Slow waves during sleep in crayfish. Origin and spread

Some invertebrates have a resting state similar to sleep in vertebrates, including circadian periods of rest, homeostatic rebound, responses to pharmacological manipulations, decrease of threshold for sensory stimuli and expression of specific genes (Kaiser and Steiner-Kaiser, 1983; Kaiser, 1988; Tobler and Stalder, 1988; Hendricks et al., 2000; Shaw et al., 2000; Shaw and Franken, 2003; Sauer et al., 2004). However, notwithstanding efforts made with bees and flies (Kaiser and Steiner-Kaiser, 1983; Nitz et al., 2002; van Swinderen and Greenspan, 2003; van Swinderen et al., 2004), the brain electrical activity of the same invertebrate before and during sleep periods has been continuously recorded only in crayfish (Ramón et al., 2004). This is arguably the most important sign characterizing sleep, as it shows that this state is unique and different from the awake one, giving basis to the statement that sleep is of, by and for the brain (Hobson, 2005).

In the crustacean crayfish we have recorded the brain electrical activity continuously for periods of 24 h (Ramón et al., 2004). These records show that such activity in alert, active animals is comprised by numerous spikes (frequencies of 100–300 Hz) riding on an almost flat baseline. However, during some resting periods, slow waves unique to this state substitute the spikes recorded from alert animals. These slow waves are in the range of 15–20 Hz (Mendoza-Angeles et al., 2007).

Vertebrate sleep is well characterized and many of its generating and regulatory mechanisms are known, in spite that to date we do not know why we (or any other organism) sleep (Hobson, 2005; Mignot, 2008). Nonetheless, a common sign of sleep in vertebrates with and without cortical structures is a change in the brain electrical activity (Mignot, 2008), and currently there is only one invertebrate in which such correlation has been established – crayfish (Ramón et al., 2004).

During sleep the vertebrate brain also generates slow waves with frequencies ranging from 0.5 Hz to 4 Hz and are due to the synaptic activity of thousands of synchronized thalamic and cortical neurons (Volgushev et al., 2006). As crayfish do not have cortex- or thalamic-like structures, questions arise as to the origin of the slow waves and relationship with sleep. Thus, to study how sleep might have developed during evolution, it is important to determine how the slow waves appear and move through the invertebrate brain.

To study the slow waves generated during sleep, we have conducted experiments with a preparation of tethered crayfish that sleeps and that allow us an extent view of the brain, such that we can record the electrical activity of well-identified brain regions. Results show that slow waves are generated at the central complex of the brain in awake animals and that sleep occurs when these waves spread to deuto- and then to protocerebrum.

Experiments were performed on healthy adult male crayfish Procambarus clarkii, Girard (7–10 cm rostrum to tail) with complete appendages. Since their arrival to the laboratory, animals were maintained in communal aquaria with well-aerated tap water under natural light–dark cycles, and fed twice a week with cat chow.

Records were obtained from the brains of 50 unrestrained animals using a procedure described elsewhere (Ramón et al., 2004). Briefly, in cold-anesthetized crayfish we opened a hole in the dorsal cephalothorax to introduce a stainless-steel cannula carrying up to three Teflon-coated platinum–iridium wires (75 μm diameter) protruding 0.1 mm. The cannula (1 mm, external diameter) was introduced under visual control through a stereoscopic microscope (Wild Heerbrugg, Gais, Switzerland), placed on identified brain regions (protocerebrum, deutocerebrum, tritocerebrum) and fixed to the cephalothorax with dental cement (Acrimin Autocurable, México City, México); the cannula was used as a reference electrode. After surgery animals were returned to their aquarium and kept unhandled for at least 24 h. For recording, animals were connected to the amplifiers via a thin flexible cable that did not restrain walking, changes in position or tail flipping responses; animals behaved as normal, unrestrained ones.

To obtain records from identified areas of the brain we used a tethered crayfish preparation as follows. The day before the experiment we glued a cork to the lateral carapace of an intact crayfish. During the experiment the cork was held by an adjustable bar, keeping the animal ‘floating’ in water on a fixed position and immersed up to the rostrum, unable to displace around the aquarium, to change position or even to touch the bottom or the walls of the aquarium.

