Mechanisms of Venom Injection and Behavioral Modulation of a Cockroach Prey by a Parasitoid Wasp

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY”

Ram Gal

Submitted to the Senate of Ben-Gurion University of the Negev

March 2010

Beer-Sheva

Mechanisms of Venom Injection and Behavioral Modulation of a Cockroach Prey by a Parasitoid Wasp

Thesis submitted in partial fulfillment of the requirements for the degree of “DOCTOR OF PHILOSOPHY”

Ram Gal

Submitted to the Senate of Ben-Gurion University of the Negev

Approved by the advisor ______Approved by the Dean of the Kreitman School of Advanced Graduate Studies ______

March 2010

Beer-Sheva

This work was carried out under the supervision of Prof. Frederic Libersat

In the Department of Life Sciences

Faculty of Natural Sciences

Acknowledgments

Throughout my seven years at the Libersat lab, more than a few people have helped and supported me personally and scientifically. I would like to thank a handful of them.

First and foremost, I would like to express my utmost gratitude to my advisor, Prof. Frederic Libersat, for his invaluable help and support throughout my studies. I consider myself privileged to have been given the opportunity to work and learn from Fred for so long. His high standards and ambition, as well as his vast knowledge and sincere passion for science, have inspired me throughout the years and have shaped my perception of science for years to come.

I would also like to thank Jose Gustavo Glusman for his technical help, valuable advice and companionship throughout the years; Gal Haspel, who was my mentor as a young scientist, for teaching me the very essentials of the scientific rationale and academic writing; the Zlotowski Center for Neuroscience for providing help, financial support and a fruitful scientific environment; and the Journal of Experimental Biology for granting me a travel fellowship for a two-month research project in The Netherlands.

A special thanks to former colleagues and students in the lab, to my friends (and especially Yael Lavi, Idan Harpaz and Rotem p. Uzan) and to my brother Idan and my sister Bar, for brainstorming, advice and help throughout the years.

Last, but not least, I want to thank my loving parents, without whose support I would have never made it through. This work is dedicated to them.

To my parents, Gidi and Orit Gal, who made it possible. Table of Contents

List of abbreviations...... i List of figures ...... ii Abstract ...... 1 1. General introduction 1.1. Background ...... 4 1.2. From stimulation to action: a behavioral model...... 11 1.3. Prior mechanistic investigations of the venom-induced hypokinesia...... 15 1.4. Research goals...... 16 2. Published work 2.1. Synopsis of the published work...... 17 2.2. New vistas on the initiation and maintenance of insect motor behaviors revealed by specific lesions of the head ganglia 2.3. A parasitoid wasp manipulates the drive for walking of its cockroach prey 2.4. A wasp manipulates neuronal activity in the sub-esophageal ganglion to decrease the drive for walking in its cockroach prey 2.5. Parasitoid wasp uses a venom cocktail injected into the brain to manipulate the behavior and metabolism of its cockroach prey 3. Unpublished work: Sensory mechanisms mediating host CNS localization 3.1. Background ...... 21 3.2. Materials and Methods ...... 22 3.3. Results and Discussion 3.3.1. Behavioral experiments...... 30 3.3.2. Electron Microscopy ...... 32 3.3.3. Electrophysiology...... 35 4. General discussion 4.1. Parasite-induced manipulation of host behavior in insects...... 38 4.2. Hunting strategies of parasitoid wasps...... 42 4.3. Overview and general discussion of the published work 4.3.1. The descending influence of cerebral ganglia on cockroach motor behaviors...... 45 4.3.2. The venom depresses the cockroach’s drive for walking...... 47 4.3.3. Involvement of the SEG in venom-induced hypokinesia...... 49 4.4. How the Jewel Wasp ‘hijacks the free will’ of cockroaches: Current mechanistic hypothesis...... 50 6. References...... 57 List of abbreviations

CBC: central body complex CNS: central nervous system CPG: central pattern generator DA: dopamine Df: fast coxal depressor Ds: slow coxal depressor DUM: dorsal unpaired medial EMG: electromyogram GI: giant interneuron OA: octopamine SEG: sub-esophageal ganglion SEM: scanning electron microscope SupEG: supra-esophageal ganglion TAG: terminal abdominal ganglion TI: thoracic interneuron TEM: transmission electron microscope VNC: ventral nerve cord

~ i ~

List of figures

Figure 1: The stinging behavior and life cycle of the Jewel Wasp, Ampulex compressa …………………………………………………………………………. 6 Figure 2: Ampulex compressa stings directly into the cerebral ganglia of its cockroach prey ……………………………………………………………………. 10 Diagram 1: A simplified model for the homeostatic control of behavior ……….. 12 Figure 3: The basic escape circuitry of the cockroach and hypothesized sites of action of the venom ………………………………………………………….... 14 Figure 4: Surgical procedures used throughout my research …………………….. 18 Figure 5: Wasp recording set-up ………………………………………………..... 29 Figure 6: Stinging duration after cockroach CNS lesions ……………………...... 31 Figure 7: SEM analysis of the wasp’s stinger ……………………..……………… 33 Figure 8: TEM analysis of the wasp’s stinger …………………………………….. 34 Figure 9: Stinger responses to mechanical stimulation …………………………… 36 Figure 10: Stinger responses to chemical stimulation ……………………………. 37 Figure 11: Examples of fatal interactions between parasites and their insect hosts ..……………………………………………………………………………... 41 Figure 12: Current model of the neurophysiological events leading to venom- induced hypokinesia in cockroaches stung by Ampulex compressa .……………… 55

~ ii ~

Abstract

The parasitoid Jewel Wasp Ampulex compressa hunts cockroaches as live food supply for its offspring. In doing so, the wasp stings a cockroach in the head and injects a venom cocktail directly and precisely into the cerebral ganglia, which are considered ‘higher-order’ neuronal centers in insects. The sting induces a unique behavioral manipulation of the cockroach prey. Although not paralyzed, the stung cockroach becomes hypokinetic and fails to initiate spontaneous or evoked locomotion for several days. The wasp grabs one of the cockroach’s antennae and steers the cockroach, walking backwards with the cockroach following in a docile manner like a dog on a leash, into a pre-selected nest or burrow. Intoxicated by the wasp’s venom, the stung cockroach does not put up a fight nor try to escape when the wasp lays an egg on its cuticle, seals the nest with debris and leaves. A larva hatches from the egg two days later and feeds on the live cockroach, protected in its nest from potential predators, for four more days. Then, when ready to pupate, the larva weaves a cocoon inside the abdomen of its cockroach host - which is at this stage not more than an empty shell - and undergoes metamorphosis. It hatches a month later as an adult ready to continue its life cycle. The purpose of my study was to characterize the mechanisms which allow the Jewel Wasp to induce this precise behavioral manipulation and ‘hijack the free will’ of its cockroach prey, rendering the cockroach a docile ‘automaton’ unwilling to escape its fate.

