Crayfish (Procambarus clarkii) have bilateral pairs of giant interneurons that control rapid escape movements in response to predatory threats. The medial giant neurons (MGs) can be made to fire an action potential by visual or tactile stimuli directed to the front of the animal and this leads to an escape tail-flip that thrusts the animal directly backward. The lateral giant neurons (LGs) can be made to fire an action potential by strong tactile stimuli directed to the rear of the animal, and this produces flexions of the abdomen that propel the crayfish upward and forward. These observations have led to the notion that the receptive fields of the giant neurons are locally restricted and do not overlap with each other. Using extra- and intracellular electrophysiology in whole animal preparations of juvenile crayfish, we found that the receptive fields of the LGs are far more extensive than previously assumed. The LGs receive excitatory inputs from descending interneurons originating in the brain; these interneurons can be activated by stimulation of the antenna II nerve or the protocerebral tract. In our experiments, descending inputs alone could not cause action potentials in the LGs, but when paired with excitatory postsynaptic potentials elicited by stimulation of tail afferents, the inputs summed to yield firing. Thus the LG escape neurons integrate sensory information received through both rostral and caudal receptive fields, and excitatory inputs that are activated rostrally can bring the LGs' membrane potential closer to threshold. This enhances the animal's sensitivity to an approaching predator, a finding that may generalize to other species with similarly organized escape systems.
Juvenile crayfish typically inhabit the more shallow areas of streams and ponds and are most commonly preyed on by fish, wading birds, and mammals (Correia 2001; Davis and Huber 2007; Englund and Krupa 2000). The response to attacks from these predators must be prompt and appropriate to ensure survival and future reproductive success (Lima and Dill 1990). Studies that examined predation on crayfish in natural habitats reveal little detail of the actual escape behavior but emphasize the ecological implications of predator-prey interactions (e.g., Blake and Hart 1995; Garvey et al. 1994; Söderbäck 1994; Stein and Magnuson 1976). Studying escape behavior of crayfish in the laboratory, however, has been a popular choice. In addition to stereotyped behaviors that can be repeatedly activated, the underlying neural circuits are characterized by large, identifiable neurons that can be easily accessed in dissected preparations (Edwards and Herberholz 2005; Edwards et al. 1999; Herberholz 2007; Krasne and Edwards 2002a; Wine and Krasne 1982). Moreover, escape circuit activation can be recorded in freely behaving animals allowing for a noninvasive measure of neural activity patterns under more natural conditions (Herberholz 2009; Herberholz et al. 2001, 2004; Liden and Herberholz 2008). The crayfish escape systems also provide a rare opportunity to test the effects of neuromodulators on identified neurons and small neural networks (Antonsen and Edwards 2007; Edwards et al. 2002; Teshiba et al. 2001; Yeh et al. 1996).
The fastest and most powerful escape responses in crayfish are generated by pairs of bilateral command neurons, the lateral giant interneurons (LGs) and the medial giant interneurons (MGs). More than 60 years ago, it was first discovered that firing of any of the crayfish's giant interneurons caused stereotyped tail flexions and that the LGs could only be recruited by stimulation of the posterior half of the body and the MGs only by stimulation of the anterior half (Wiersma 1947, 1961). Since then, numerous behavioral and physiological studies have confirmed this relationship between stimulus orientation and activation of the giant interneurons. Only tactile stimuli directed to the rear of the crayfish can fire the LGs and elicit an escape tail-flip that pitches the animal up- and forward away from the stimulus. MG-mediated tail-flips, however, only happen in response to strong tactile or visual stimuli directed to the front and propel the crayfish backward (Edwards et al. 1999; Liden and Herberholz 2008; Wine and Krasne 1972, 1982). In addition, attacks from natural predators evoke LG tail-flips when the attack is directed at the tail and abdomen or MG tail-flips when the attack is directed toward the head and thorax (Herberholz et al. 2004).
