The effects of sleep stages and time of night on NREM sleep ERPs

Abstract

Event-related potential (ERP) is one of the best techniques for studying information processing during sleep because it does not require behavioral responses or consciousness awareness. Several ERP components have been identified during non-rapid eye movement (NREM) sleep, but the associated underlying processes of these waveforms remain unclear. The present study examines the effect of sleep stage and time of night on the NREM ERPs to further understand these processes. An oddball paradigm was conducted in 11 healthy subjects to elicit ERPs throughout the night. Polysomnographic recordings were also applied to identify sleep stages. The results showed that P220, N350, and P900 decreased during the second half of the night, when the NREM sleep drive is partially satiated. This finding is consistent with the notion that the NREM ERPs reflect an inhibitory process associated with sleep drive. P220 and P900 were also found to increase as subjects entering deep sleep. However, the N350 was not affected by the deepening of sleep and peaked earlier during stage 1 sleep. Although these components are all related to the process for sleep preservation, the N350 may be more associated with sleep–wake transition and the P220 and P900 with the process of deepening of sleep.

Introduction

Sleep is an altered state of consciousness in which the processes of external stimuli are attenuated from the waking state. In spite of the absence of behavioral responses to environmental stimuli during sleep, arousal thresholds have been shown to be lower for meaningful stimulus than neutral stimulus, indicating that the sleeping brain does process external information beyond the sensory level (e.g., McDonald et al., 1975, Shanon, 1979). However, the studies on information processing during sleep have been limited by the lack of measurable indices during sleep. One technique, event-related potential (ERP), has long been used to investigate information processing with or without overt behavioral responses, and with or without attention to the stimuli. Through recording and averaging the brainwaves immediately following stimulus presentation, it is possible to obtain a series of deflections in brainwaves that represent the brain activities associated with the processing of the stimulus presented. Since the analysis of ERP does not require behavioral responses and/or conscious awareness, it is an ideal technique to study the processing of sensory stimuli by the brain during sleep.

Several sleep-specific ERP components have been reported in previous studies. With the use of oddball paradigm, the standard P300 component is known to disappear around the time of sleep onset and is replaced by a series of sleep-specific ERP components during non-rapid-eye-movement (NREM) sleep. Components reported to form part of the NREM-specific ERP include P220, N350, P450, N550, and P900 (e.g., Atienza et al., 2001, Harsh et al., 1994, Hull and Harsh, 2001, Ujszaszi and Halasz, 1988, Winter et al., 1995; for a review, see Bastuji and Garcia-Larrea, 1999).

Although these NREM ERP components have been consistently observed across different studies, the underlying processes associated with theses waveforms have not been fully explored. Nonetheless, some preliminary hypotheses have been generated based on the features of the different components. For example, the NREM ERPs have been shown to increase their amplitudes in response to deviant stimuli compared to standard ones with oddball paradigm, indicating that these waveforms do not simply reflect a general reaction to sensory stimuli, but are at least associated with a primitive discriminating process of information (Bastien and Campbell, 1994, Harsh et al., 1994, Hull and Harsh, 2001, Nielsen-Bohlman et al., 1991). Moreover, when the subjects were asked to attend or respond to a target stimulus while they were still awake, after falling asleep the NREM ERPs (especially N350, N550 and P900) were consistently found to elevate to the target stimulus compared to non-target stimulus (Nielsen-Bohlman et al., 1991, Bastien and Campbell, 1994, Harsh et al., 1994). This elevation might reflect further process of the meaning of the stimulus. Based on the findings, some researchers have suggested that these waveforms are primarily associated with an arousal process that orientates the individual to process the psychological or biological relevance of sensory stimuli during sleep (Halasz, 1998, Bastuji and Garcia-Larrea, 1999, Atienza et al., 2001). However, habituation and/or refractory period might offer alternative explanations of this phenomenon. The shorter intervals between two standard stimuli might decrease the level of firing in the neuronal population because some of the neurons might be refractory after reacting to the previous stimulus (Bastien and Campbell, 1992, Harsh et al., 1994). When the probabilities of target and non-target were held constant through experimental manipulation, the amplitudes of P450, N550 and P900 elicited were found to be similar between the two types of stimuli (Colrain et al., 2000, Hull and Harsh, 2001). The stimulus type effect is more associated with the probability of occurrence than the psychological or cognitive relevance of the stimulus. Furthermore, recent studies on the effects of sleep deprivation and sleep disruption on NREM ERPs also did not support the point of view that the waveforms reflect an arousal process. In one study, the amplitudes of the N350 and the P900 during sleep and the N550 during wake-to-sleep transition were found to increase during evening naps following one night of sleep deprivation (Peszka and Harsh, 2002). In another study, the amplitudes of N350 increased during the undisturbed sleep subsequent to disrupted ones (Nicholas et al., 2002). Given that sleep deprivation and sleep disruption enhances sleep drive (and therefore increases the threshold for arousal) during subsequent sleep, it is unlikely that the elevation of the amplitudes of these waveforms reflects an arousal reaction. On the contrary, they are more likely to reflect an inhibitory process that prevents or minimizes cognitive processing and/or cortical activation following the detection of a sensory event. These NREM specific ERP components have been suggested to represent the manifestation of underlying sleep pressure, as a mechanism that facilitates the initiation and maintenance of sleep (Nicholas et al., 2002, Peszka and Harsh, 2002). One study in particular reported that the percentage of slow wave sleep (SWS) increased in association with the N350 amplitude and suggested that this component may be associated with pressure of SWS (Nicholas et al., 2002).

