|Home | About | Journals | Submit | Contact Us | Français|
Prolonged wakefulness or a lack of sleep lead to cognitive deficits, but little is known about the underlying cellular mechanisms. We recently found that sleep deprivation affects spontaneous neuronal activity in the neocortex of sleeping and awake rats. While it is well known that synaptic responses are modulated by ongoing cortical activity, it remains unclear whether prolonged waking affects responsiveness of cortical neurons to incoming stimuli. By applying local electrical microstimulation to the frontal area of the neocortex, we found that after a 4-hour period of waking the initial neuronal response in the contralateral frontal cortex was stronger and more synchronous, and was followed by a more profound inhibition of neuronal spiking as compared to the control condition. These changes in evoked activity suggest increased neuronal excitability and indicate that after staying awake cortical neurons become transiently bistable. We propose that some of the detrimental effects of sleep deprivation may be a result of altered neuronal responsiveness to incoming intrinsic and extrinsic inputs.
An essential function of the brain is to perceive complex external and internal stimuli and translate them into adequate behavioral responses. Fast and adequate responding to incoming stimuli is on one hand vital for survival and on the other hand is highly energy-demanding (Attwell and Laughlin, 2001). Therefore, continuous modulation of neuronal responsiveness is crucial to achieve a fine balance between speed, efficiency, selectivity and economy in the interaction of the brain with the outside world.
Various aspects of behavior, brain state and network activity account for the trial-to-trial variability in the responses to stimuli (Fontanini and Katz, 2008, Rector et al., 2005, Vyazovskiy et al., 2009a). Several animal studies reported consistent differences in the cortical evoked responses to sensory or electrical stimulation between wake and sleep. For example, cortical evoked potentials were larger in NREM sleep compared to waking in response to auditory clicks (Hall and Borbely, 1970), visual flashes (Galambos et al., 1994) or electrical pulses applied intracortically (Vanderwolf et al., 1987, Vyazovskiy et al., 2008). Such differences can be accounted for by neuronal bistability, manifested in an alternation between periods of neuronal depolarization (UP) and hyperpolarization (DOWN) states during sleep (Hill and Tononi, 2005). Indeed, studies in the rodent somatosensory system showed that postsynaptic potentials to brief whisker stimulation were largest and most reliable when evoked from DOWN states (Sachdev et al., 2004, Petersen et al., 2003). Consistently, slow waves, induced by local cortical electrical stimulation in spontaneously sleeping rats, were virtually absent if the stimulus was delivered immediately after a large spontaneous slow wave (Vyazovskiy et al., 2009a). Similarly, in humans the phase of the cortical slow oscillation determined the response evoked by auditory clicks (Massimini et al., 2003, Schabus et al., 2012). The differences in the neuronal responsiveness between UP and DOWN states might arise from the changes in various cellular properties, such as the fluctuations of the input resistance (Contreras et al., 1996, Leger et al., 2005).
Spontaneous cortical activity does not only reflect the current level of arousal or ongoing behavior, but is also influenced by preceding sleep-wake history (Borbely et al., 1984, Vyazovskiy et al., 2009b). For example, during sleep deprivation slower EEG activity (~2–6Hz), leaks into periods of waking (Franken et al., 1991, Vyazovskiy and Tobler, 2005), and it does so in a region-specific manner (Leemburg et al., 2010). We found recently that after sleep deprivation, local populations of cortical neurons in awake rats start undergoing brief OFF periods similar to those in NREM sleep, and their occurrence was associated with decreased behavioral performance (Vyazovskiy et al., 2011b). There are two basic questions that need to be addressed next. First, what are the factors leading to an occurrence of individual DOWN states at the individual neuron level in awake, but sleep-deprived animal? It has been suggested that DOWN states during sleep, as a network event, occur due to disfacilitation (Timofeev et al., 2001). Indeed, it is possible that a withdrawal of excitatory inputs impinging on a given cells could shift the balance towards inhibition (Stepanyants et al., 2009), resulting in a local bistability. However, while disfacilitation is likely crucial for the maintenance of the cell in a hyperpolarized state, active inhibition may play a crucial role in the initiation of an individual DOWN state. Indeed, it was found that in physiological conditions synaptic inhibition controlled the duration and the synchrony of active state termination (Chen et al., 2012). The second crucial question relates to the mechanisms underlying the link between “local sleep” and behavioral deficits after sleep deprivation (Vyazovskiy et al., 2011b). We hypothesize that the occurrence of local isolated DOWN states (i.e. periods of neuronal hyperpolarization), regardless of the specific underlying mechanisms, will likely affect the network activity and ultimately behaviour, by affecting neuronal responsiveness.