Of the two giant circuits, the LG circuit has received most experimental attention due to the accessibility of its individual components, and it is now considered one of the best understood neural circuits in the entire animal kingdom (Edwards et al. 1999; Krasne and Edwards 2002a,b). Previously identified excitatory inputs to the LGs are purely mechanosensory, and they are generated by primary afferents and sensory interneurons located in the abdomen and tail (Wine and Krasne 1972). The primary afferents are excited by movement of sensory hairs and by mechanical displacement of tail segments (Ebina and Wiese 1984; Newland et al. 1997), and they are coupled to each other to amplify the excitatory input to the LGs (Antonsen et al. 2005; Herberholz et al. 2002). The afferents connect directly to the LGs (Araki and Nagayama 2003; Zucker et al. 1971) and to mechanosensory interneurons; if a critical number of these interneurons is recruited, they cause an action potential in the LGs (Zucker 1972; Zucker et al. 1971). The LGs then make powerful electrical output connections to pairs of giant motor neurons (MoGs), which innervate flexor muscles that bend the abdominal segments (Mittenthal and Wine 1973).
Together, these experiments further supported the notion that the receptive fields of the giant interneurons are nonoverlapping and the LGs' excitatory inputs are confined to the abdomen and tail. However, the possibility that the LGs receive excitatory inputs that descend from rostral sensory systems was never excluded because most studies on the LG escape circuit were performed in isolated crayfish abdomens where any descending inputs were eliminated (e.g., Araki and Nagayama 2005; Krasne 1969; Roberts 1968; Yeh et al. 1997), or the existence of excitatory inputs from the front was not explicitly measured (Herberholz et al. 2004; Wine and Krasne 1972). Descending inputs to the LGs that originate in the rostral areas of the nervous system have been described; however, they are considered to be entirely inhibitory (Krasne and Teshiba 1995; Krasne and Wine 1975; Vu and Krasne 1992; Vu et al. 1993). Here, we show for the first time that the LGs receive subthreshold excitatory inputs from descending interneurons that are activated by stimulation of rostrally located sensory systems. These additional inputs bring the LG closer to threshold and, when activated by an approaching predator, could prepare the animal for an attack that is ultimately directed toward its abdomen or tail. Parts of the results have been published previously in abstract form (Herberholz and Liu 2008).
Animals and dissecting procedure
Juvenile crayfish (Procambarus clarkii) of both sexes were obtained from a commercial supplier and kept in communal tanks before the experiments. Animals were fed medium-sized shrimp pellets twice a week and kept under a constant 12 h:12 h light-dark cycle. Crayfish used in experiments (n = 36; size: 3.7 – 4.3 cm; measured from rostrum to telson) were thoroughly checked for intactness, and those missing limbs or sensory appendages and those with soft exoskeletons, indicating a recent molt, were excluded.
Minimal invasive surgery was used to provide local access to the nervous system in largely intact whole animal preparations. The dissecting procedure was modified using a previously described technique (Glantz and Viancour 1983; Herberholz and Edwards 2005; Herberholz and Liu 2008). Animals were anesthetized by placing them into small plastic tubes and inserting the tubes into crushed ice. The temperature inside the tubes was monitored using a thermometer, and animals were kept in the ice until the temperature in the tube had cooled down to 1.5°C. All dissections and experiments were performed in buffered (pH 7.4) crayfish saline (in mM: 202 NaCl, 5.37 KCl, 13.53 CaCl2, 2.6 MgCl2, and 2.4 HEPES). Animals were firmly pinned ventral side up in a Petri dish lined with silicone elastomer (Sylgard), and the entire preparation was bathed in fresh, noncirculating saline. The temperature in the dish stayed constant at 18–19°C throughout the experiments.
The exoskeleton directly above the ventral nerve cord was removed so that all abdominal segments and primary tail afferents were exposed (Fig. 1A). Nerves 1–3 were cut in all ganglia except the last abdominal ganglion (A6). The ventral nerve cord was flipped over and pinned down by the nerves so that the dorsal side containing the LG neurons was facing up. We paid attention to keep the ventral artery intact, although some damage may have occurred in some preparations when the cord was flipped over. To expose the caudal area of the brain and provide access to the initial segments of the brain connectives, the first, second, and third maxillipeds, a small part of exoskeleton located between the remaining mouthparts and the nephropores, as well as both antennal glands, were removed (Fig. 1A). For stimulation of rostral sensory systems, a small window was cut into the fifth basal segment of the right antenna (posterior to the first segment of the flagellum) to expose the antenna II nerve or into the caudal part of the right eyestalk and the underlying sheath to expose the protocerebral tract. Status of the animals was consistently monitored by observing spontaneous neural activity in the sensory nerves, brain connectives and the ventral nerve cord. Experiments were only conducted in preparations that appeared healthy and lively, which most remained for several hours after the dissection.