As a first step, and since the sting is directed at the two cerebral ganglia of the cockroach prey, I explored how each of these ganglia individually regulate locomotion in non-stung cockroaches. To this end, I developed and validated a methodology for selectively removing descending inputs from each cerebral ganglion to thoracic motor centers in behaving cockroaches. By combining behavioral and electrophysiological tools, I constructed a functional model to describe how different stimuli evoke different motor patterns in cockroaches, and how these are separately regulated by each cerebral ganglion. I found that walking-related behaviors are differentially regulated by the two cerebral ganglia, with the sub-esophageal ganglion (SEG) up-regulating and the supra-esophageal ganglion (SupEG, or ‘brain’) down- regulating locomotion.

~ 1 ~

Next, I performed a detailed examination of stung cockroaches to study which and how motor behaviors are altered as a result of the sting. I employed behavioral paradigms originally developed to study motivational disorders in mammals, combined with electrophysiological recordings from leg muscles during ongoing motor tasks. The behavioral and electrophysiological data obtained in these experiments demonstrate that the venom selectively depresses the cockroach’s motivation or ‘drive’ to initiate and maintain walking-related behaviors, rather than inducing an overall decrease in arousal or a ‘sleep-like’ state. As such, I found that the threshold for the initiation of walking related behaviors, specifically, is increased in stung cockroaches. Further, I showed that stung cockroaches cannot maintain such behaviors even after these are evoked by supra-threshold stimuli, although the central pattern generators responsible for the spatio-temporal pattern of locomotion are intact. These deficits can be explained by a decrease in descending excitatory inputs from the cerebral ganglia to thoracic motor centers, which was indeed noted using electromyographic recordings from leg muscles.

Thirdly, since the venom decreases the propensity of occurrence of walking related behaviors specifically, and since the SEG was found to up-regulate such behaviors, I hypothesized that the venom decreases neuronal activity in (at least) the SEG of stung cockroaches to induce hypokinesia. To test this hypothesis, I used behavioral, electrophysiological and neuropharmacological tools. My results show that (1) the wasp actively and specifically searches for the SEG inside the cockroach’s head capsule when inflicting the head sting; (2) selective inhibition of neuronal activity in the SEG, either by venom or by a local anesthetic, decreased spontaneous and evoked locomotion in non-stung cockroaches similar to a natural wasp sting; (3) spiking spontaneous and stimulus-evoked neuronal activity in the SEG, as recorded with an extracellular bipolar microelectrode, is decreased in stung compared with non-stung control cockroaches. My data thus confirm a long-standing hypothesis suggesting that the primary neuronal target for the venom-induced hypokinesia is the SEG, and highlight the role of this little-studied ganglion in the up-regulation of locomotion in insects.

~ 2 ~

Finally, I have investigated the sensory mechanisms which allow the wasp to precisely locate the two cerebral ganglia inside the head capsule of its cockroach prey during the head-sting. By combining behavioral studies, electrophysiological experiments and high-resolution electron microscopy, I showed that the wasp’s stinger possesses specialized cuticular sensory organs, most probably of a mechano- chemosensory nature (i.e., gustatory sensilla), which are capable of discriminating nervous from non-nervous tissue inside the cockroach’s head cavity. Although sensory organs are commonly found on the stingers of parasitoid wasps, such ‘brain- sensors’ capable of identifying neuronal tissue appear to be unique to the Ampulex stinger, in which they are crucial for successful parasitization and behavioral manipulation of the prey.

In this thesis, I will describe the rationale, methodology and results obtained during my six-year research of the intricate wasp-cockroach interaction. Based on these results and on previous studies, I will also suggest a mechanistic model to describe how venom injected in the cerebral ganglia modulates the initiation and maintenance of walking-related behaviors, thus depriving stung cockroaches of their ‘free will’.

Keywords: Wasp; Cockroach; Parasitoid; Neuroethology; Motor Behavior; Venom; Motivation; Sub-esophageal Ganglion; Sensillae.

~ 3 ~

1. General introduction

1.1. Background Animals constantly interact with their environment, responding to external and internal stimuli so as to increase their fitness. Such responses are realized using sensory systems to perceive different aspects of the environment, and motor systems to produce appropriate behavioral outputs. While sensory inputs and motor outputs are carried by the peripheral nervous system – sensory afferents and motor neurons, respectively – the processing and organization of the motor response is performed by the Central Nervous System (CNS). The CNS integrates sensory inputs from various sources and controls motor neurons which, in turn, produce the muscle activity required for the execution of movement. Since the majority of inter-cellular communication in the nervous system is mediated through chemical synapses, the system can be profoundly modulated by endogenous and exogenous neuromodulators. Such disposition allows for remarkable plasticity in the nervous system, but also presents vulnerability since chemical synapses are convenient targets for toxins. Indeed, in the perpetual struggle between predator and prey, some organisms have acquired elaborate chemical tools to manipulate the nervous system of their prey or predator. In many cases in the animal kingdom, neurotoxins are used to interfere with normal ongoing activity of the nervous system. A common target for neurotoxins is the peripheral nervous system, and in particular the neuromuscular junction, where neurotoxins act to induce transient or long-lasting tonic or flaccid muscle paralysis in the prey or predator [Adams and Swanson 1996, Olivera 1999, Rapuoli and Montecucco 1997]. However, through millions of years of co-evolution, some venomous predators have developed neurotoxins that target the central, rather than the peripheral nervous system of their prey. In some unique instances, these toxins are used to dramatically manipulate the behavior of the prey, forcing it to modify its normal behavior for the benefit of the predator or its offspring [Libersat et al. 2009]. My study has focused on one such instance, where an Ampulicine parasitoid wasp uses a neurotoxic venom cocktail to ‘hijack the free will’ of its cockroach prey for the benefit of the wasp’s offspring.

