Ambulatory 24 hour sleep-wake monitoring in narcolepsy-cataplexy compared to matched controls. Electroencephalogr ClinNeurophysiol 70:473–481.

Broughton R, Krupa S, Boucher B, Rivers M, Mullington J (1998)Impaired circadian waking arousal in narcolepsy-cataplexy. SleepRes Online 1:159–165.

Chemelli RM, Willie JT, Sinton CM, Elmquist JK, Scammell T, Lee C,Richardson JA, Williams SC, Xiong Y, Kisanuki Y, Fitch TE, Naka-zato M, Hammer RE, Saper CB, Yanagisawa M (1999) Narcolepsyin orexin knockout mice: molecular genetics of sleep regulation.Cell 98:437–451.

Dantz B, Edgar DM, Dement WC (1994) Circardian rhythms innarcolepsy: studies on a 90 minute day. Electroencephalog ClinNeurophysiol 90:24–35.

De Lecea L, Kilduff TS, Peyron C, Gao X, Foye PE, Danielson PE,Fukuhara C, Battenberg EL, Gautvik VT, Bartlett FS, Frankel WN,Van den Pol AN, Bloom FE, Gautvik KM, Sutcliffe JG (1998) Thehypocretins: hypothalamus-specific peptides with neuroexcitatoryactivity. Proc Natl Acad Sci USA 95:322–327.

Denoyer M, Sallanon M, Buda C, Kitahama K, Jouvet M (1991) Neu-rotoxic lesion of the mesencephalic reticular formation and/or theposterior hypothalamus does not alter waking in the cat. Brain Res539:303.

Gencik M, Dahmen N, Wieczorek S, Kasten M, Bierbrauer J, Anghe-lescu I, Szegedi A, Menezes Saecker AM, Epplen JT (2001) Aprepro-orexin gene polymorphism is associated with narcolepsy.Neurology 56:115–117.

Gerashchenko D, Kohls MD, Greco M, Waleh NS, Salin-Pascual R,Kilduff TS, Lappi DA, Shiromani PJ (2001) Hypocretin-2-saporinlesions of the lateral hypothalamus produce narcoleptic-like sleepbehavior in the rat. J Neurosci 21:7273–7283.

Greco MA, Shiromani PJ (2001) Hypocretin receptor protein andmRNA expression in the dorsolateral pons of rats. Brain Res MolBrain Res 88:176–182.

Hara J, Beuckmann CT, Nambu T, Willie JT, Chemelli RM, Sinton CM,Sugiyama F, Yagami K, Goto K, Yanagisawa M, Sakurai T (2001)Genetic ablation of orexin neurons in mice results in narcolepsy,hypophagia, and obesity. Neuron 30:345–354.

Horvath TL, Diano S, Van den Pol AN (1999) Synaptic interactionbetween hypocretin (orexin) and neuropeptide Y cells in the rodentand primate hypothalamus: a novel circuit implicated in metabolicand endocrine regulations. J Neurosci 19:1072–1087.

Kaitin KI, Kilduff TS, Dement WC (1986) Evidence for excessivesleepiness in canine narcoleptics. Electroencephalogr Clin Neuro-physiol 64:447–454.

Kales A, Cadieux RJ, Soldatos CR, Bixler EO, Schweitzer PK, PreyWT, Vela-Bueno A (1982) Narcolepsy-cataplexy. I. Clinical andelectrophysiologic characteristics. Arch Neurol 39:164–168.

Lin L, Faraco J, Li R, Kadotani H, Rogers W, Lin X, Qiu X, de Jong PJ,Nishino S, Mignot E (1999) The sleep disorder canine narcolepsy iscaused by a mutation in the hypocretin (orexin) receptor 2 gene.Cell 98:365–376.

Lugaresi E, Medori R, Montagna P, Baruzzi A, Cortelli P, Lugaresi A,Tinuper P, Zucconi M, Gambetti P (1986) Fatal familial insomnia

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235234

and dysautonomia with selective degeneration of thalamic nuclei.New Engl J Med 315:997–1003.

Marini G, Gritti I, Mancia M (1988) Ibotonic acid lesion of the thalamicmedialis dorsalis nucleus in the cat: effects on the sleep-wakingcycle. Neurosci Lett 89:259–264.

McGinty DJ (1969) Somnolence, recovery and hyposomnia followingventromedial diencephalic lesions in the rat. ElectroencephalogClin Neurophysiol 26:70–79.

Nauta JH (1946) Hypothalamic regulation of sleep in rats: an experi-mental study. J Neurophysiol 9:285–316.