The basic organization of electrode placement was as follows (Fig. 1B): two silver wire hook electrodes were placed on the ventral nerve cord (VNC), one rostral and one caudal to the LG recording site to monitor evoked neural activity in the nerve cord. Silver wire hook electrodes were also placed on the brain connectives (BC) for recordings of interneuron activity. For stimulation of sensory afferents and interneurons, hook electrodes were placed on the peripheral nerves of A6 to stimulate primary tail afferents and on the antenna II nerve in the fifth basal segment of the antenna or on the protocerebral tract just posterior of the lateral protocerebrum. The antenna II nerves contain axons of primary afferents from mechanoreceptors and chemoreceptors located on the flagellum, and axons of proprioreceptors in its basal segments, all of which project to the antenna II neuropils (Sandeman et al. 1992). The protocerebral tract contains axons of primary visual afferents, second order visual interneurons, multimodal sensory interneurons, all of which descend to the median protocerebrum; in addition, the protocerebral tract also contains the axons of ascending visual interneurons from the contralateral eye, and—within the olfactory globular tract—small diameter axons of ascending olfactory interneurons which originate in the accessory and olfactory lobes (Mellon 2000; Sullivan and Beltz 2001).
Intracellular microelectrodes for recording and current injection had resistances of 15–35 MΩ and were filled with 3 M potassium chloride (KCl) for most experiments. Because inhibitory postsynaptic potentials in LG are mediated by GABA through an increase in chloride conductance (Roberts 1968; Vu et al. 1993), leak of chloride into the impaled neuron could potentially change an inhibitory postsynaptic potential (IPSP) into an excitatory PSP (EPSP). Because this may cause a problem for unambiguous identification of LG EPSPs in small animals like the ones we used in our study, we also used microelectrodes filled with 2 M potassium acetate (KAc; 15–40 MΩ) to conduct several control experiments.
LG neurons were impaled with one or two microelectrodes in close proximity and anterior to the fourth (A4) or sixth (A6) abdominal ganglion. MGs and other interneurons of interest were impaled with electrodes in the brain connectives close to the caudal part of the brain. Giant neurons were identified by their response to rostral or caudal sensory nerve stimulations; unambiguous identification was also warranted by measurements of size and conduction velocity of giant neuron action potentials evoked with suprathreshold current injections.
We used microelectrode amplifiers (Axoclamp900A; Molecular Devices) for current clamp experiments, a Grass stimulator (Model S88) and an A-M Systems differential amplifier (Model 1700) for stimulation and recording through hook electrodes. Analog data were digitized using a Digidata 1440A (Molecular Devices) and stored and analyzed using pClamp10.0 and Clampfit 10.0 software (Molecular Devices).
LG receives inputs from rostrally located sensory systems
To measure evoked neuronal activity and directionality of action potential propagation, we placed hook electrodes on the exposed antenna II nerve or protocerebral tract for stimulation, on the ipsilateral BC, and on two separate locations on the ventral nerve cord either anterior (VNC_rostral) or posterior (VNC_caudal) to the LG intracellular recording site. One intracellular electrode was inserted into the MG neuron near its initial segment close to the caudal part of the brain, and a second intracellular electrode was inserted into the LG ipsilateral to the respective stimulation site (i.e., antenna II nerve or protocerebral tract). Recordings from the LG in abdominal segments A4 (n = 22) and A6 (n = 8) of different animals were obtained with similar results using KCl- or KAc-filled electrodes.