~ 4 ~

The use of neurotoxic venoms to incapacitate the prey is common in parasitoid wasps. These Hymenopternas are solitary wasps of the families Ampulicidae, Sphecidae, Pompilidae, Mutillidae or Bethylidae [Rathmayer 1978], in which the adult is free- living and feeds on nectar, while the larva is parasitic on insects or spiders. Parasitism in these species begins when an adult female wasp locates a suitable host and then paralyzes it with a venomous sting. The venom usually contains neurotoxins which interfere with the initiation or propagation of action potentials in the nervous system or with synaptic transmission, most often in the neuromuscular junction. The wasp then carries the paralyzed host to a nest or burrow, where it lays an egg on the prey and leaves. When the wasp’s larva hatches, it develops by feeding on the immobile host until it is ready to pupate [for reviews, see Piek and Spanjer 1986, O’Neill 2001, Libersat and Gal 2007].

In the specific case of the Jewel Wasp, Ampulex compressa (also known as the Emerald Cockroach Wasp), a large tropical Ampulicine parasitoid, the adult female hunts cockroaches (Periplaneta americana) for use as a live food supply for its offspring. Unlike most parasitoid wasps, and since the development of its larva requires feeding on a live cockroach for several days [Haspel et al. 2005, Libersat and Gal 2007], the adult Jewel Wasp does not paralyze its cockroach prey like most other parasitoid wasps. Instead, it performs a delicate chemical ‘brain surgery’ to manipulate the cockroach behavior in a most precise manner, which have intrigued naturalists centuries ago [de Reaumur 1742, Bingham 1897, Maxwell-Lefroy 1909, Williams 1942]. With a venomous sting into the cockroach head, the wasp renders its prey an obedient ‘zombie’ which follows the wasp into a nest and waits submissively as the larva feeds on its internal organs for almost a week (Figure 1).

~ 5 ~

A B C

D E F

G H I

Figure 1: The stinging behavior and life cycle of the Jewel Wasp, Ampulex compressa. The wasp (A) stings a cockroach twice, first in the thorax (B) and then in the head (C), to evoke three consecutive behavioral changes. Initially, as a result of the thoracic sting, the cockroach’s front legs are transiently paralyzed, as can be seen in the change of body posture (D). As paralysis weakens, the cockroach begins grooming excessively (E) and then enters a long lasting hypokinetic state. The wasp grabs one of the cockroach’s antennae and leads the cockroach to a pre-selected burrow (F), with the cockroach following in a docile manner, like a dog on a leash. Inside the burrow, the wasp lays an egg on the hypokinetic cockroach (G, arrow). The egg later develops into a larva (H), which feeds on the docile cockroach for several days. The larva then pupates inside the cockroach’s abdomen and hatches a month later (I) to leave the burrow and continue the life cycle.

~ 6 ~

To induce this behavioral modulation, the Jewel Wasp stings the cockroach twice, first in the thorax and then in the head [Fouad et al. 1994]. Grabbing the cockroach by the pronotum, the wasp directs its stinger into the cockroach’s thorax to inject venom in and around the first thoracic ganglion [Haspel 2003, PhD thesis]. The venom interferes with cholinergic and GABAergic transmission in the first thoracic ganglion to prevent the recruitment of the motor neurons innervating legs muscles, thereby indirectly inducing a flaccid paralysis of the front legs which lasts several minutes [Piek et al. 1984, 1989, Haspel and Libersat 2003, Moore et al. 2006]. With its front legs paralyzed, the cockroach cannot fight off the wasp, which pulls out its stinger and directs it at the cockroach’s head, penetrating the head capsule through the soft neck cuticle. During this head sting, the wasp injects a venom cocktail directly inside the cockroach's cerebral ganglia [Haspel et al. 2003], namely the supra-esophageal ganglion (SupEG, or ‘brain’) and the sub-esophageal ganglion (SEG) (Figure 2), which comprise the ‘higher-order’ neuronal centers of the insect nervous system [Kien and Altman 1992a, Strauss and Heisenberg 1993, Schaefer and Ritzmann 2001, Strauss 2002, Ridgel and Ritzmann 2005, Gal and Libersat 2006, Wessnitzer and Webb 2006]. As a result of this head sting, and soon after recovering from paralysis, the cockroach begins grooming excessively for 20-30 minutes non-stop, during which time the wasp leaves the cockroach unattended to search for a suitable nest or burrow nearby. The stereotypic though excessive grooming behavior is presumably evoked by dopamine (DA) and dopaminergic agonists in the venom, agents that have been shown to induce grooming when injected into the head of non-stung cockroaches [Wiesel-Eichler et al. 1999]. The precise adaptive value for the wasp (if any) in inducing excessive grooming in its prey is yet unclear. For instance, it might serve to remove ectoparasites and fungi, potentially hazardous for the developing larva, from the cockroach’s cuticle; alternatively, grooming might serve to keep the cockroach occupied at the stinging site while the wasp searches for a nest to store the stung cockroach in. Upon locating an oviposition site, the wasp returns to the stung cockroach, which gradually ceases grooming and enters the next phase of envenomation, the hypokinetic state. In this state, although it shows no signs of muscular paralysis, the stung cockroach behaves like a submissive automaton, acting as if the wasp’s sting had ‘hijacked its free will’ [Fouad et al. 1994]. Under natural conditions, this state lasts for several days [Haspel et al. 2005], during which time the cockroach is unable to self-initiate locomotion and is unresponsive to aversive stimuli,