To expose the brain and place the electrodes on it we opened a hole in the dorsal carapace of a cold-anesthetized crayfish in the experimental chamber. A plastic retractor introduced through the hole was used to push muscles and stomach away. With a microscope we visualized and identified different regions of the brain, which was not desheathed to maintain its hemolymph circulation. The ensemble electrode had two or three glued stainless-steel cannulae of 1 mm external diameter, separated by 1.5 mm and carrying inside a Teflon-coated platinum–iridium wire (75 μm diameter). The wire tips protruded about 0.5 mm from the cannulae and penetrated the brain sheath. The cannulae were used as reference electrodes (Fig. 1).

The cavity was filled with saline solution and, when the electrode was not moved, covered with Parafilm (American Can Co., Greenwich, CT, USA). Animals survived surgery well and could be recorded the following 2–3 days (N=20).

This preparation allowed us to move the electrodes and simultaneously record from two or three brain regions. Records were obtained from the lateral protocerebrum (E1 and E3), protocerebral bridge (E2), central complex of the brain (E4), deutocerebrum (mainly accessory lobe, E5 and E6) and tritocerebrum (E7 and E8) of 20 animals (Fig. 2). As symmetrical brain regions generate very similar records, only those from one side are shown in the figures below.

The tethered preparation also allowed us to obtain records from the areas previously mentioned, while a selected region was cooled by placing on it a Teflon-coated (75 μm diameter) platinum wire, whose other end was submerged in ice slush at 4°C. We found the temperature in the tip of the wire to be about 10°C, which importantly reduced the temperature of the recorded areas. Chilling effects disappeared approximately 30 min after the wire was removed from the cooled region.

Sensory stimulation was accomplished by trains of light flashes of 10 μs duration at a frequency of 10 Hz, delivered by a reflector–diffuser lamp, 13 cm in diameter, connected to a photic-stimulator (PS22, Grass Instruments, Quincy, MA, USA) and placed 20–30 cm in front of the crayfish. Visual evoked potentials (VEPs) and omitted stimulus potentials (OSPs) were recorded. Typical experiments involved recording segments of 3 s duration: 1 s control prestimulus, 1 s during the stimuli, and 1 s after the train (delivered every 10 s). We did not average the responses and show the raw data.

The brain electrical activity recorded from both free (N=50) and tethered (N=20) animals was band-pass filtered (3–3000 Hz) and a 60-Hz notch filter was sometimes used. Records were amplified with AC preamplifiers (P511, Grass Instruments), displayed on an oscilloscope (DSO 1604, Gould Electronics, Essex, UK) and in parallel sampled by an A/D converter (NI-DAQ, PCI-6024E, NI Corp., Austin, TX, USA) at 2 kHz. Data were acquired with a Matlab (The MathWorks Inc., Natick, MA, USA) program-based software developed at our laboratory, and stored on a personal computer for off-line analysis. For detection of slow waves and spikes, a sampling rate of 2 kHz and filtering gave us enough information, as we have demonstrated previously (Ramón et al., 2004).

To analyze the data we used the wavelet transform (WT) described elsewhere (Mendoza-Angeles et al., 2005; Mendoza-Angeles et al., 2007). Briefly, the algorithm uses the Morlet wavelet as the mother wavelet to calculate the WT of a discrete time series f(n) with n=0, …N, where N is the number of points in the time series with equal time spacing δt. With this WT we obtained 3-D graphs that include samples, frequency and power (in the Figures below, only the 0–60 Hz band of interest is represented, and power is coded in color using the visible light spectrum).

Unrestrained animals were video recorded with an infrared video camera while simultaneously recording the brain electrical activity. Position versus time (day–hour) graphs allowed us to locate specific segments of the brain electrical signals corresponding to the two more relevant positions, i.e. standing up and lying on one side.