Single-pulse electrical stimulation of the right antenna II nerve caused depolarizations in both the ipsilateral MG and the ipsilateral LG (Fig. 2). Electrical stimulation of this nerve also produced action potentials in interneurons, which are recorded in the right BC and in both locations of the VNC. With increasing stimulus intensity (Fig. 2A; gray traces: 3 V, black traces: 3.5 V), an increasing number of interneurons was recruited and corresponding changes of the MG PSP and the LG PSP were seen (Fig. 2A).
Stimulating the antenna II nerve at low intensity caused smaller depolarizations in MG and LG and the activation of a large single descending unit recorded in the BC and VNC (Fig. 2B). By comparing the positions of the recorded action potential in the BC and at different locations of the VNC, it became apparent that the action potential originated in the brain region and descended from head to tail. Weak stimulation (1.2 V, 0.3 ms) produced no measurable activity in the BC or VNC and produced no visible depolarization of LG (gray traces). A slightly stronger stimulus (1.4 V, 0.3 ms) resulted in the recruitment of an action potential that was seen in the BC with a delay of 9.3 ms after the stimulus onset and arrived 3 ms later at the rostral recording site of the VCN (between the 1st and 2nd abdominal ganglia) and 1.1 ms later in the caudal recording site of the VCN (between the 4th and 5th abdominal ganglia). The temporal pattern of spike propagation is consistent with the possibility that this was in fact a single descending unit (Fig. 2B; arrows), which caused the depolarization (0.7 mV) in the LG (black traces). Using electrodes filled with 2 M KAc (n = 4) instead of 3 M KCl produced very similar results (Fig. 2B1). A weaker stimulus (3.6 V) applied to antenna II nerve had no measurable effect (gray traces), whereas a slightly stronger stimulus (4.0 V) led to recruitment of a single descending unit that elicited a unitary PSP (0.85 mV) in LG (black traces).
Within each preparation, the temporal relationship between stimulus onset, recorded interneuron spike and corresponding LG PSP was highly consistent for each stimulus intensity. Five separate stimuli (2.1 V, 0.3 ms) successively applied to the antenna II nerve recruited the same interneuron and the corresponding LG PSP each time (Fig. 2C; 5 traces superimposed). The same result was observed when using KAc-filled microelectrodes (n = 2) to record LG PSPs (Fig. 2C1). Five repetitive stimuli (4.0 V, 0.3 ms) to the antenna II nerve recruited the same interneuron and led to corresponding PSPs in LG.
Conduction times between stimulus site and LG (∼13 ms) were similar for most experiments (Fig. 2, B and C), suggesting that the same interneuron may have been recruited in these cases. We only recorded inputs to the ipsilateral LG near A4 and A6; although we did not measure LG responses in A5 or in any of the ganglia that are located more anterior to A4, results are not expected to be different.
LG also received inputs from descending units activated by stimulation of the protocerebral tract (n = 4, Fig. 3). By varying the intensity of the stimulus, an increasing number of descending units was activated. A weak stimulus (4.6 V, 0.3 ms) produced one measurable action potential in the VNC and produced only a small depolarization of LG (0.5 mV; light gray traces in Fig. 3A). A second interneuron was recruited with stronger electrical stimulation (5 V, 0.3 ms) and was found propagating through the BC (not shown) and VNC, resulting in a larger depolarization of LG (1.1 mV; dark gray traces in Fig. 3A). Higher stimulus intensity (5.4 V, 0.3 ms) resulted in activation of more descending units, producing a larger, compound PSP in LG (1.7 mV; black traces in Fig. 3A). The experiment was repeated twice with KAc solution in the electrode and produced similar results (Fig. 3A1). Gradual increases in stimulus voltage (light gray traces: 2 V, dark gray traces: 3 V, black traces: 4 V) applied to the protocerebral tract led to the activation of additional descending interneurons and produced an increase in both the number and size of the LG PSPs.