~ 7 ~ although it still remains ambulatory and can even walk for short distances if stimulated strongly enough [O’neill 2001]. The wasp then cuts both the cockroach’s antennae and drinks some hemolymph from the cut end, presumably to induce oviposition [Keasar et al. 2006]. Under the influence of the wasp’s venom, the cockroach does not try to fight or resist the host-feeding process and remains docile when the wasp then grabs one of the cut antennal stumps and, walking backwards, steers it to the pre-selected nest for oviposition. It is at this stage that the adaptive value of the non-paralytic hypokinesia is probably best illustrated. The stung cockroach, typically several-fold heavier than the wasp, expresses a normal walking pattern as it follows its predator’s lead into the nest, like a dog on a leash (movie available online [Gal and Libersat 2008]: http://www.cell.com/current- biology/supplemental/S0960-9822(08)00611-8). This significantly simplifies host- transportation for the wasp, as the preferred nesting site is often elevated from the ground (e.g., holes in trees or elevated burrows between rocks) and requires moving through uneven terrain. Upon reaching the nest, the wasp navigates its prey through the opening and lays one egg on its leg. As the stung cockroach waits passively in its tomb, the wasp seals the nest with debris and then leaves the cockroach in the nest, protected from other potential predators. The larva hatches from the egg two days later, perforates a hole in the cockroach’s cuticle and feeds on its hemolymph for another three days. Then, it penetrates through the cockroach’s exoskeleton and feeds on internal organs for two more days. The docile cockroach remains alive, although it does not feed or drink, throughout the long larval developmental process. This is because in addition to hypokinesia, the venom also induces a decrease in the cockroach’s metabolic rate to preserve water and sustain nutrients for the developing larva [Haspel et al. 2005]. After feeding for almost a week, after which the cockroach finally meets its inevitable fate and dies, the larva is ready to pupate. It weaves a cocoon inside the abdomen of the dead cockroach (which is, at this stage, nothing more than an empty shell), and undergoes metamorphosis. An adult wasp emerges from the pupa a month later, leaves the nest and flies off, ready to continue the parasitoid life cycle. The precise time course of the larval development can be found in [Haspel et al. 2005], and a link to a short online movie describing this process can be found in the Supplementary Material section of [Gal and Libersat 2008].

~ 8 ~

The first two phases of cockroach envenomation, i.e., transient front-leg paralysis and excessive grooming, have been thoroughly investigated in the Libersat lab for more than a decade [for recent review, see Libersat et al. 2009]. Thus, I focused my research on the third and more elusive phase of envenomation, namely the long-term hypokinetic state. This change in cockroach responsiveness is not mechanistically related to the larval developmental process. In fact, if the wasp is experimentally separated from the cockroach immediately after the head sting, the cockroach recovers from the hypokinetic state after several days to resume apparently normal activity [Fouad et al. 1994]. Hence, the behavioral manipulation must be induced by the wasp’s venom injected inside the cockroach’s cerebral ganglia. Since these ganglia house neuromodulatory circuits that regulate the expression of motor behaviors, the venom-induced manipulation enables a unique glance into the mechanisms which mediate self-driven behaviors in cockroaches. To understand such behavioral mechanisms, one must first consider the peripheral and central mechanisms which mediate between stimulation and action in animals.

~ 9 ~

A B

Figure 2: Ampulex compressa stings directly into the cerebral ganglia of its cockroach prey. A. The wasp stings a cockroach in the head. B. A micrograph of the stinger (st) is shown over a schematic lateral view of a cockroach head drawn to scale (cerebral nervous system in yellow). The wasp’s stinger penetrates through the ventral head cuticle to reach the cockroach brain (br) and sub-esophageal ganglion (SEG, s). Scale bar: 0.5 mm. Figure modified from [Gal et al. 2005]. C-E: Work of a former PhD student, Haspel et al. [2003], demonstrating the location of the sting within the cockroach head. Wasps were injected with radiolabeled amino acids that were incorporated in the venom to allow tracing in the cockroach head after a sting. C. In stung cockroaches (black bars), the levels of radiolabeled venom are significantly higher in the brain and SEG than in non-neuronal head tissue. In contrast, when radioactive amino acids are injected manually into the head cavity of control cockroaches as control (open bars), they do not cross the protective sheath around the cerebral ganglia. Hence, radioactivity levels in this case are significantly higher in non-neuronal head tissue than inside the cerebral ganglia. D, E: Representative sections of the brain (D) and SEG (E) of a cockroach stung by a radiolabeled wasp. Venom is located in the brain (posterior to the Central Body Complex and around the mushroom bodies) and around the middle of the SEG. Scale bar = 0.25 mm.

~ 10 ~

1.2. From stimulation to action: a behavioral model Animals are not automatons which react in the same manner every time they encounter the same stimulus. Rather, an animal must choose a proper behavioral response from a wide behavioral repertoire, based on its current state in space and time. A given stimulus may thus produce entirely different behavioral responses when the animal is in different behavioral states, e.g., hungry, in search of a mate, resting, escaping from an oncoming predator, etc. A crucial element in the life of every animal is, therefore, the ability to make correct behavioral choices and respond to a given stimulus by selecting the optimal behavioral output. Such behavioral choices are a function of the CNS, which integrates sensory inputs and activates motor neurons to produce muscle activity. Hence, the behavioral state of the animal at a given moment, and thus the predicted response to a given stimulus, can be reflected in, and is largely determined by, processes occurring in the CNS.

Below, I describe a simplified homeostasis-based behavioral algorithm for the selection of appropriate behavioral outputs to a given general sensory input. Understanding the exact functions described in this model and their underlying neural mechanisms lay at the core of neuro-ethology. This was my main motivation in studying the wasp-cockroach interaction, where stung cockroaches fail to respond appropriately to naturally-occurring stimuli. For simplicity, key functions of the nervous system in the model are presented in italics, and the model is also schematically described in Diagram 3.

