1 INSERM U480, Université Claude Bernard, Lyon, France ,
Correspondence: Jean-Louis Valatx, INSERM U480, University Claude Bernard, 8 Avenue Rockefeller, 96373 LYON Cedex 08, France.
Recently, insulin synthesis and the presence of an insulin receptor have been demonstrated in the brain. Intracerebroventricular infusion of insulin causes a selective increase in the amount of slow-wave sleep. In the present study, the sleep-wake cycle of transgenic mice, with or without habenular neuronal expression of the human insulin gene, was studied to investigate the possible role of brain insulin as a sleep modulator. Slow-wave sleep duration was increased in those mice expressing human insulin in the habenula. However, it is possible that this effect was not due to expression of the insulin transgene, but to the genetic background of one of the parental strains (CBA) used for insertion of the transgene. Users of transgenic mice should be aware of this possibility and be cautious in interpreting results when hybrid embryos are used as transgene recipients.
human insulin, paradoxical sleep, slow-wave sleep, transgenic mice
Insulin or insulin-related peptides are found in both vertebrates and invertebrates (Smit et al. 1998). Their classical role in energy (glucose) metabolism has now been extended to include many behavioural functions, such as food intake, motor activity, and memory (Mayer et al. 1990; Douhet et al. 1997; Wickelgreen 1998). The existence of insulin receptors in the brain (Pezzino et al. 1996; Doré et al. 1997) and the transport of insulin across the blood-brain barrier may explain these behavioural effects. However, using RT-PCR, amplification and in situ hybridization, both preproinsulin 11 and insulin messenger RNA have been shown to be present in the hypothalamus (Young 1986; Devaskar et al. 1993, 1994; Tsuji et al. 1996), and central insulin may therefore be involved in the regulation of several aspects of behaviour. In terms of sleep, Danguir and Nicolaēdis (1984) showed, in the rat, that intracerebroventricular insulin infusion results in a specific increase in the amount of slow-wave sleep (SWS). In transgenic mice bearing a human insulin transgene (HIg) with a modified 5'-flanking sequence (transcription unit), the length of the sequence (258, 168, or 58 nucleotides upstream of the transcription starting point; DELTA-258, DELTA-168, or DELTA-58, respectively) determines whether the transgene is expressed in the pancreatic P-islets, in the brain, or not at all (Fromont-Racine et al. 1990). In DELTA-168 mice, human insulin is preferentially expressed in the brain (Douhet et al. 1993). Cerebral insulin expression has been observed in only one structure, the median habenular nucleus. Using double immunohistochemical staining, human insulin was shown to be colocalized in cholinergic habenular neurons (Douhet et al. 1995). Given that the habenular nucleus is connected to the sleep regulating network, the aim of the present work was to study the effects of extra brain insulin on the sleep-wake cycle.
MATERIALS AND METHODS
In a first series of experiments, nine adult male transgenic mice expressing HIg in the median habenular nucleus (DELTA- 168 mice) and six age-and sex-matched control C57BL/6 J (B6) mice were used. In a second set of experiments, two groups of transgenic mice were used, containing the HIg with altered 5'-flanking sequences resulting in either expression of the transgene only in pancreatic P-islets (DELTA-258) (n = 4) or in its nonexpression (A58) (n = 5) (Itier et al. 1996).
Surgery and recordings
Under pentobarbital anaesthesia (80 mg/kg, i.p.), mice were implanted with five cortical and three muscular electrodes, then caged singly in a Plexiglas jar and housed in a light-and temperature-controlled room (light on, 07.00 h; light off, 19.00 h; 23 + PC) with free access to water and food. All animals were treated according to guidelines approved by the European Communities Council Directive of 24 November 1986 (86/609/EEC). After 10 days of habituation to experimental conditions, continuous recordings were performed for at least 5 days, then the mice were subjected to a 10-h sleep deprivation period (09:00-19:00), using the water tank technique (Kitahama and Valatx 1980), and their recovery monitored over a further 2 days.
All recordings were scored visually, in 30-s epochs, for wakefulness (W), SWS, and paradoxical sleep (PS) using standard criteria (Valatx 1971). The data were then stored in a computer for further analysis using software designed in our own laboratory. The results are expressed as the mean + SD. Statistical analyses for the day, night, or full 24-h periods were performed by multiple factor analysis of variance, followed by multiple range test analysis. For sleep rebound analysis (first 24 h), Student's t-test was used, the significance level being set at P < 0.05.
Figure 1(A) shows that the amount of SWS over the 24-h period was significantly higher in transgenic (DELTA-168) mice (+ 15.60% + 0.7%, P < 0.001); this increase was mainly due to the increase in the amount of SWS during the night period. The amount of PS was not affected. After 10 h of sleep deprivation, the rebound was seen only during the first night and was lower in transgenic mice than in controls (SWS, + 71.1 + 16.6 min vs. + 136.7 + 5.6 min; PS, + 33.0 + 2.0 min vs. + 44.4 + 3.1 min), representing a sleep debt payback of 13.6% vs. 31.4% for SWS and 61.6% vs. 81.7% for PS (Fig. 2). During the second 24-h period, SWS and PS amounts were not different from controls.