Combined sensory input to LG can elicit an action potential
To verify the excitatory nature of the descending inputs to LG, two different experiments were performed. First, antennal sensory inputs were combined with sensory inputs from primary afferents of the tail. Descending inputs in these experiments could not alone cause action potentials in the LGs, but when paired with EPSPs elicited by stimulation of tail afferents, the inputs summed to yield firing (Fig. 4A). Hook electrodes were placed on the antenna II nerve and on sensory nerves 2–4 of the last abdominal ganglion that contain many of the primary afferents that excite the LGs (Antonsen et al. 2005; Herberholz et al. 2002). One intracellular electrode was used to record LG membrane potentials on the ipsilateral side of the sixth abdominal ganglion and neural activity was monitored with a hook electrode placed on the ipsilateral BC. Electric stimulation (5.6 V, 0.3 ms) applied to the tail afferents was adjusted so that the resulting EPSP in the LG (7 mV) was close to firing threshold (dark gray traces in Fig. 4A). A stimulus applied to the right antenna II nerve (5 V, 0.3 ms) alone resulted in a small depolarization of the LG (light gray traces in Fig. 4A). When the onsets of the rostral and caudal stimuli were timed so that their resulting inputs to LG overlapped, the combined input from both sources was sufficient to bring the LG above threshold (black traces in Fig. 4A; note that the LG action potential has been truncated). In these experiments, the antenna II nerve was stimulated several milliseconds before the sensory nerves of the tail to allow enough time for the descending units to reach the sixth abdominal ganglion and to coincide with the peak of the EPSP received from caudal inputs. Control experiments with KAc-filled electrodes (n = 2) produced the same results (Fig. 4A1). A stimulus to the right antenna II nerve (5.0 V, 0.3 ms) caused a small depolarization in LG; when combined with subthreshold stimulation of tail afferents (1.5 V, 0.3 ms), the inputs summed and fired the LG neuron.
In a second set of experiments, subthreshold current (179 nA, 5 ms) was injected into the LG via a second KCl-filled intracellular electrode positioned in the same interganglion segment and in close proximity to the recording electrode (gray traces in Fig. 4B). The right antenna II nerve was stimulated with hook electrodes (5 V, 0.5 ms), causing the resulting depolarization in the LG to overlap with the depolarization produced by current injection. The combined depolarization was sufficient to produce an action potential in the LG, which could be recorded with the second intracellular electrode and an extracellular recording electrode placed on the BC (black traces in Fig. 4B). When suprathreshold current injection was used to elicit an action potential in the LG and this current injection was combined with stimulation of the ipsilateral antenna II nerve, the threshold for LG activation was always reached earlier than with current injection alone (data not shown).
Identifiable descending neurons produce EPSPs in LG
The BC was impaled near the brain with an intracellular electrode to identify individual descending interneurons that were activated by electrical stimulation of the antenna II nerve and produced EPSPs in the LG (Fig. 5). Several neurons that matched both criteria were found in different preparations (n = 14). Electrical stimulation of the antenna II nerve (3 V, 0.5 ms) caused firing of many of the descending interneurons, some of which produced EPSPs in LG. In one such example, an action potential was recorded in one of these interneurons in the ipsilateral brain connective 7 ms after the stimulation (Fig. 5A).
Prolonged current injection (50 nA, 20 ms) into the same interneuron caused it to fire multiple action potentials (Fig. 5B). The spikes were observed in the extracellular electrodes placed on the VNC, and they evoked corresponding unitary EPSPs in the ipsilateral LG (Fig. 5B). Injections of lower currents or for shorter time periods often caused single action potentials in individual interneurons and corresponding EPSPs in LG. The temporal relationship between interneuron spikes and LG EPSPs was very consistent for each individual interneuron but varied across preparations and according to LG recording site. The temporal delay measured between the peak amplitudes of interneuron spikes elicited by current injection and corresponding LG EPSPs was 4.8 ± 0.5 (SD) ms for recordings obtained in A4 (n = 7) and 6.6 ± 0.5 ms for recordings obtained in A6 (n = 5). Similar results were obtained from experiments using microelectrodes filled with KAc (Fig. 5B1; n = 2). Current injection (60 nA, 20 ms) into a descending mechanosensory interneuron caused it to fire several action potentials which produced corresponding EPSPs in the ipsilateral LG. When descending inputs from a single interneuron were paired with tail afferent inputs, the LG EPSP increased in size (Fig. 5C). Activation of the BC interneuron with current injection (60 nA, 25 ms) produced a series of small (0.5 mV) PSPs in LG (light gray trace). Stimulation of the A6 sensory nerve alone (0.7 V, 0.3 ms; dark gray trace) also caused a PSP in LG (1.5 mV). When rostral and caudal inputs coincided, they summed to produce a larger EPSP in LG (2 mV; black trace).