After placing the electrodes on selected regions of the brain the tethered animal was left undisturbed for at least 30 min. After that, for the first couple of hours the crayfish displays two main behaviors, i.e. active (alert) and motionless. During the alert periods, lasting up to 20–30 min, the animal moves the appendages in an apparently purposeless way, raising the chelae, waving the antennae or flicking the antennulae. Occasionally it becomes motionless and at this time their chelae and walking legs hang, while their antennae and antennulae float on the water surface. The tail extensor muscles tone decreases, as judged by the tail hanging in a slightly curled position. 1–10 min later the crayfish actively moves its appendages again.

A motionless, tethered crayfish could be sleeping but because they cannot lie on one side [the position with which we clearly identify this state (Mendoza-Angeles et al., 2007)], we needed a different way to assess sleep. For this purpose, we recorded the electrical signal indicative of a mild cognitive processing generated by the brain of alert, unrestrained animals in response to a train of light flashes (OSPs, see below), and that disappears when sleeping (Ramón et al., 2001).

Alert, unrestrained animals chronically implanted with electrodes on the brain detect flashes of light and process the resulting nervous signal. Detection of stimuli is ascertained by the presence of VEPs generated after each light flash, while processing is indicated by an OSP, a discharge at the end of the train with a fixed delay after the due-time of the first omitted stimulus. When unrestrained animals are lying on one side sleeping, processing by the brain ceases and there are no OSPs after the train of light flashes. Thus, to test such detection and processing of light in a tethered crayfish, we first stimulated the eyes with a light flash and recorded VEPs (Fig. 3A); we then stimulated with a train of light flashes at 10 Hz for 1 s and searched for the occurrence of OSPs (Fig. 3B).

Fig. 3A shows VEPs (inside the dotted frame) after a single light flash, recorded from the central complex (E4) and protocerebral bridge (E2) from the brain of alert, tethered crayfish, to a lesser degree from lateral protocerebrum (E1) and the accessory lobe of deutocerebrum (E6), and not from tritocerebrum (E7). By contrast, OSPs (inside the continuous frame) with characteristics similar to those from unrestrained animals were clearly recorded (Fig. 3B) only from the central complex of the brain (E4), the protocerebral bridge (E2), and somewhat from the accessory lobe of deutocerebrum (E6). These records are not averaged and correspond to a single experiment representative of those from 20 animals.

Motionless, tethered crayfish do not generate OSPs in response to a train of light flashes (data not shown), and because this result is indicative of sleep in an unrestrained crayfish, we interpreted it as showing that tethered crayfish are also able to sleep. Thus, we conducted experiments aimed to record the electrical activity from different regions of the crayfish brain.

The brain electrical activity from tethered crayfish during active periods is similar to that previously recorded from alert, unrestrained animals, i.e. numerous spikes riding on an almost flat baseline. The main difference with records from unrestrained animals is that spikes are larger and more numerous, probably due to electrode tips penetrating the brain sheath. Fig. 4A shows a typical record obtained from the accessory lobe of deutocerebrum of a tethered crayfish. Fig. 4B is a 3-D graph obtained from a WT analysis of a 600 s segment of brain electrical activity. Note that the highest power (red) corresponds to a frequency around 40 Hz and it is nearly constant during the 600 s recorded. This result is typical of alert animals, unrestrained or tethered.

After ascertaining that records from the brains of alert, tethered crayfish are similar to those from unrestrained animals, we recorded from the brain areas indicated in Fig. 2. Records from different brain areas of alert, tethered crayfish show a pattern of activity similar to that observed in unrestrained animals, i.e. numerous spikes on an almost flat baseline. Fig. 5A shows records from the following brain areas: lateral protocerebrum (E1), protocerebral bridge (E2), central complex of the brain (E4), accessory lobe of deutocerebrum (E6), and tritocerebrum (E7). Fig. 5B is the wavelet analysis of those records. Contralateral areas (E3, E5, E8) produced similar records and are not shown. Records from the lateral protocerebrum (E1), protocerebral bridge (E2) and accessory lobe of deutocerebrum (E6) of alert, tethered animals (Fig. 5A) show the pattern of activity characteristic of unrestrained crayfish, abundant spikes and no slow waves.