To test this hypothesis, we have chosen to use local cortical electrical stimulation for two reasons. First, electrical stimulation elicits periods of neuronal silence and hyperpolarization, similar to spontaneously occurring DOWN states (McCormick, 1992, Contreras and Steriade, 1995, Logothetis et al., 2010, Chung and Ferster, 1998, Vanderwolf et al., 1987). It is possible that the mechanisms underlying OFF periods evoked by electrical stimuli share some similarities with those behind spontaneous OFF periods, such as the recruitment of synaptic inhibition at its onset in response to initial excitation brought about by the electrical pulse or by increased network activity during the UP state. Second, local electrical cortical microstimulation is a promising approach to probe cortical responsiveness. While electrical stimulation may seem relatively unspecific, as it engages relatively large populations of neurons, involves both orthodromic as well as antidromic responses and entails feedforward and recurrent inhibition and excitation (Logothetis et al., 2010), it is not affected by the ecological significance of the stimulus, which is typical of sensory stimuli. In fact, it can be considered to some extent “physiological” inasmuch as it can efficiently elicit perception and even drive behavior (Otto et al., 2005, Romo et al., 1998). Moreover, external electrical stimulation can entrain cortical neurons in a behaviorally-specific manner (Frohlich and McCormick, 2010, Ozen et al., 2010), and efficiently interferes with responsiveness to naturalistic stimuli (Chung and Ferster, 1998). Crucially, electrical stimulation can be delivered in vivo, with standardized protocols that allow comparing neuronal responsiveness across a repertoire of different behavioral states. Therefore, this approach is useful to provide novel knowledge that complements that obtained by studying spontaneous cortical activity (Vyazovskiy et al., 2011b, Vyazovskiy et al., 2009b).
In this study, we measured cortical responses to brief electrical pulses at the beginning and the end of sleep deprivation at the level of local field potentials and individual neurons. We found that neuronal responsiveness is affected by sleep-wake history. We suggest that the well-known deficits in sensory, psychomotor and cognitive aspects of behavior after sleep deprivation may arise as a result of altered neuronal responsiveness to incoming stimuli.
Adult male WKY rats (total n=9) were used for this study. All rats were housed individually in transparent Plexiglas cages. Lighting and temperature were kept constant (LD 12:12, light on at 10am, 23±1°C; food and water available ad libitum and replaced daily at 10am. All procedures related to animal handling, surgery, recording etc. followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were in accordance with institutional guidelines. The animals were implanted with microwire arrays in the deep (V–VI) cortical layers of the frontal (B: +1–2 mm, L: 2–3mm) and the contralateral parietal cortex (B: −2–3mm, L: 4–5mm) for local field potential (LFP) and neuronal activity recordings, as previously described (Vyazovskiy et al., 2011b). Bipolar concentric electrode for stimulation was placed in the frontal cortex contralateral to the frontal microwire array (Fig. 1). The ground and reference screw electrodes were placed above the cerebellum as previously (Vyazovskiy et al., 2011b, Vyazovskiy et al., 2009b). Data acquisition and online spike sorting were performed with the Multichannel Neurophysiology Recording and Stimulation System (TDT). Subsequent offline spike sorting was performed by PCA followed by SMEM clustering algorithm as previously (Vyazovskiy et al., 2011b, Vyazovskiy et al., 2009b).
The total number of recorded neurons (n=9 rats) in all experiments was 504 in the frontal derivation and 312 in the parietal derivation, including well-isolated units and multiunit clusters. The individual neurons were subdivided into specific subtypes based on the wave-shape of their action potential and the firing pattern. Previous in vitro and in vivo studies classified neurons into three main categories: regular-spiking (RS), intrinsically bursting (IB) and fast-spiking (FS) neurons (Connors and Gutnick, 1990, Bartho et al., 2004). It is generally accepted that broad action potentials occurring at low frequency are produced by pyramidal excitatory cells, while narrow spikes occurring at faster frequency belong to GABA-ergic interneurons. In our recordings all individual neurons were visually screened and classified as a specific subtype (Fig. 2 A,B,C) based on the shape of the extracellular action potential and the histogram of interspike intervals. The specific measures that were taken into account were: i) the width of the negative and the positive deflection, ii) the overall spike shape, e.g. the slopes of each of the deflections; and iii) the peak and the width of the ISI distribution.