Escape behavior in crayfish
Fast and powerful escape tail-flips in crayfish are mediated by neural circuits that consist of pairs of giant interneurons, giant motor neurons, and electrical coupling between these components. The lateral giant (LG) interneurons are the key components of the LG circuit, and it has previously been assumed that they can only be excited by stimuli directed to the rear of the crayfish's body. Using largely intact preparations of juvenile crayfish, we have now shown for the first time that the LG interneurons have more extensive receptive fields than previously thought, and they receive excitatory inputs following rostral sensory stimulation. These descending inputs lower the threshold for LG-mediated escape tail-flips and are therefore likely to enhance the animal's ability to respond adaptively. This is because any attack that is ultimately directed to rear of the animal will be preceded by multimodal sensory stimulation (e.g., visual and hydrodynamic) before the actual physical contact occurs, and if this information is detected by widely distributed sensory systems and integrated with sensory stimulation to the rear, it will enhance the animal's chance of surviving. Moreover, during exploration of their environment (e.g., during foraging), any “suspicious” signal detected by antennal contact or by the eyes could lead to an increase of the LGs' excitability and thus enhance the escape system's sensitivity for a possible attack. Such a mechanism has been shown in cockroaches where antennal palpation provides cues for predator detection, and such textural information (rather than chemical information) is used to modify the escape behavior (Comer et al. 1994, 2003).
Organization of escape giant neurons' receptive fields
After electrical stimulation of rostral sensory nerves, action potentials were seen in the BCs within 10 ms after stimulation, and they arrived in the abdomen only a few milliseconds later. Thus information transfer is fast, and spikes that originate rostrally arrive in the abdomen fast enough to affect caudal events that happen shortly after. The relatively short time it takes for action potentials to descend from rostral centers to the abdomen and to elicit EPSPs in the abdominal LGs, implies that the information is transmitted via individual descending interneurons that project from the brain to the last abdominal segment and make monosynaptic connections with the LGs. The descending inputs from several activated interneurons can summate and produce long-lasting compound EPSPs in the LGs. Likewise, single interneurons that produce multiple spikes cause several milliseconds of depolarization in the LGs thus providing an “open time window” during which subsequent excitatory inputs from tail afferents can combine with this depolarization. Although we only recorded relatively small LG EPSPs in response to rostral stimuli, we cannot exclude the possibility that summation of rostral inputs after repetitive stimulation of multiple sensory channels are sufficient to fire the LGs, especially in freely behaving animals.
Sensory stimulation of the protocerebral tract or the antenna II nerve also elicit EPSPs in the MGs; because the MGs receive their inputs in the brain, interneuronal activity in the BCs or nerve cord cannot be directly linked to MG EPSPs. However, both MG EPSPs and LG EPSPs appeared with the first recruitment of descending units, and thus it is possible that the same interneurons innervate the MGs and LGs. Another interesting question is whether the MGs have receptive fields that extend to the abdomen and receive subthreshold excitatory inputs from this area. Preliminary data indicate that this may be the case (Herberholz and Liu 2008). More experiments toward this end are planned for the future.
Using broad electrical stimulation of the antenna II nerve and the protocerebral tract, we were unable to unambiguously determine the identity of activated sensory modalities. This is because the antenna II nerve contains both mechano- and chemosensory afferents, and the protocerebral tract contains axons of many sensory modalities, including visual, mechanosensory, and chemosensory, some of which are descending and some of which are ascending (Sandeman et al. 1992; Sullivan and Beltz 2001). Although most of the axons found in the antenna II nerve belong to mechanosensory afferents (Sandeman 1989), and most of the large descending axons in the protocerebral tract (also referred to as the “optic tract” in the older literature) belong to visual interneurons (Cooper et al. 2001; Wiersma and Yamaguchi 1966, 1967), future experiments will have to use natural visual and natural tactile stimuli for identification of specific sensory modalities that are responsible for the increase of the LGs' excitability in the abdomen.