Records from the central complex of the brain (E4) have characteristics similar to those from unrestrained, sleeping animals, i.e. few spikes and slow waves of about 15–20 Hz. Records from tritocerebrum (E7) have fewer spikes, usually smaller in amplitude, and also slow waves with a frequency of 2–4 Hz, generated by the scaphognatites (Ramón et al., 2004).

Wavelet analysis of records from lateral protocerebrum, protocerebral bridge and accessory lobe of deutocerebrum show a high power for frequencies between 20 Hz and 40 Hz (Fig. 5B) and power values relatively low at frequencies over 50 Hz for all areas, except tritocerebrum. Such results are similar to those obtained from unrestrained animals. However, at the central complex of the brain (E4, Fig. 5B) there is a shift of the dominant frequencies, from 20–40 Hz to 15–20 Hz, and a large decrease in power at frequencies of 30–45 Hz (gray frame), typical of sleeping animals.

Motionless, tethered crayfish have a reduced number of brain spikes, slow waves in the recordings and no OSPs, an indication of processing visual stimuli. VEPs and OSPs are absent or strongly reduced (not shown), as it occurs in unrestrained crayfish lying on one side.

Furthermore, records from protocerebral bridge (E2) and accessory lobe of deutocerebrum (E6) are clearly similar to those from an unrestrained animal lying on one side (Fig. 6A). A wavelet analysis of these records (Fig. 6B) shows a shift of the dominant frequencies to 15–20 Hz and a decrease in power in the range 30–45 Hz (gray frames in Fig. 6B, E4 and E6), as seen also in unrestrained animals lying on one side.

In tethered crayfish restoration of movements (awakening), stiffening or moving of walking legs and chelae, elevation of antennulae and antennae, and rectification of the tail position is accompanied by the reappearance of spikes, shifting of frequencies and increase in power in the range of 30–45 Hz for all recorded areas but the central complex of the brain.

The records from the central complex of the brain show slow waves in both alert and sleeping animals, and the wavelet analysis exposes the similarity of their characteristics. However, in alert animals the waves are limited to the central complex of the brain, while in sleeping ones they are also recorded from the rest of the brain (excluding tritocerebrum). Therefore, we searched for the dynamics of the waves.

When one of a pair of electrodes was placed on the central complex of the brain and the other one on the accessory lobe of deutocerebrum, records obtained during the transition from an active to a motionless animal showed a surprising result (Fig. 7). Before the transition, the electrode on the central complex (E4) recorded slow waves, while that on the accessory lobe of deutocerebrum (E6) had only spikes, similar to records from an awake animal; however, after the transition (arrow in Fig. 7A) both electrodes recorded prominent slow waves. The wavelet analysis of the activity from the central complex of the brain before (Fig. 7B left, before the gray frame) and after the transition (Fig. 7B left, after the gray frame) shows a persistent high power for frequencies in the range 15–20 Hz, and a low power for those between the 30–45 Hz characteristic of sleep. The wavelet analysis for deutocerebral activity (E6) before transition (Fig. 7B right, before the gray frame) is typical of an alert animal. After transition (Fig. 7B right, after the gray frame), there is a reduction in power for all frequencies, a shift in the dominant frequency to 15–20 Hz and a strong decrease in power for the frequencies between 30 Hz and 45 Hz.

Results shown in Fig. 7 were clearly reproduced in eight animals. However, because they could also be due to our tethering of the crayfish, we went back to record unrestrained animals, placing the chronic recording electrodes on the same brain regions, central complex of the brain and accessory lobe of deutocerebrum.

Records from the central complex and accessory lobe of deutocerebrum of unrestrained crayfish during the transition from an alert animal to an animal lying on one side are similar to those obtained from tethered crayfish with electrodes on the same brain regions. These records are shown in Fig. 8A, where an electrode placed on the central complex of the brain (E4) shows continuous slow waves similar to those recorded from animals lying on one side. By contrast, the record from accessory lobe of deutocerebrum (E6) is initially comprised by spikes on a flat baseline, as in those from alert animals, but after the crayfish lies on one side (arrow in Fig. 8), slow waves similar to those from the central complex of the brain (E4) appear.