Sleep duration in transgenic mice. (A) SWS and PS during the night (SWS-N, PS-N) or day (SWS-D, PS-D) periods or over 24 h (SWS-24 h, PS-24 h) in DELTA- 168 (stippled bars, n = 9) and control B6 (white bars, n ~ 6) mice. (B) SWS and PS duration in mice expressing the human insulin transgene in the brain (DELTA-168) or in pancreatic islets (DELTA-258, n = 4), in mice not expressing the transgene (DELTA-58, n = 5), and in CBA mice (black bars, n = 10). Mean values + SD of the five baseline days.
In DELTA-258 mice, the amount of SWS and PS during baselines (Fig. 1B) and recovery (Fig. 2) was not statistically different from that seen in DELTA-58 mice. In addition, neither strain showed a significant difference in these parameters compared with A168 mice, whereas both showed a significant difference when compared with control B6 mice. In both the DELTA-258 and the DELTA-58 strains, the rebound after sleep deprivation was not different statistically from that seen in DELTA-168 mice, but there was considerable variation between the individual mice in each strain (Fig. 2).
Recovery after a 10-h sleep deprivation period in control B6 mice (white bars) and in three transgenic mice with differing expression of the human insulin gene (see legend to Fig 1 B). The same mice as in Fig 1 were submitted to sleep deprivation. Mean values + SD.
The first experiment shows that expression of the human insulin transgene in habenular neurons caused a specific increase in the amount of SWS and a decreased rebound for both states of sleep. Before discussing the possible mechanisms of these variations, it is interesting to take into consideration the results of the second experiment.
This experiment was designed to study appropriate control mice, i.e. transgenic mice with a modified 5'-flanking sequence leading either to expression of the transgene only in the pancreas (DELTA-258 mice) or to its total nonexpression (DELTA-58 mice). In these mice, the amount of SWS did not differ from that seen in the transgenic mice (DELTA-168) expressing human insulin in habenular neurons. We then asked how these results could be explained and whether the transgenic process (insertion of the transgene) alters the expression of other genes involved in sleep regulation; such a change has been previously described when the human interferon P gene was fortuitously inserted into the MAO-A gene, resulting in its inactivation by deletion of two exons and a resultant altered phenotype (Cases et al. 1995, 1996). Insertion of the human insulin gene does not cause a dramatic change in phenotype. However, it might possibly cause alterations in some candidate genes for sleep regulation recently localized on the mouse chromosome map (Tafti et al. 1997). Unfortunately, the precise localization and number of the insulin transgene copies in the mouse genome are still unknown.
Another explanation may be the strain into which the transgene was injected. In the original publication concerning these mice, the authors mentioned that the transgene was injected into the male pronucleus before the first division of the egg from an F2 hybrid C57BL/6 x CBA (Fromont-Racine et al. 1990). Thus, mice bearing the transgene have a genetic background that is a recombination of the B6 and CBA genomes. In terms of sleep, these two strains differ markedly. Several years ago, we demonstrated that all CBA sleep parameters were dominantly transmitted when these mice were crossed with C57 mice (Valatx et al. 1972), and it would therefore be possible that the sleep characteristics of the three transgenic lines were inherited from the CBA parent of the injected hybrid embryo. Two arguments support this hypothesis. Firstly, in CBA mice, during SWS, EEG recordings show high amplitude spindles (8-10Hz, 800 uV) that are characteristic of several strains derived from BALB/c mice (e.g. C3H and CBA) and dominantly transmitted to F1 hybrids (Valatx et al. 1974), whereas different sleep spindles are seen in B6 mice (12-15Hz, 400 uV); CBA-type spindles were recorded during SWS in several of our transgenic mice (Fig. 3). Secondly, comparison of the present results with previously obtained sleep data from CBA mice (Valatx et al. 1974) shows that the duration of SWS and PS sleep in CBA mice is very similar to that seen in the transgenic mice (Fig. 1B). However, the sleep rebound in transgenic mice was intermediate between that seen in B6 and CBA mice, suggesting that the regulatory mechanisms for rebound are inherited differently from those for spontaneous sleep.
Because of the gene insertion process, it is difficult to interpret the present results in terms of a specific effect of human brain insulin on sleep regulation. Potential users of transgenic mice should be aware of such difficulties and should carefully examine the entire process of transgenic mouse construction. Many mice containing a knockout gene deletion are constructed by alteration of embryonic stem (ES) cells from the 129 strain, which are then injected into a hybrid embryo or another strain. To obtain a stable genetic background, at least 10 or more backcrosses to the recipient strain are necessary. Unfortunately, strain 129 is not often used in behavioural studies and therefore the heritability of most of its behavioural traits is unknown. In conclusion, transgenic mice are potentially good models for sleep studies, as long as certain rules are followed during their construction and breeding, e.g. using the same mouse strain for the entire process of transgenesis. Recent progress has made it possible to perform conditional gene deletion or activation, which becomes effective only in the adult animal, and each animal can therefore be used as its own control.
Examples of 30-s EEG recordings during SWS in a control mouse (A) and in a transgenic mouse (B). Note the high amplitude spindles in the transgenic mouse (vertical bars: 100 uV). Record derived from the fronto-parietal derivation: 11 min frontal to bregma and lambda, and 1 mm lateral to midline.
This work was supported by INSERM U52 and U257. We would like to thank Tom Barkas for proof-reading the manuscript.