In naturally sleeping animals extracellular recordings in the neocortex reveal periods of synchronous activity among neuronal populations, interrupted by periods of population silence of variable duration (e.g. (Ji and Wilson, 2007, Luczak et al., 2007, Vyazovskiy et al., 2009b)). Consistently, we observed similar periods of generalized neuronal activity (ON periods) and silence (OFF periods) in our recordings from the frontal and the parietal cortex. Such neuronal OFF periods occur also in waking (Okun et al., 2010, Poulet and Petersen, 2008, Vyazovskiy et al., 2011b). In order to detect spontaneous and evoked OFF periods, all time stamps corresponding to individual detected spike occurrences were concatenated across all recording channels showing robust single- or multiunit activity. Next, onset and offset of the periods with no unit activity were identified as previously (Vyazovskiy et al., 2011b, Vyazovskiy et al., 2009b). OFF periods were quantified separately before the electrical stimuli (spontaneous OFF periods) and after the stimuli (evoked OFF periods). To investigate the changes in neuronal synchronization, we computed the latency of the first and last spike of each unit from the onset of population ON or OFF periods, respectively. The variability was defined as the standard deviation in the time of entry into ON and OFF periods between the individual neurons. Since higher variability would correspond to reduced synchrony, we defined the latter as 1/variability as previously (Vyazovskiy et al., 2009b). Several experiments (1–3 per animal) have been performed to assure stability and reproducibility of the results across time (at least 5 days between consecutive experiments). In only n=7 out of n=9 rats both the signals in the frontal and in the parietal derivation could be analyzed. Statistical differences were assessed with paired t-tests (specific p-values are reported).
Prior to the analysis of neuronal activity, vigilance states were identified for consecutive 4-s epochs. To do so, signals were loaded with custom-written Matlab programs using standard TDT routines, and subsequently transformed into the European Data Format (EDF) with Neurotraces software (www.neurotraces.com). Sleep stages were scored off-line by visual inspection of 4-sec epochs (SleepSign, Kissei), where the EEG, LFP, EMG and spike-activity were displayed simultaneously. Waking was characterized by low voltage, high frequency EEG pattern and high-amplitude, phasic EMG activity. Epochs of eating, drinking and intense grooming were carefully excluded (< 5%), since during those periods MUA is contaminated by movement artifacts, for example due to chewing, precluding reliable isolation of individual spikes. NREM sleep was characterized by the occurrence of high amplitude slow waves and low tonic EMG activity (Leemburg et al., 2010, Vyazovskiy et al., 2009b). During REM sleep the EEG/LFP was similar to that during waking, but only heart beats and occasional twitches were evident in the EMG signal.
About one week was allowed for recovery after surgery, and experiments were started only after the sleep/waking cycle had fully normalized, as evidenced by the entrainment of sleep and wake by the light/dark cycle and the homeostatic time course of sleep SWA. After a stable baseline, animals were recorded during 4 h period of continuous wakefulness starting at light onset. The animals were spontaneously awake and their behavior as well as their polysomnographic recordings were under constant visual observation. Sleep was prevented by providing the rats with novel objects because this method mimics naturalistic conditions of wakefulness, is effective, ethologically relevant, and does not appear to stress the animals (Palchykova et al., 2006, Tobler and Borbely, 1986, Vyazovskiy et al., 2007). The rats were well habituated to the experimenter and to the exposure to the novel objects (exposure at light onset for 30min/day for several days; new objects every day) prior to the experiment. Novel objects included nesting and bedding material from other rat cages, wooden blocks, small rubber balls, plastic, metallic, wooden, or paper boxes and tubes of different shape and color. Video recordings were performed continuously with infrared cameras (OptiView Technologies, Inc) and stored in real time with 25 frames/sec resolution.