The wavelet analysis of records in Fig. 8A shows a large power for frequencies around 20 Hz in the record from E4 (Fig. 8B, left) before the animal changes position and lies on one side (gray frame). After the transition, the only change is a further decrease in power in the 40 Hz band. In the wavelet analysis of deutocerebral records, before the transition there is high power for frequencies in the 40 Hz band, while after the transition between the two behavioral states (gray frame in Fig. 8B, right) there is a decrease in power at those frequencies.

Results in Fig. 8 show that before the transition position slow waves are recorded only from the central complex (E4), while after the transition both, central complex (E4) and accessory lobe of deutocerebrum (E6), show similar slow waves. To statistically compare these results we performed a cluster analysis that searches for correlated data in an n-dimensional distribution. Each cluster is characterized by a centroid around which correlated data accumulates.

For this analysis we divided the WT time series in four frequency bands: B1, from 1 Hz to 5 Hz; B2, from 6 Hz to 15 Hz; B3, from 16 Hz to 30 Hz; and B4, from 30 Hz to 60 Hz. Because the B1 band corresponds to scaphognatite activity we eliminated it and worked only with the remaining bands. For each band we calculated the centroid and the minimum radius. The results of this clustering analysis show that deutocerebral data from standing up alert animals (triangles) form a cluster completely separate of data from animals lying on one side (diamonds) (Fig. 9). Furthermore, when the data from the central complex of the brain of limp tethered crayfish are added (squares), they become closely correlated with those from unrestrained sleeping animals (diamonds).

Records obtained from protocerebrum, central complex of the brain and the accessory lobe of deutocerebrum, with electrodes at E2, E4 and E6 from tethered or unrestrained sleeping crayfish, are remarkably similar in their low frequencies. This is shown in Fig. 10 that contains filtered records (cut-off at 50 Hz) from tethered crayfish after they transit between the two body positions (active and passive) that we used as indicative of alert and sleep brain states.

Fig. 10 shows that slow waves recorded by electrodes at the central complex of the brain and accessory lobe of deutocerebrum are more synchronous than between the central complex of the brain and protocerebrum. Choosing some prominent peaks (arrows) from these records we calculated a propagation delay, which between the central complex and accessory lobe of deutocerebrum has a mean value of 32 ms (range 2–68 ms, N=15) and between accessory lobe of deutocerebrum and protocerebrum a mean value of 2.34 ms (range 1–6 ms, N=6).

To further ascertain the behavior of slow waves in tethered crayfish before and after the behavioral transition from active (alert) to passive (sleep), we obtained the cross correlation between signals from proto- and the accessory lobe of deutocerebrum with those from the central complex. Fig. 11 shows that before the transition the correlation between the central complex of the brain and both, protocerebrum and the accessory lobe of deutocerebrum, is the same and low, while after transition (arrow) the accessory lobe of deutocerebrum is slightly better correlated to the central complex than the protocerebrum.

These results strongly suggest that the central complex of the brain is the main oscillating structure; therefore, if we block the activity of this area, cooling it for example, the rest of the brain would not oscillate. When in a restrained animal we chilled the central complex of the brain, the amplitude of the slow waves decreased and it can even disappear, they were not seen in deutocerebrum nor in protocerebrum, and the animal did not show any behavioral sign of sleep (Fig. 12). These results confirm the role of this area in the generation and sustaining the slow waves that spread to the whole brain during sleep.

In previous work we have addressed the issue of the origin of the electrical signals recorded with electrodes placed on the crayfish brain (Hernández et al., 1996; Ramón et al., 2004). As the electrodes are electrically isolated from the surrounding tissue by a steel cannula (reference electrode), they record only the local field potentials generated in the close vicinity, and this is also the case for the three cannulae ensemble that we used to record simultaneously from different brain areas. Therefore, the activity generated by other tissues (mainly muscles) is not included in our recordings, which are similar to those previously obtained from invertebrates (Bullock, 1945; Bullock and Basar, 1988).