The model considers the following: (1) The state of an animal, with respect to its internal and external environment at a given moment, is defined by an array of intero- and exteroceptors, which represent the different features of the environment. A given change in the environment must, therefore, exceed a sensory threshold to be perceived as a stimulus. This threshold is determined by the physical properties of the relevant sensory organs and their accompanying sensory cells, which transduce a physical change in environmental conditions into a specific neuronal code. Afferent axons of the sensory cells convey this code to designated regions in the CNS, where it is integrated with an array of sensory inputs arriving from other sensory organs; (2) A supra-threshold external stimulus deflects the animal from homeostasis. In response, the animal must make a decision as to whether or not to initiate a motor response to

~ 11 ~ the stimulus. The decision is based largely upon the type and intensity of the stimulus, and on the current state of the animal in space and time (e.g. its motivation, current activities, etc.). A stimulus which surpasses the ‘behavioral threshold’ determined by the state of the animal (and thus not to be confused with the ‘sensory threshold’) optimally results in selecting the least resource-consuming motor output that will reinstate homeostasis; (3) Next, the CNS engages in a series of neuronal activities to initiate the selected behavior and suppress conflicting behaviors, a function achieved through subsets of designated initiatory circuits in the CNS, to (4) activate motor systems, i.e. motor neurons which set in motion the relevant musculature, through sets of descending inter-neurons. The nature of the motor response, even in relatively ‘simple’ animals such as invertebrates, is often extremely complex and involves numerous muscles acting in specific and tightly-coordinated spatio-temporal patterns. This is mediated through Central Pattern Generators (CPGs), a network of sensory, inter- and motor neurons which locally orchestrate the spatio-temporal components of locomotion (such as moving legs in a coordinated fashion during walking or running, flapping wings during flying, etc.); (5) The motor behavior is ideally maintained until the goal of the behavior is reached or until another stimulus occurs, which may again deflect the animal from homeostasis.

Change in internal / S Stimulus ens external environment > Threshold ory B 1 eh Sensory System PNS av io ral 2 Decision CNS 3 Maintenance 5 Initiation

4 Motor Systems PNS

Change in external Response environment

Diagram 1: A simplified model for the homeostatic control of behavior. Functions of the peripheral and central nervous system (PNS and CNS, respectively) are numbered as in the text.

~ 12 ~

The chain of events leading from stimulation to action, and the complexity of the motor response, can be comprehensively illustrated in the thoroughly investigated escape system of the cockroach, which is impaired in stung cockroaches (Figure 3) [Camhi 1984, Camhi and Levi 1988, Schaefer et al. 1994]. Under uncompromised conditions, a wind gust generated by an incoming predator might surpass the sensory threshold of wind-sensitive hairs on the cerci (two specialized appendages on the rear end of the abdomen), evoking action potentials in the sensory afferents which innervate the terminal abdominal ganglion (TAG) of the ventral nerve cord (VNC). These, in turn, evoke action potentials in the Giant Interneurons (GIs) that send their axons to thoracic ganglia and synapse onto specific thoracic interneurons (TIs). The TIs are part of the locomotory CPG in the thorax and activate motor neurons, either directly or indirectly (see Figure 3). The CPG also receives descending neuromodulatory inputs from the cerebral ganglia, which integrate numerous sensory inputs representing the state of the cockroach in relation to the environment (e.g., visual, tactile or chemical inputs, etc.) [Ritzmann et al. 1991, Casagrand and Ritzmann 1992]. Under appropriate conditions, in an awake and alert cockroach (see, for instance, [Watson and Ritzmann 1994]), the wind gust applied to the cerci evokes a biphasic escape behavior. First, the insect produces a stereotypic ballistic movement (known as the ‘startle’ response), where the cockroach rapidly turns away from the stimulus source. Then, a rhythmic running behavior is initiated, designed to allow the cockroach to escape its predator. Running is maintained until the cockroach finds a hiding place or otherwise escapes its predator.

Failure to escape aversive stimuli, such as a wind gust applied to the cerci, is a hallmark of the venom-induced hypokinetic state observed in stung cockroaches. Which of the aforementioned steps in the behavioral algorithm are impaired in stung cockroaches? What are the neuronal underpinnings of these deficits? In the following section, I will describe how former investigations (i.e. prior to my arrival at the laboratory in 2003) addressed these issues and what were their main findings. Investigations of other aspects of cockroach envenomation, i.e., the transient leg paralysis, excessive grooming and changes in the metabolic rate, are described in detail in Libersat et al. [2009] and thus will not be discussed here.

~ 13 ~

Figure 3: The basic escape circuitry of the cockroach and hypothesized sites of action of the venom, as postulated by Fouad et al. (1996). 1,2 represent mechanoreceptors of the antennae and cerci, respectively. 3 represents the descending mechanoreceptive interneurons (INs), 4 the giant INs, 5 the thoracic INs, 6 the leg motor neurons, 7 the thoracic INs which are part of the locomotion Central Pattern Generator (CPG). Cell bodies labeled with an M represent populations of neuromodulatory cells that modulate the synapses between sensory and thoracic INs, and/or the synapses between thoracic INs and specific motor neurons, in the CPG. These cell populations receive descending inputs from the cerebral ganglia, namely the supra-esophageal ganglion (‘brain’) and sub-esophageal ganglion (SOG), which are known to regulate locomotion. In stung cockroaches, the venom is hypothesized to affect neurons in cerebral ganglia to repress the activity of neuromodulatory neurons in the thoracic ganglia, thereby repressing the locomotory CPG.

~ 14 ~

1.3. Prior mechanistic investigations of the venom-induced hypokinesia The first mechanistic studies of venom-induced hypokinesia in cockroaches stung by the Jewel Wasp were published in 1994 and 1996. In their 1994 article, Fouad et al. described the course of events in the wasp-cockroach interaction and quantitatively showed that approximately 30 minutes after the head sting (but not after the thoracic sting alone), cockroaches failed to initiate characteristic escape responses to supra- threshold wind stimuli applied to the cerci. Yet, signal transmission between wind afferents and the Thoracic Interneurons (TIs) appeared intact, as wind stimuli evoked rigorous action potentials which travelled normally through the Giant Interneurons (GIs) (Figure 3). Hence, the head sting appears to depress the strength of the connection between the GIs and the TIs, or between the TIs and motor neurons. This was further investigated in 1996, when Fouad et al. recorded activity in the slow and fast coxal depressor motor neurons (Ds and Df motor neurons, respectively) in control and stung cockroaches. Neuronal activity in these motor neurons control the coxal depressor leg muscles which, upon different patterns of activation, mediate the maintenance of body posture and the stance phase during rhythmic motor behaviors (such as walking and escape/running). As discussed above, in normal cockroaches, a typical motor response to a supra-threshold wind stimulus consists of an initial stereotypic ‘startle response’, in which the cockroach jumps in place and rapidly rotates away from the stimulus source, followed by running. The initial ‘startle’ is characteristically accompanied by a burst in Ds and Df activity, followed by their rhythmic discharge during running. Fouad et al. [1996] found that in stung cockroaches, a wind stimulus evokes a burst in Ds alone which is not followed by rhythmic activity in either motor neuron. This is in agreement with behavioral observations, where a stimulus sometimes evokes a ‘jump’ in place, not followed by running. However, Df and Ds motor neurons did receive stimulus-mediated synaptic inputs from the TIs, demonstrating that the deficit in escape is not a motor deficit per se (Figure 3). Taken together, these two important studies highlighted that it is not the peripheral sensory or motor circuitries (i.e., steps #1 or #4 in the behavioral algorithm described above, respectively) that the head sting depresses. Since the venom is injected into the CNS, the venom must, therefore, manipulate circuitries involved in the decision, initiation and/or maintenance of locomotion (i.e., steps #2, 3 and/or 5 in the behavioral algorithm).