Local electrical stimulation is a suitable tool to investigate sensitivity, or responsiveness of neurons to afferent stimuli in a living brain and in a highly standardized and precise manner (Logothetis et al., 2010, Vyazovskiy et al., 2008, Vyazovskiy et al., 2009a). The collection of LFPs and neuronal evoked responses occurred at light onset (W0) and after about 4 h of continuous waking (W4, Fig. 1). On a control day, the responses were collected at light onset and then after 4 hours of undisturbed waking and sleep (the mean amount of NREM sleep during this interval was 143.7±8.5 min). Responses were collected from the contralateral homotopic frontal cortex and from the ipsilateral parietal area (Fig. 1) to the site of stimulation. This stimulation paradigm was chosen because the corpus callosum is the major brain commissure, and is composed of well organized, uniform excitatory fibers that establish direct monosynaptic excitatory cortical connections between homotopic areas, increasing the likelihood of obtaining monosynaptic responses in the contralateral frontal cortex (Vanderwolf et al., 1987, Vyazovskiy et al., 2008). Moreover, contralateral evoked responses have been shown to increase after sleep deprivation (Vyazovskiy et al., 2008, Vyazovskiy et al., 2011a). Prior to the experiment, input-output tests were performed in each rat. For the final analysis one intensity level was selected, corresponding to ~ 50% of the intensity eliciting maximal response below the motor threshold. S88 Dual Output Square Pulse Stimulator and stimulus isolating unit (PSIU-6; Grass-Telefactor, AstroMed, Inc.) were used for electrical stimulation, which consisted of a monophasic squared pulse of 0.1 ms duration. In an attempt to stimulate, primarily, the bodies of cortical pyramidal cells, the electrical pulse was applied to the deep part of the bipolar electrode, and was referenced to the superficial part of the bipolar electrode. Thus, the current flow was oriented in a plane perpendicular to the surface of the cortex, approximating the flow along the cell body and the apical dendrites. Each recording session lasted up to 3–5 min, during which a sufficient number of artifact-free responses was collected (W0=39.5±1.7, W4=41.6±1.4, on average 7.18±1.2 s between two consecutive stimuli). Stimulation was performed always in quiet immobile waking state. To standardize this behavioral state as much as possible, behavior, EEG, and EMG activity of each rat were under constant visual observation for the preceding 10 minutes and for the duration of the entire stimulation session. Direct contact with the animals to induce arousal was always avoided. Stimuli were delivered manually only when the rat was quietly awake (immobile, eyes open, low EMG tone, low-voltage, high-frequency cortical EEG activity). We always tried to minimize the number of stimuli per session, to avoid potential damage to the tissue or the induction of long-lasting changes in excitability. There was no noticeable systematic change in any parameter of the neuronal or LFP responses within a recording session. In addition, in the main experiment, where inter-stimulus intervals were long (> 4 s) the responses to a given stimulus were not affected by the timing to the preceding stimulus. A separate experiment was specifically designed to assess the effects of induced OFF periods on the neuronal responsiveness, and a 50-ms inter-stimulus interval was chosen (Fig. 5A). Based on the average duration of the evoked OFF period in the present and previously published studies (~100 ms) (e.g. (Vanderwolf et al., 1987, Vyazovskiy et al., 2011a)), we chose intervals of 50 ms, to insure that in most cases the second stimulus would occur when most neurons were silent and presumably hyperpolarized. The intensity of the stimulation was always low (see above), so that it did not produce any noticeable changes in the motor output and did not affect behaviour in any way. Importantly, the LFP signals recorded from the stimulation electrode before and after the stimulation session did not show any systematic changes in the amplitude, indicating that the local stimulation neither resulted in a cortical damage nor produced any long-lasting changes in local cortical excitability. After each session, all individual responses were visually screened, and only those that were recorded on a background of stable low-amplitude, high-frequency cortical activity were retained for further analysis.
Most neurons in both cortical areas responded to the electrical pulses with a short latency ranging from 3 to 20 ms, followed by prolonged (~50–150 ms) generalized neuronal silence or suppressed neuronal firing (Fig. 1C,D). The initial early response of cortical neurons (Fig 1D) was characterized by an increased firing frequency – the average ISIs during the first 20 ms after the electrical pulse was on the order of a few milliseconds (5.8±0.3ms), suggesting that neurons often responded with a burst. The latency and the duration of the initial response indicate that it was a mixture of monosynaptic and polysynaptic responses.