Two other technical points deserve some discussion. Firstly, many spikes riding on an almost flat baseline comprise the brain electrical activity from alert crayfish and during some resting periods, this activity is substituted by a pattern of slow waves unique to this state. Ramon et al. reported the frequency of these signals in the range of 4–8 Hz (Ramon et al., 2004), which contrasts with that found in this work in the range of 15–20 Hz. The discrepancy can be explained because the first values were obtained analyzing the data with a Fast Fourier Transform (FFT) which, as we have previously demonstrated (Mendoza-Angeles et al., 2007), is not a satisfactory method for analysis of the non-stationary signals recorded from the crayfish brain. Because of this, we have now analyzed the data with a WT and obtained a value of 15–20 Hz for the dominant frequency of slow waves during sleep. We believe that the latter is the correct value.

Secondly, it is important to emphasize that the range of frequencies of the slow waves recorded from the crayfish brain is different from that recorded in vertebrate brains during sleep (0.5–4 Hz). Our use of the term ‘slow waves’ is based on that in crayfish, the brain electrical activity is comprised by high-frequency, non-synchronized events (mostly spikes with frequencies up to 200 Hz) (Hernández et al., 1996; Ramón et al., 2004); therefore, the waves characterizing sleep in crayfish are of a clearly lower frequency than the spikes, and we call them ‘slow waves’ with no intention to equate them with those from vertebrates or imply a similar generating mechanism.

Sleep has been seen as a behavior pertaining only to vertebrates, and there is an immense body of work related to its origin, nervous structures involved and possible functions. These results have been summarized in the phrase ‘sleep is of, by, and for the brain’ (Hobson, 2005; Mignot, 2008) and, as this statement implies, only animals with a brain would sleep. The problem of defining a brain can be resolved following Bullock and Horridge, who describe it as, ‘the enlarged and specialized rostralmost or highest ganglion of the central nervous system…’ (Bullock and Horridge, 1965). However, such a corollary does not take into account invertebrates, which then would not sleep, and to solve this it has been proposed that sleep shoud be redefined in such a way as to include vertebrates as well as invertebrates (Mignot, 2008). Thus, sleep can be defined by the association of rest behavior, reduced sensory response and the presence of a brain electrical activity unique for this state, criteria that crayfish fulfills (Ramón et al., 2004).

We have previously shown that unrestrained, chronically implanted crayfish with electrodes on the brain sleep (Ramón et al., 2004; Mendoza-Angeles et al., 2005; Mendoza-Angeles et al., 2007). However, the constraints of placing the electrodes in appropriate areas of the brain are limiting, as surgery has to be made rapidly, vision of the brain is partial because the hole for the electrode has to be small, and once the electrodes are implanted their position on the brain cannot be changed. Therefore, to improve this situation we developed a tethered crayfish preparation.

Tethered preparations are commonly used to study behaviors that would normally take the animal away from the recording site [i.e. locust (Robert and Rowell, 1992); fly (Liu et al., 2006); grasshopper (Hedwig and Heinrich, 2004)], and also for recording the electrical activity of the brain of small invertebrates [i.e. snails (Balaban and Maksimova, 1993); flies (Nitz et al., 2002; van Swinderen and Greenspan, 2003); ants (Ramón and Gronenberg, 2005)]. As these preparations substitute chronically implanted animals, which are very difficult to achieve in small invertebrates, they do not have the benefits that comparisons with records from free animals provide. To ameliorate this problem we have developed a chronically implanted, unrestrained crayfish preparation to investigate general characteristics of the sleep behavior, and a tethered preparation to record the electrical behavior from different brain areas during such periods.

Tethering crayfish did not prevent them from moving their appendages and only anchored them at a single location in the aquarium, as shown by the similarity between records of their brain electrical activity and those from alert, unrestrained animals (Figs 3, 4 and 5); relaxed tethered crayfish show slow waves similar to those recorded from unrestrained sleeping animals. Furthermore, wavelet analysis of records obtained from tethered preparations allowed us to identify the same features found in unrestrained animals, i.e. a general decrease in power for all the analyzed frequencies, particularly in the range 30–45 Hz (Fig. 6). Moreover, because OSPs are not generated in limp tethered preparations, implying a diminished ability of the animal to detect its environment, we conclude that tethered crayfish have periods of activity and also frequent and variable periods of time sleeping.