Descending excitation versus depolarizing inhibition
Previously described descending inputs to the LGs that originate in the thoracic ganglia and brain were considered to be purely inhibitory; they provide “tonic” inhibition that modulates the LG's threshold during specific behaviors such as feeding, restraint and possibly also during social interactions (Krasne and Edwards 2002b; Krasne and Lee 1988; Krasne and Wine 1975). Although descending inhibitory neurons have not been identified, tonic inhibition is thought to result from GABA-mediated changes of chloride channel activity at inhibitory synapses that are located distally on the LG's dendrites (Vu and Krasne 1992, 1993). Tonic depolarizing inhibition is characterized by a persistent depolarization of LG without rapid fluctuations in membrane potential indicating very slow decay rates for IPSPs (Vu et al. 1993). Moreover, “recurrent” GABAergic inhibition of the LG neurons, which happens proximal near the spike initiation zone, causes large depolarizing IPSPs that are much longer than the brief depolarizing events seen in our study (Roberts 1968; Vu and Krasne 1992). Thus long and persistent depolarizing IPSPs are in sharp contrast to the short depolarizations seen in our experiments. The excitatory nature of these brief events is further evidenced by the lower input needed from the tail afferents to fire the LGs when caudal inputs coincide with rostral inputs. We performed these experiments with both KCl- and KAc-filled electrodes and obtained identical results; this strongly suggests that the LGs receive excitatory inputs from rostral sensory regions. How these excitatory inputs are integrated with inhibitory inputs to orchestrate the most adaptive escape responses remains to be determined.
Comparison to other escape systems
Fast escape responses of decapods are typically controlled by neural circuits that contain LG and MG giant interneurons (Espinoza et al. 2006). Given the many similarities in escape behavior and underlying neural circuitry among these closely related species, it is likely that our findings generalize to other decapods with giant escape circuits. Escape responses have also been investigated in a number of noncrustacean model systems. Extensive work has been done in insects where escape behavior and underlying neural circuitry parallels the crayfish escape system to a certain extent. For example, cockroaches and crickets escape from wind stimuli by turning, running, and/or jumping. Wind stimuli cause deflection of hairs located on the cerci, and sensory afferents activate multiple giant interneurons that ascend from the abdomen to the thoracic region to initiate the escape maneuvers (Gras and Hörner 1992; Kanou et al. 1999; Ritzmann 1984; Ritzmann and Pollack 1994; Tauber and Camhi 1995). Interestingly, directional escape behavior in cockroaches can also be initiated by antennal stimulation. Such stimulation leads to the activation of descending mechanosensory interneurons that project to the thoracic ganglia and trigger or modulate escape turns and running (Burdohan and Comer 1990; Comer et al. 2003; Schaefer and Ritzmann 2001; Ye and Comer 1996).
Although speculative at this point, rostrally activated sensory interneurons of cockroaches and crickets (and possibly other insects) could project all the way to the abdomen and directly affect the excitability of the giant interneurons. The “decision neurons” for escape in insects are situated in the thoracic ganglia and not in the abdomen; however, this does not exclude the possibility that ascending inputs to the decision neurons are affected by preceding rostral stimulation. If true, this could modify the responses to stimuli directed toward the rear and thus provide an additional level of behavioral plasticity during escape.
This work was in part supported by Research Grant IOS-0919845 from the National Science Foundation. Y.-C. Liu received support from the National Taiwan University and was additionally supported by a stipend from the Psychology Department at the University of Maryland, College Park, MD.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Dr. David Yager and W. H. Liden for helpful discussions and V. Medley for help with some of the experiments.
Present address of Y.-C. Liu: Dept. of Organismal Biology and Anatomy, University of Chicago, Chicago, IL 60637.
- Copyright © 2010 the American Physiological Society