~ 15 ~

Considering that the wasp’s stinger penetrates the cockroach’s head from the ventral neck cuticle (see Figures 1C and 2A), Maxwell-Lefroy suggested as early as 1909 that the sting might be directed at the sub-esophageal ganglion (SEG). This anecdotal suggestion was investigated experimentally by Libersat et al. [1999], where wind- evoked escape responses were compared between stung cockroaches, ‘brainless’ cockroaches (in which the supra-esophageal ganglion was surgically removed) and sham-operated cockroaches. The group found that the responses of brainless and sham-operated cockroaches were similar and intact, and dramatically different from the responses to wind realized in stung cockroaches. This indirectly implied that the venom’s effect is primarily on the SEG and that this ganglion must, therefore, play an important role in modulating the thoracic portion of the escape circuitry. In addition, the study also showed that Df motor neurons can still be recruited in stung cockroaches with bath application of the muscarinic agonist, pilocarpine. The conclusions from this study were that descending input, presumably from the SEG, modulates premotor TI to motor neuron connections in the thoracic escape circuitry. A part of my own study [Gal and Libersat 2008, 2010] was aimed at confirming this initial hypothesis directly.

1.4. Research goals

The hypokinetic state characteristic of stung cockroaches is a direct result of neurotoxins precisely injected inside the cerebral ganglia (brain and SEG) during the head sting. These neurotoxins must, therefore, modulate descending inputs from the brain and/or SEG to thoracic ganglia, where motor patterns are generated. The goal of my study was thus three-fold: First, to characterize in more detail the venom-induced hypokinesia and learn which aspects of the stimulation-to-action behavioral algorithm described above are impaired in stung cockroaches; second, to localize the neuronal substrates responsible for the venom-induced hypokinesia; and third, to identify the mechanisms which allow the wasp to precisely locate the cerebral ganglia inside the cockroach’s head capsule during the head sting. It was my hope that identifying the neuronal underpinnings of the venom-induced hypokinesia would reveal how locomotion is regulated in cockroaches, in general, and probably in other insects as well.

~ 16 ~

2. Published work

2.1. Synopsis of the published work It was previously shown that, in stung cockroaches, some motor behaviors (such as walking and escape) are severely impaired while others (such as righting and flying) appear to be spared. To understand how venom injected inside the cerebral ganglia manipulates the expression of some behaviors but not others, one must first understand how these ganglia, in general, modulate the expression of different motor patterns. My first objective was thus to characterize the descending effects of the brain and SEG on different motor patterns in non-stung cockroaches. To this end, I employed lesion experiments in which I removed the descending input from one cerebral ganglion at a time and quantified the resulting motor deficits in a battery of behavioral paradigms. A pre-requisite for such investigation, however, was to selectively ablate descending input from the SEG while leaving descending inputs from the brain intact. This proved problematic since the SEG is anatomically located between the brain and the thoracic ganglia, and thus cutting the neck connectives caudal to the SEG would remove descending inputs from both cerebral ganglia indiscriminately. Thus, I first established a methodology to remove descending inputs from the SEG with minimal interference to inputs descending from the brain. Briefly, as demonstrated in article #1 (“New vistas on the initiation and maintenance of insect motor behaviors revealed by specific lesions of the head ganglia”), the method was based on the fact that descending neurons from the SEG have cell bodies contralateral to their axons, while descending neurons from the brain run through the periphery of the SEG to reach the thorax. Thus, as I demonstrated in the article, a mid-sagittal section through the SEG (Figure 4) removes descending inputs from the SEG while leaving those axons descending from the brain anatomically and functionally intact. The technique allowed me to methodologically test different motor behaviors after selectively ablating the brain (by cutting the circumesophageal connectives caudal to the ganglion), the SEG (by a mid-sagittal section), or both cerebral ganglia (by cutting the neck connectives caudal to the SEG). Based on the behavioral and electrophysiological data collected, I presented a functional model to describe how different stimuli evoke different motor behaviors in cockroaches, and how these are regulated by each cerebral ganglion. One important finding with particular regard to the venom-induced manipulation of cockroach locomotion was that the brain and

~ 17 ~

SEG exert opposite tonic descending inputs onto thoracic locomotory centers, with the brain down-regulating and the SEG up-regulating the expression of walking- related behaviors. As will be discussed later in this thesis, this has provided evidence confirming a long-standing hypothesis in the neurobiology of insect locomotion.

A B

Stinging direction 4 3 1 2

C D Mx Mx AF

Ant SEG Eye SEG NC NC

Figure 4: Surgical procedures used throughout my research. A. Schematic drawing of the ventral head capsule. Dashed lines and arrows represent the locations and directions of cuticular incisions used to expose the SEG and neck connectives. For experiments not involving a wasp sting subsequent to surgery, the SEG was exposed by cutting the submentum (largest part of the labium) in the order 1Æ2Æ3. This formed a flap that opened rostrally to expose the ventral nervous system. For SEG removal prior to wasp stinging, the SEG was exposed by following incisions 2Æ3Æ4, forming a flap that was opened laterally. The wasp was encouraged to attack from the right side of the cockroach (black arrow) and to penetrate the head through the intact part of the cuticle. B. Schematic drawing of the dorsal head capsule. Dashed lines represent the location of cuticular incisions used to create a flap and expose the brain. C-D. The Split-SEG preparation. Ventral view (D: close-up detail of C) of a cockroach demonstrating the mid-sagittal section through the SEG. For this picture, the ventral neck cuticle, labium and neck muscle were removed post mortem to expose the SEG. A small piece of aluminum foil (AF) is inserted between the two halves of the split SEG to demonstrate the location of the lesion. Mx: maxilla; Ant: antenna; Eye: compound eye; NC: neck connective.