Individual neurons were highly variable with respect to the latency of responding to the electrical pulses: some neurons failed to respond, some exhibited early phasic increase in spiking activity and some showed tonic sustained response (Fig. 2A,B). Consistent differences were found between neuronal subtypes when all individual well-isolated units were subdivided based on the shape of the action potential and ISIs (Fig. 2A,B,C). Notably, putative excitatory neurons (type B, 182 neurons, Fig. 2B) responded with a somewhat shorter latency than putative inhibitory neurons (type A, 44 neurons, Fig. 2A). Specifically, at W0 the corresponding latencies were 9.5±0.4 and 8.8±0.2 ms for types A and B respectively (p=0.14) and the difference became significant at W4 (10.1±0.4 vs 8.6±0.2 ms, p=4.8038e-004). Apart from the neurons with narrow and broad spikes, we consistently observed a third distinct subtype that responded to the stimuli with a latency, similar to type B neurons (W0: 8.7±0.2 ms, W4: 8.6±0.2 ms, 47 neurons) or with a biphasic response (Fig. 2C). The average firing rates of the three neuronal subtypes were as follows: A: 10.5±1.2, B: 7.4±0.5, C: 13.0±1.3 Hz. The same three neuronal subtypes were also observed in the parietal derivation, where they responded with a somewhat longer latency as compared to the frontal neurons (W0: 10.4±0.4, 11.7±0.3 and 11.8±0.3 ms; W4: 11.2±0.4, 11.0±0.3 and 11.5±0.3 ms for types A (41 neuron), B (118 neurons) and C (32 neurons) respectively). Thus, the data suggest that electrical stimulation does not lead to a stereotypic unspecific response of an entire neuronal population in the contralateral cortical area, but leads to a characteristic sequence of involvement of different neuronal subtypes that unfolds in time. Specifically, the initial excitatory response is followed by a recruitment of putative inhibitory neurons, which may mediate subsequent prolonged neuronal silence (Fig. 2D,E). This interpretation was confirmed by computing a firing profile of all neurons with broad and narrow spike waveforms (Fig. 2E) that showed an earlier increase in firing and stronger subsequent suppression of spiking activity for putative excitatory neurons, but more sustained spiking activity in putative inhibitory neurons.
An inspection of individual trials revealed that after 4-h wakefulness (W4) the electrical pulse often induced a stronger initial increase in neuronal firing, followed by a longer silent period (Fig. 3A). Moreover, quantitative analysis showed that after the period of waking neurons in the frontal cortex responded to the stimuli with moderately, but significantly shorter latency (by ~10%; Fig. 3B), while evoked firing was significantly enhanced by 32% (p<0.01) (Fig. 3C). In the parietal derivation, the latency to the first spike decreased from W0 to W4 at a statistical tendency level (p=0.06). Stimulation, performed on the control day, when the animals were allowed to sleep ad lib (on average ~ 2 h of sleep) between the two recording sessions did not reveal significant changes neither with respect to the latency nor with respect to the evoked firing (not shown).
Thus, electrical stimulation revealed that, consistent with the findings in spontaneous brain activity (Vyazovskiy et al., 2011b, Vyazovskiy et al., 2009b), cortical neurons after prolonged waking respond more readily, suggesting increased cortical excitability.
The initial strong increase in neuronal firing after the stimulus was often followed by a conspicuous suppression of activity (evoked OFF-period) in the time window starting from ~10 ms and lasting ~100 ms after the electrical pulse, corresponding to a positive LFP wave (Fig. 1C, ,4A).4A). In order to establish the relationship between neuronal activity and the size of the LFPs, we subdivided all individual responses into ten 10% percentiles according to the neuronal firing rates in the window between 10 and 100 ms post stimulus. During this window, some neurons were completely silent (see example in Fig. 1C). However, in many cases there was some residual firing activity, and the individual trials and individual neurons were highly variable with respect to the duration of the silent period (Fig. 2). As expected, larger LFP amplitudes were consistently associated with longer and more pronounced suppression of neuronal firing (Fig. 4A,B). This result suggested that, consistent with spontaneous slow waves (Vyazovskiy et al., 2009b), higher amplitude of evoked LFP events correspond to longer periods of neuronal inactivity.
Since the evoked silent period followed after a distinct enhancement in putative inhibitory neurons activity (Fig. 2), it is possible that it was brought about by GABA-mediated inhibition. We hypothesized that if common mechanisms underlie spontaneous OFF periods and evoked OFF periods, then electrical stimuli of the same intensity should result in a longer and more synchronous suppression of neuronal firing at W4 than at W0. Therefore, next, we compared the average amplitudes of the LFP and evoked OFF periods between W0 and W4 conditions. We found that the amplitude of the LFP after waking was enhanced moderately, but significantly by ~7.0±2.3% (p=0.01), suggesting a more profound inhibition of neuronal activity evoked by electrical pulses when physiological sleep pressure is increased. Consistently, as compared to W0, the evoked OFF periods at the end of the waking period were also substantially longer by 23.6±6.5% (Fig. 4C) and occurred more frequently (increase by 32.6±14.1%). In the parietal derivation, the duration of the evoked OFF period also increased moderately (p=0.05) from W0 to W4. Notably, at W4 the onset of the evoked OFF period in the frontal derivation was also better synchronized between individual neurons (by 16.3±5.8%, Fig. 4C), as it was also the case for spontaneous OFF periods (Vyazovskiy et al., 2009b).