By simultaneously recording several areas of the alert, tethered crayfish brain, we show in this work that most regions generate the pattern of multiple spikes riding on an almost flat baseline common to invertebrates [p. 317 in Bullock and Horridge (Bullock and Horridge, 1965)]. By contrast, low-frequency signals vary among brain regions. The central complex of the brain (E4; Figs 5 and 6) generates slow waves in animals that are standing up (active) or animals that are lying on one side (passive). However, when a crayfish lies on one side or becomes limp, slow waves similar to those from E4 are recorded also at E6 (accessory lobe of deutocerebrum) in both animal preparations.

The differences in the electrical pattern of the central complex of the brain and the accessory lobe of deutocerebrum become more apparent when animals transit from an alert excited state to a motionless relaxed one. Before the position transition the central complex shows slow waves, and lateral deutocerebrum and protocerebrum shows numerous spikes on a flat baseline. After the position transition, spikes and slow waves are still seen at the central complex, while similar waves appear in the accessory lobe of deutocerebrum electrode, as confirmed by the results of the wavelet analysis (Fig. 7).

Our results show that independently of the brain state, active or passive, the central complex generates continuous oscillations that are not seen at other brain regions. Slow waves in the accessory lobe of deutocerebrum are seen while the animal is limp, and disappear when it becomes active. Lateral protocerebrum is not initially involved in these waves and a few milliseconds are required for them to invade it. Tritocerebrum never shows these slow waves, only those related to the scaphognatites respiratory activity (Ramón et al., 2004). However, occasionally a communication between the central complex and the rest of the brain opens up, and slow waves similar to those of the central complex (E4) appear at the deutocerebrum (E5 and E6), while the animal also shows all other signs of sleep. The blockade by cooling the central complex area stops spreading of the slow waves to the rest of the brain and other behavioral signs of sleep, indicating that this is the area where the slow waves originate.

A comparison between records in Fig. 10 makes apparent a non-uniform propagation of low frequency signals throughout the brain. Slow waves recorded from the central complex (E4) and the accessory lobe of deutocerebrum (E6) have a mean time difference of 32 ms, while between the accessory lobe of deutocerebrum and protocerebrum the difference is only of 2.3 ms. As the two electrodes used to record from those regions were the same and separated by 500–600 μm, it is clear that there is a delay in propagation between E4 and E6, as compared with E6 and E2. Assuming that fibers at the border region between the central complex and deutocerebrum of the crayfish brain are coupled by electrical synapses, a decrease in coupling would result in such propagation delay. We speculate that during sleep the slow waves reach the brain through neuron–glia interactions or by electrotonical coupling (Coles and Abbott, 1996; Traub et al., 2001). Cross-correlation shows that records from the accessory lobe of deutocerebrum are always better correlated to the central complex than those from protocerebrum, corroborating that in passive or active animals slow waves move from the central complex to deuto- and then to protocerebrum (Fig. 11).

Results from the clustering analysis (Fig. 9) show a clear difference between data from sleeping and alert animals, giving further support to the idea of two completely different brain states. Thus, the central complex seems to act as an oscillator that synchronizes the deutocerebrum during sleep.

Slow waves in response to appropriate stimulus have been recorded in many preparations from invertebrates to vertebrates and receive different interpretations (Gelperin and Tank, 1990; Delaney et al., 1994; Laurent and Naraghi, 1994; Sandeman and Sandeman, 1998; Prechtl et al., 2000; Greenspan and van Swinderen, 2004). However, spontaneous, self-sustained and continuous oscillations seem present only in the hippocampus of vertebrates (Fischer, 2004), where an intrinsic oscillator with synaptic interactions and facilitating gap junction mediate axo-axonic interactions. Models of spontaneous and self-sustained oscillations through axo-axonal gap junctions have been developed to explain such behavior (Lewis and Rinzel, 2000; Maex and De Schutter, 2007).