~ 18 ~

The next step in my research was to examine in detail which and how subsets of motor behaviors are impaired in stung cockroaches (article #2, “A parasitoid wasp manipulates the drive for walking of its cockroach prey”). The rationale and aim behind these experiments was two-fold; first, to correlate motor deficits of stung cockroaches with motor deficits of non-stung cockroaches with cerebral ganglia lesions, and second, to identify which subsets of behavioral processes, i.e., steps in the behavioral algorithm described in Diagram 1, are impaired. To characterize venom- induced deficits in the selection, initiation and maintenance of locomotion in stung cockroaches, I employed behavioral paradigms traditionally used to test motivation in mammalian models of depression. These were complemented with electromyograms (EMG) recorded from coxal muscles during ongoing motor behaviors. Briefly, this project revealed that (1) motor deficits in stung cockroaches highly resemble those observed in cockroaches with SEG lesions, and (2) the threshold for initiation of walking-related behaviors (but not the other motor behaviors investigated) is increased in stung cockroaches and their ability to maintain locomotion, once initiated, is impaired. Thus, the sting specifically affects the motivation or ‘drive’ of the cockroach to initiate and maintain walking-related behaviors, rather than inducing a general ‘sleep-like’ state of decreased arousal.

Next, based on the results of the two previous studies, I focused on identifying the neuronal substrate for the venom’s hypokinetic effect (article #3, “A wasp manipulates neuronal activity in the sub-esophageal ganglion to decrease the drive for walking in its cockroach prey”). Since stung cockroaches resemble cockroaches with SEG lesions in many of the behavioral paradigms tested, and since cockroaches with brain lesions, in contrast, show increased locomotor activity, I hypothesized that the venom manipulates circuitries in the SEG to decrease the drive for walking. The hypothesis was confirmed using behavioral, neuro-pharmacological and electrophysiological methods which highlight the role of the SEG in venom-induced hypokinesia, as well as in the general determination of the ‘rest state’ of cockroaches. This work was also the first in which venom milked from wasps was successfully injected directly into the cockroach cerebral ganglia to reproduce the deficits observed in naturally-stung cockroaches.

~ 19 ~

Throughout my studies, I also took an active part in writing three scientific review papers. The first publication [Gal et al. 2005], which I also included in this thesis (article #4, “Parasitoid wasp uses a venom cocktail injected into the brain to manipulate the behavior and metabolism of its cockroach prey”), summarizes recent advancements in the study of the various behavioral manipulations induced in cockroaches by the A. compressa venom. It also presents some of my preliminary results regarding the mechanisms of venom injection, which were not published elsewhere. The second review [Libersat and Gal 2007] is a book chapter which discusses neuro-manipulation of hosts by parasitoid wasps, in general, and the unique interaction between the Jewel Wasp and its cockroach prey as a specific case-study. The third review [Libersat et al. 2009] summarizes our current knowledge of the behavioral manipulation induced in hosts by parasitic insects and insect parasites. It is currently, to my knowledge, the most thorough and updated review of the subject in scientific literature. I strongly encourage the reader to examine these three publications and gain a wider perspective of the subject of this thesis.

~ 20 ~

J Comp Physiol A (2006) 192:1003–1020 DOI 10.1007/s00359-006-0135-4

ORIGINAL PAPER

New vistas on the initiation and maintenance of insect motor behaviors revealed by specific lesions of the head ganglia

Ram Gal · Frederic Libersat

Received: 5 March 2006 / Revised: 11 April 2006 / Accepted: 23 April 2006 / Published online: 30 May 2006 © Springer-Verlag 2006

Abstract In insects, thoracic pattern generators are Keywords Periplaneta americana · Brain · modulated by the two head ganglia, the supraesopha- Subesophageal ganglion · Locomotion · Pattern geal ganglion (brain) and the subesophageal ganglion, generator which act as higher-order neuronal centers. To explore the contribution of each head ganglion to the initiation Abbreviations and maintenance of speciWc motor behaviors in cock- B-DINs Brain descending interneurons roaches (Periplaneta americana), we performed speciWc CirC Circumesophageal connectives lesions to remove descending inputs from either the Df Fast coxal depressors brain or the subesophageal ganglion or both, and quan- DMI Descending mechanosensory interneuron tiWed the behavioral outcome with a battery of motor Ds Slow coxal depressors tasks. We show that ‘emergency’ behaviors, such as EMG Electromyogram escape, Xight, swimming or righting, are initiated at the PG Pattern generator thoracic level independently of descending inputs from SEG Subesophageal ganglion the head ganglia. Yet, the head ganglia play a major SEG-DINs SEG descending interneurons role in maintaining these reXexively initiated behaviors. By separately removing each of the two head ganglia, we show that the brain excites Xight behavior and Introduction inhibits walking-related behaviors, whereas the sub- esophageal ganglion exerts the opposite eVects. Thus, Locomotion in animals is generated by two functional control over speciWc motor behaviors in cockroaches is units: a unit to generate muscle activity and a decision- anatomically and functionally compartmentalized. We making unit. First, diVerent muscles associated with a propose a comprehensive model in which the relative given appendage must be rhythmically activated in a permissive versus inhibitory inputs descending from speciWc spatio-temporal pattern to allow coordinated the two head ganglia, combined with thoracic aVerent movements. In all animals studied thus far, this func- sensory inputs, select a speciWc thoracic motor pattern tion is generated by local networks commonly referred while preventing the others. to as pattern generators or PGs (Marder and Bucher 2001; Grillner and Wallen 2002; Marder et al. 2005). These local networks include sensory, motor and central neuronal components (Bässler and Büschges 1998; Dickinson et al. 2000; Duysens et al. 2000; Pearson R. Gal · F. Libersat (&) 2004; Zill et al. 2004; Büschges 2005). Rhythmicity in Department of Life Sciences and the Zlotowski these local networks can be expressed spontaneously Center for Neuroscience, Ben-Gurion University of the Negev, (Knop et al. 2001) but more often requires an appropri- P.O. Box 653, Beer-Sheva, 84105, Israel ate mechanical or pharmacological stimulus (for recent e-mail: [email protected] reviews: Arshavsky et al. 1997; Marder and Bucher