Finally, we computed a ratio between the number of spikes per ms in the first 10 ms after the stimulus, e.g. when induced firing is strongest, and the number of spikes in subsequent 50 ms, when spiking was suppressed in most cases. This ratio increased substantially (by 23.5 ±4.9%) and significantly (p=0.0006) from W0 to W4, indicating that after prolonged waking the electrical stimuli elicit stronger initial excitation followed by a more profound inhibition. No significant differences in any of these parameters were found on the control day, when the animals were allowed to sleep ad lib (not shown).
The crucial question is: what are the consequences of the occurrence of neuronal OFF periods (or prolonged periods of hyperpolarization) on neuronal function? On one hand, the occurrence of such periods of inhibition have been shown to temporarily disrupt local cortical interactions (Logothetis et al., 2010), and affect neuronal responsiveness to sensory stimuli (Haider et al., 2007). On the other hand, several studies suggest that synaptic responsiveness to stimulation arriving during the hyperpolarized state is actually enhanced (Crochet et al., 2005).
In order to test neuronal responsiveness specifically during evoked OFF periods we applied double electrical pulses with inter-pulse intervals of 50 ms, so that the timing of the second pulse corresponded to the period of neuronal inhibition (Fig. 5A). The initial response to the first pulse was not markedly higher than the background activity in either derivation (frontal: background activity: 4.9±1.0 Hz, response: 8.2±2.4 Hz, p=0.17; parietal: background: 3.8±1.1 Hz, response: 4.1±1.2, p=0.75), and not different between the derivations (p=0.37, unpaired t-test). However, the responses to the second pulse were consistently increased in both the frontal and the parietal areas as compared to the first pulse, albeit to a different degree. Specifically, we found that the initial response to the 2nd pulse in the contralateral frontal cortex was stronger for most neurons, but only to a moderate extent (n=11 neurons, 2nd vs 1st pulse: +18.4±8.2%, p=0.05). Instead, in the ipsilateral parietal cortex the second pulse, applied during the evoked OFF period, elicited much stronger responses as compared to the first pulse (n=7 neurons, 2nd vs 1st pulse, +73.0±19.6%, p=0.0097). The difference between the derivations in the 2nd to 1st pulse ratio reached statistical tendency (p<0.1). Thus, the data suggest that the responsiveness of cortical neurons is determined by the occurrence of OFF periods.
Next, we hypothesized that if individual neurons are indeed bistable after prolonged waking and show repeated brief episodes of hyperpolarization, then their responsiveness should be affected. As previously (Vyazovskiy et al., 2011b), we found that the number of spontaneous OFF periods before the stimuli (Fig. 5B) increased significantly from W0 to W4 (p<0.05). When we compared those trials with only a few spontaneous OFF periods (‘low’, bottom 20% of the distribution) with the trials with many spontaneous OFF periods (‘high’, top 20% of the distribution), we found that evoked firing rates (relative to pre-stimulus level), were significantly higher for the latter (Fig. 5C, left). Moreover, the trials with many spontaneous OFF periods showed tighter synchrony between the responses of individual neurons (Fig. 5C, right), suggesting more stereotypic neuronal responses to afferent stimuli on the background of a sleep-like state.
To rule out the possibility that the changes in the relative increase in firing rates after stimulus are driven solely by pre-stimulus firing activity, and is unrelated to the actual occurrence of OFF periods, we performed additional analyses. Specifically, we removed all the trials with OFF periods before the pulse, and subdivided the remaining trials based on the post-stimulus absolute firing rates into those with high and low firing (20.7±3.7 vs 32.6±4.7 Hz, ~60% difference, p=3.0728e-006). The corresponding pre-stimulus firing rates were virtually identical (12.4±1.1 vs 13.1±1.3 Hz, p=0.17), suggesting that pre-stimulus firing rates do not account for the post-stimulus differences in responses. Consistently, when we subdivided again all the trials without pre-stimulus OFF periods into those with high and low relative increase in post-stimulus firing (computed as a ratio to pre-stimulus firing rates: 1.5±0.2 vs 2.4±0.2, p=2.4777e-007), the corresponding pre-stimulus firing rates were again similar (13.0±1.4 vs 12.3±1.0, p=0.44). This analysis suggests that the capacity of the network to respond to the stimulus is not merely an artifact of pre-stimulus firing rates, but is determined primarily by the occurrence of OFF periods.