The present study aimed to further explore the functional significance of the NREM ERPs by examining the effects of sleep stage and time of night on the different waveforms. Several issues were addressed in this study. Firstly, NREM ERPs were compared between the first and second halves of the night. It was reported that the auditory arousal thresholds during sleep were decreasing across the night, with increasing of frequency of awakening by auditory stimuli (Busby et al., 1994, Rechtschaffen et al., 1966, Watson and Rechtschaffen, 1969, Williams et al., 1964, Zimmerman, 1970). It indicates an elevation of arousal process or a reduction of inhibitory process to auditory inputs during the second part of the night when sleep drive has been partially satiated. Therefore, if the ERP waveforms reflect an inhibitory process associated with sleep pressure, they should be attenuated during the later part of the night. In contrast, if the waveforms are associated with an arousal process, they should be enhanced during the second half of the night. There were limited number of studies looked into the effects of time of night on NREM ERPs. The results were not very consistent. For example, both P220 (P2) and N350 during stage 2 sleep was reduced in amplitude during the second half of sleep as compared to the first half, as shown in the figure of an earlier study (Campbell et al., 1992, Fig. 7.5). In another study, the amplitude of N550 during stage 2 sleep was also reported to be higher during the first than the second part of the night, but the amplitude of P220 (P240) was not different between the two parts of the night (Plihal et al., 1996). Since the previous researches on the time-of-night effects on NREM ERPs were conducted only during stage 2 sleep and were limited to some ERP components, the present study further examined the time-of-night effects for different waveforms and during all NREM sleep stages.

Secondly, NREM ERP components in different sleep stages were compared. Several of the NREM ERPs were suggested to reflect the process associated with sleep deepening (Nicholas et al., 2002) or to be the phasic manifestations of SWS pressure (Peszka and Harsh, 2002). If this is the case, the waveforms should be enhanced as an individual gets into deeper sleep stages. Most of previous studies on NREM ERPs focused on the wake to sleep transition or stage 2 sleep under the context of K-complex; only a few studies have compared data derived from all NREM stages. An earlier study reported that the N350 (N300) and the P900 had a tendency to increase with the deepening of sleep (Ujszaszi and Halasz, 1988). The other study showed higher P900 (P700) amplitude during SWS than during stage 2 sleep (Wesensten and Badia, 1988). However, a later study showed that the N350 was not affected by sleep stage (Bastien and Campbell, 1992). The figure of another study showed a tendency of increasing amplitudes of P220 (P2) and N350 (N2) from stage 2 to SWS (Bastuji et al., 1995, Fig. 4). Overall, P220 and P900 were consistently found to be elevated in deeper sleep, N350 showed inconsistent findings. Although these studies have reported the stage effects, they did not consider the effect of time of night at the same time. Since the distributions of different sleep stages are not equal across the night, the stage differences could be confounded by the effects of time of night. Therefore, the present study examined the effects of depth of sleep and time of night at the same time to clarify this issue. It was expected that if a certain ERP component reflects processes associated with sleep deepening, it should be increased in deeper sleep regardless of the proportions of the night.