Sleep in invertebrates has been difficult to demonstrate and to be accepted by vertebrate physiologists. Since the early proposal from Kaiser and Steiner-Kaiser (Kaiser and Steiner-Kaiser, 1983), and from Tobler (Tobler, 1983) and Tobler and Stalder (Tobler and Stalder, 1988), for almost two decades there were no references to sleep in invertebrates despite efforts made on the issue (Kaiser, 1988). Perhaps one of the main reasons resides in the absence, in those works, of brain electrical recordings from unrestrained animals, which was difficult to obtain because of technical constraints. The multiple methods used by Hendricks et al. (Hendricks et al., 2000) and by Shaw et al. (Shaw et al., 2000) have provided more evidence on the existence of this function in invertebrates. Notwithstanding brain recordings from tethered mutant flies performed by Nitz et al. (Nitz et al., 2002), reversible changes in the brain electrical activity from unrestrained or tethered invertebrates have been demonstrated only in crayfish (Ramon et al., 2004; Mendoza-Angeles et al., 2007).

The slow waves recorded in crayfish do not have the same frequency as those recorded from vertebrate brains, and the brain activity from alert invertebrates is also different from similar records in vertebrates. Nevertheless, there are common features in both groups of animals. During sleep, the threshold for sensory stimuli increases, cognitive processes are reduced and there is a homeostatic regulation and changes in the brain electrical activity (Volgushev et al., 2006; Ramon et al., 2004; Sauer et al., 2004; Nitz et al., 2002; Hendricks et al., 2000; Shaw et al., 2000). However, because sleep is a function of, by, and for the brain (Hobson, 2005), this is the organ whose changes define the function. Changes in brain activity pattern, from high frequency to slow waves are common to both groups of animals and, giving the different structure of both kinds of brains (Bullock and Horridge, 1965), it would be naïve to expect that the waves (whether slow or fast) would have the same frequency values. However, it is possible to speculate that a similar central oscillator is at the core of the synchronization producing the slower frequency waves in both groups of animals.

A puzzling finding during this work is the continuous rhythmic activity in the central complex of the brain, although persistent oscillatory electrical activity is not uncommon. For example, neurons from several nervous regions show electrical oscillations even when isolated, such as those from the hippocampus (Buzsáki, 2002; Goutagny et al., 2009; Rácz et al., 2009), cerebellum (D'Angelo et al., 2009) and nucleus accumbens (Hunt et al., 2009). In some cases the oscillations change during the sleep–wake cycle and persist even under sedation (Hunt et al., 2009), originating speculations about their origin and function. However, there are examples of neuronal spontaneous electrical oscillations whose function is known, such as neurons from the isolated suprachiasmatic nucleus, which become coupled to perform two main functions, i.e. timekeeping and controlling the circadian overt rhythm (Colwell, 2000; Silver and Schwartz, 2005; Hu et al., 2007; Bernard et al., 2007).

Although we do not know the functions of the oscillations recorded from neurons in the central complex of the crayfish brain, the following speculation seems reasonable. In crayfish the central complex of the brain, with its multiple synaptic arborizations (Sandeman et al., 1995; Sandeman and Sandeman, 1998; Utting et al., 2000), sensory inputs, and the possibility of axo-axonal and neuron–glia interactions via gap junctions, could provide the basis for a dynamic oscillator that synchronizes wide areas of the brain. We speculate that as the ‘need for sleep’ increases in the animal, the central complex oscillations begin to block the sensory inputs, then they spread, more sensory inputs are blocked and, finally, the whole brain (proto- and deutocerebrum) oscillates at the same frequency of the central complex. The oscillations decrease as synchronization eliminates an accumulated ‘factor’, the system reaches a reset point and the animal wakes up, only to repeat the cycle each time that a similar excess occurs. In vertebrates adenosine has been proposed as a regulating factor of sleep (Basheer et al., 2004).

Supported by DGAPA IN202207-3 and Facultad de Medicina, UNAM.

 
• OSP

  omitted stimulus potential

 
• VEP

  visual evoked potential

 
• WT

  wavelet transform