Then, in order to investigate absolute firing rates evoked by electrical pulses, we subdivided all trials for the W0 and W4 conditions into ten 10% percentiles as a function of the number of spontaneous OFF periods pre-stimulus, and computed the corresponding numbers of spikes in the first 10 ms post-stimulus. During baseline, as expected, a significant positive correlation was apparent, with more evoked spiking during those trials with frequent OFF periods (Fig. 5D, left). In contrast, after prolonged waking the evoked firing rates were consistently elevated above the baseline level (Fig. 5D, right). Finally, we investigated whether prolonged waking leads to more variable responses to electrical stimuli. Even during baseline, it was often apparent that some electrical stimuli did not elicit a noticeable response, while others evoked a burst of firing followed by a large full-fledge LFP wave and suppressed neuronal activity. We found that inter-trial variability in the initial increase of firing rates as compared to prestimulus level was 18% higher during W4 as compared to W0 (p=0.02), while the increase in the variability of the evoked OFF period duration reached 19% (p=0.07). Thus, frequent occurrence of local sleep-like states after wakefulness seems to lead to increased and more variable responsiveness of cortical neurons.
There is substantial evidence that spontaneous brain activity both in wake and in sleep changes as a function of preceding sleep/wake history. The main finding of this study was that prolonged wakefulness also affects cortical responsiveness to incoming stimuli, and these changes parallel those observed in spontaneous activity of cortical neurons. The data suggest that altered neuronal states after prolonged wakefulness lead to changes in the balance of local and global intracortical interactions. Such effects may be a primary consequence of sleep deprivation, and account for at least some of its detrimental effects on behavior and cognition.
Sleep/wake history has profound effects on brain activity and cognitive functions (Borbely et al., 1984, Leemburg et al., 2010, Vyazovskiy et al., 2011b, Dijk et al., 1992, Van Dongen et al., 2003, Van Dongen et al., 2011). However, little is known about the neuronal mechanisms underlying the behavioral and cognitive deficits typical for sleep deprivation. We found recently that neuronal firing rates and synchrony increase after waking and decrease after sleep (Vyazovskiy et al., 2009b), and that after sleep deprivation cortical neurons start exhibiting brief synchronous periods of silence, encompassing most or all neurons, similar to the OFF periods in NREM sleep (Vyazovskiy et al., 2011b). However, neither the mechanisms underlying these changes nor their consequences on the neuronal function are clear. The aim of this study was to investigate whether the changes in neuronal activity incurred during prolonged waking would be accompanied by altered neuronal responsiveness to afferent stimuli. As a tool to probe neuronal responsiveness, we used electrical microstimulation, which produces complex excitatory and inhibitory effects locally and in distant targets (Logothetis et al., 2010).
First, we found that after a sustained period of wakefulness most cortical neurons respond to the stimulation faster, stronger and more synchronously than in the control condition. This result is compatible with our earlier findings on spontaneous cortical activity (Vyazovskiy et al., 2009a, Vyazovskiy et al., 2009b, Vyazovskiy et al., 2007) and supports the notion that cortical excitability after prolonged waking is increased, and/or the intracortical connectivity is stronger (Vyazovskiy et al., 2008, Huber et al., 2012). It is unlikely that these effects are confounded by a circadian factor, as control recordings, performed at the same time of the day (4h after light onset), but after allowing spontaneous sleep ad lib, revealed no systematic changes in the neuronal responses to electrical stimuli. Moreover, the lack of effect in the absence of prolonged waking rules out possible changes in excitability or long-term plasticity induced by the procedure of electrical stimulation per se. The state-dependent change in the responsiveness may occur at the level of individual neurons, for example, via changes in membrane potential, synaptic strength and intrinsic conductances (Crochet et al., 2005).
Our second major observation was that after staying awake cortical neurons become more bistable, as they were more easily activated if they were relatively silent, and on the other hand, more prone to transition into a prolonged silence, reminiscent of a spontaneous down state (Chung and Ferster, 1998, Contreras et al., 1996), after a period of activity. It has been shown that such “evoked OFF periods” are brought about by GABAB-mediated inhibition (Butovas et al., 2006), that produces a clear-cut prolonged hyperpolarization of the membrane potential indistinguishable from a spontaneous intracellular down-state (Contreras et al., 1996, McCormick, 1992).