Nishino S, Riehl J, Hong J, Kwan M, Reid M, Mignot E (2000a) Isnarcolepsy a REM sleep disorder? Analysis of sleep abnormalitiesin narcoleptic Dobermans. Neurosci Res 38:437–446.

Nishino S, Ripley B, Overeem S, Lammers GJ, Mignot E (2000b)Hypocretin (orexin) deficiency in human narcolepsy. Lancet 355:39–40.

Nobili L, Besset A, Ferrillo F, Rosadini G, Schiavi G, Billiard M (1995)Dynamics of slow wave activity in narcoleptic patients under bedrest conditions. Electroencephalogr Clin Neurophysiol 95:414–425.

Overeem S, Mignot E, Gert VD, Lammers GJ (2001) Narcolepsy:clinical features, new pathophysiologic insights, and future per-spectives. J Clin Neurophysiol 18:78–105.

Paxinos G, Watson C (1986) The rat brain in stereotaxic coordinates.New York: Academic Press.

Peyron C, Faraco J, Rogers W, Ripley B, Overeem S, Charnay Y,Nevsimalova S, Aldrich M, Reynolds D, Albin R, Li R, Hungs M,Pedrazzoli M, Padigaru M, Kucherlapati M, Fan J, Maki R, Lam-mers GJ, Bouras C, Kucherlapati R, Nishino S, Mignot E (2000) Amutation in a case of early onset narcolepsy and a generalizedabsence of hypocretin peptides in human narcoleptic brains. NatMed 6:991–997.

Peyron C, Tighe DK, Van den Pol AN, De Lecea L, Heller HC, SutcliffeJG, Kilduff TS (1998) Neurons containing hypocretin (orexin)project to multiple neuronal systems. J Neurosci 18:9996–10015.

Ranson SW (1939) Somnolence caused by hypothalamic lesions inthe monkey. Arch Neurol Psychiat 41:1–23.

Ripley B, Overeem S, Fujiki N, Nevsimalova S, Uchino M, Yesavage J,

Di Monte D, Dohi K, Melberg A, Lammers GJ, Nishida Y, Roe-landse FW, Hungs M, Mignot E, Nishino S (2001) CSF hypocretin/orexin levels in narcolepsy and other neurological conditions. Neu-rology 57:2253–2258.

Sakurai T, Amemiya A, Ishii M, Matsuzaki I, Chemelli RM, Tanaka H,Williams SC, Richardson JA, Kozlowski GP, Wilson S, Arch JR,Buckingham RE, Haynes AC, Carr SA, Annan RS, McNulty DE, LiuWS, Terrett JA, Elshourbagy NA, Bergsma DJ, Yanagisawa M(1998) Orexins and orexin receptors: a family of hypothalamicneuropeptides and G protein-coupled receptors that regulate feed-ing behavior. Cell 92:573–585.

Sallanon M, Sakai K, Buda C, Puymartin M, Jouvet M (1988) Increaseof paradoxical sleep induced by microinjections of ibotenic acid intothe ventrolateral part of the posterior hypothalamus in the cat. ArchItal Biol 126:87–97.

Shimada M, Tritos NA, Lowell BB, Flier JS, Maratos-Flier E (1998)Mice lacking melanin-concentrating hormone are hypophagic andlean. Nature 396:670–674.

Shoham S, Teitelbaum P (1982) Subcortical waking and sleep duringlateral hypothalamic ‘somnolence’ in rats. Physiol Behav 28:323–333.

Swett CP, Hobson JA (1968) The effects of posterior hypothalamiclesions on behavioral and electrographic manifestations of sleepand waking in cats. Arch Ital Biol 106:283–293.

Thannickal TC, Moore RY, Nienhuis R, Ramanathan L, Gulyani S,Aldrich M, Cornford M, Siegel JM (2000) Reduced number ofhypocretin neurons in human narcolepsy. Neuron 27:469–474.

Trivedi P, Yu H, MacNeil DJ, Van der Ploeg LH, Guan XM (1998)Distribution of orexin receptor mRNA in the rat brain. FEBS Lett438:71–75.

von Economo C (1930) Sleep as a problem of localization. J Nerv MentDis 71:249–259.

Wiley RG (1992) Neural lesioning with ribosome-inactivating proteins:suicide transport and immunolesioning. Trends Neurosci 15:285–290.

Wu MF, Gulyani SA, Yau E, Mignot E, Phan B, Siegel JM (1999) Locuscoeruleus neurons: cessation of activity during cataplexy. Neuro-science 91:1389–1399.

(Accepted 6 August 2002)

D. Gerashchenko et al. / Neuroscience 116 (2003) 223–235 235