We suggest that increased cortical bistability is the primary cause of the behavioral and cognitive deficits incurred after sleep deprivation, as it seems to affect neuronal responsiveness. Specifically, the occurrence of OFF periods, encompassing local neuronal populations, should on one hand result in the disruption of short-range neuronal interactions, since, by definition, all or most local neurons during such an OFF period are simultaneously deactivated or inhibited (Chauvette et al., 2011, Timofeev et al., 2001) and do not produce spikes (Logothetis et al., 2010). On the other hand, OFF periods can facilitate long-range responsiveness, because while hyperpolarized, neurons seem to be more easily excitable (Crochet et al., 2005); however, the only remaining source of the input in this case are distant neurons, not engaged in the local OFF periods. In physiological conditions, network activity and neuronal responsiveness are continuously modulated by the balance of excitation and inhibition (Haider et al., 2006), which is maintained in a broad dynamical range (Okun and Lampl, 2008). Our data suggest that after prolonged waking the balance between excitation and inhibition is maintained, but the stimuli elicit stronger excitatory responses that in turn recruit stronger inhibition (Fig. 4), leading to an occurrence of longer periods of generalized inactivity.
Thus, increased incidence of local and global OFF states after prolonged waking is likely a direct consequence of increased neuronal excitability and stronger intracortical functional connectivity. It is possible that, if local OFF periods are isolated and rare, they are quickly interrupted and dissolved by incoming sparse ongoing activity. However, when excitability, functional connectivity and bistability of cortical networks are increased, most intracortical interactions arise from distant sources and occur through strong volleys of activity often falling in the windows of increased responsiveness, thereby precipitating global cortical synchronization. While such scenario has not been proposed previously for waking state after sleep deprivation, it has been shown that cortical propagation of sleep slow waves is facilitated by long-range recruitment of distributed cortical networks in the global slow oscillation (Riedner et al., 2007, Vyazovskiy et al., 2011b).
What are the consequences of the bistability in cortical neuronal networks on behavior? We show here that inter-trial variability in neuronal responses in awake rats after prolonged wakefulness is increased. It is likely that state instability arises because after sleep deprivation, an alert and attentive waking cannot be easily sustained due to frequent emergence of local and global OFF periods (Vyazovskiy et al., 2011b, Van Dongen et al., 2011). Undoubtedly, if neuronal responsiveness to complex internal and external inputs is affected by staying awake for too long, it could account for many detrimental consequences of prolonged waking. Further experiments, exploiting specific sensory stimulation paradigms and behavioral tasks are needed to directly test this hypothesis
It should be noted that most effects found in this study are relatively modest, being in the range of 10–20%. The low magnitude of the effect is expected, as the duration of wakefulness in this study was close to the maximal duration of spontaneous waking bouts in rats. It is possible that the effects could have been more pronounced if we had used long-term instrumental sleep deprivation. However, in this study we strived to minimize the confounding factor of stress and other factors. It should also be noted that individual neurons were highly variable in their responses to the stimulation, as well as with respect to the effects of prolonged wakefulness. Such diverse effects can hardly be accounted for by stress, hormonal changes or other unspecific factors related to the procedure of keeping the animals awake. Instead, it can be hypothesized that since individual neurons in the recorded population differ with respect to their phenotype, location, pattern of connectivity, and function, they would also show different wake-and experience-related changes in their spontaneous and evoked firing. Thus, the pronounced natural variability in the responses of individual neurons can account for the relatively modest average effects. At the same time, this variability is on one hand a convincing argument against the role of unspecific factors, and, even more importantly, it represents a very powerful and yet totally unexplored tool for investigating specific activity-dependent cellular mechanisms of sleep regulation. One limitation of the current study is that the recordings were performed from the deep cortical layers only (layers V–VI). Since the cortical layers differ in the connectivity and play a differential role in sensory processing (Sakata and Harris, 2009), it will be important to extend the findings obtained in this study to the thalamorecipient and supragranular layers, and exploit more selective stimulation paradigms.
In summary, we found that cortical responsiveness to afferent stimuli is altered by preceding sleep/wake history. We suggest that the changes observed after prolonged waking in spontaneous brain activity are a manifestation of altered brain responsiveness that might underlie the well-known behavioral and cognitive deficits typical for sleep deprivation. Future studies are necessary to shed light on how the altered neuronal responsiveness translates in specific behavioral changes.
Supported by NIMH P20 MH077967 (CC), NIH Director’s Pioneer award (GT) and AFOSR FA9550-08-1-0244 (GT). We thank A. Nelson, Drs M Dash, U Faraguna and E Hanlon for help with the experiments.
COI statement: All authors indicated no financial conflicts of interest.