123 1004 J Comp Physiol A (2006) 192:1003–1020

2001; Dietz 2003; Grillner 2003; Libersat and PXueger One caveat of lesion experiments is that none of 2004). Typically, PGs that produce the spatio-temporal these procedures has addressed the role of the SEG pattern characteristic of a given locomotory behavior directly. An interesting approach used by Matsumoto reside in the “lower” portion of the central nervous sys- and Sakai (2000) to study copulation in crickets con- tem: the spinal cord in vertebrates or the thoracic and sists of performing a mid-sagittal section through the abdominal nerve cord in arthropods. The second func- SEG. This removes the eVect of SEG descending inter- tional unit is that involved in decision making. Here, a neurons (SEG-DINs) on thoracic PGs while leaving speciWc neuronal Wring pattern is generated to select the connectivity between the brain and thoracic ganglia speciWc PGs while inhibiting others (Kien and Altman (the ‘through running axons’ of brain descending inter- 1992a; Briggman et al. 2005; Grillner et al. 2005). This neurons, B-DINs) intact for the following reason. A function is presumably controlled by so-called “higher- midline cut through the SEG eliminates SEG neurons order” neuronal centers: the brain in vertebrates or the with contralateral descending axons, which comprise head ganglia in arthropods. Accumulating evidence the majority of the SEG-DINs (Kien et al. 1990), suggests that the diVerences between arthropods and thereby eliminating the descending inXuence of the vertebrates (including mammals) in brain structures SEG. The direct descending pathway from the brain to and their eVect on locomotion are, in essence, more the thorax (the ‘through running axons’) is mostly anatomical than functional (Kien and Altman 1992b; unaVected by this midline cut since this pathway runs Strausfeld 1999; Svidersky and Plotnikova 2002). through the margins of the SEG (Altman and Kien In insects, locomotory PGs reside in the three tho- 1987; Matsumoto and Sakai 2000). racic ganglia and associated appendages (Wilson 1961, Using this new approach, we sought to re-examine 1967; Pearson and Iles 1970; Büschges 2005) and their the contribution of each head ganglion separately, and expression is modulated, at least in part, by the two also the contribution of speciWc sensory inputs, to the head ganglia that comprise the insect’s “higher-order” regulation of the expression of Xight and walking- neuronal centers (Kien and Altman 1992a). Anatomi- related behaviors generated in the thorax of the Amer- cally, the Wrst, the supraesophageal ganglion (brain) is ican cockroach. To this end, we used speciWc lesions of connected to the next, the subesophageal ganglion the cockroach head ganglia, combined with alterations (SEG), via the circumesophageal connectives (CirC). in aVerent sensory inputs, and quantiWed the resulting The SEG, in turn, is connected to the thoracic ganglia motor deWcits. via the neck (cervical) connectives. Several motor behaviors in insects can be initiated with appropriate stimuli and performed without the Materials and methods head ganglia, for example Xight (Wilson 1961), walking (Roeder 1948), and grooming with the metathoracic Animals legs (Eaton and Farley 1969; Reingold and Camhi 1977; Berkowitz and Laurent 1996; Matheson 1997). Adult male cockroaches Periplaneta americana L. were Insects with the CirC cut exhibit long bouts of walking raised in crowded conditions in plastic barrels at activity (Roeder 1937; Bässler 1983; Kien 1983; Ridgel 24–26°C on a 12L:12D cycle. Water and cat food were and Ritzmann 2005), suggesting that the brain is a provided ad libitum. For behavioral assays, cock- source of inhibitory inXuence on thoracic walking cen- roaches of approximately the same age, 2–6 weeks ters. In contrast, lesions of the neck connectives dra- after the last molt, were used. matically decrease spontaneous and evoked walking in locusts (Kien 1983) and reduce excitatory input to the Lesions crawling PG in Manduca (Johnston et al. 1999). Schae- fer and Ritzmann (2001) demonstrated the importance Cockroaches were mildly anesthetized with carbon of descending neural inXuences on escape behavior by dioxide and pinned onto a Petri dish. An insect pin sta- examining cockroaches following removal of all ple was gently pressed against the cockroach neck to descending inputs from the head. The reduction of leg decrease hemolymph loss during surgery. After each movement in the ‘headless’ cockroaches is accompa- surgical procedure, the staple was slowly removed to nied by a decrease in fast motor neuron activity, sug- allow gradual recovery of the hemolymph Xow to the gesting a reduction of excitability within the thoracic head and the wound sealed itself by coagulation. To portion of the escape circuitry. These observations sug- produce “brainless” cockroaches, we disabled brain gest, indirectly, that the SEG is a source of excitatory descending interneurons (B-DINs) projecting to the inXuence on walking. thorax while leaving the SEG-DINs intact: a small Xap 123 J Comp Physiol A (2006) 192:1003–1020 1005 was opened in the dorsal cuticle between the com- USA), and the number of stained somata in the brain pound eyes and the brain was surgically removed by and SEG was counted in control (n = 8) and SEG-less cutting the CirC. Care was taken to avoid damaging (n = 10) preparations. the surrounding tissue and tracheae. To produce “SEG-less” cockroaches, the SEG was exposed by Electrophysiology opening a cuticular Xap in the ventral head capsule; then we performed a mid-sagittal slit through the SEG To verify that splitting the SEG does not functionally with a modiWed scalpel. Occasionally, a small piece of impair B-DINs projecting to the thorax, we used an aluminum foil was inserted between the two halves of insect pin connected to a piezo crystal to apply tactile the split SEG to evaluate the exact position of the slit stimuli, at 15-s intervals, to the tip of the left antenna of postmortem. One should notice, though, that splitting four cockroaches. Such stimuli recruit the descending the SEG also removes the inXuence of descending mechanosensory interneurons (DMIs) involved in trig- brain neurons that relay in the SEG and that it does gering touch-evoked escape behavior (Burdohan and not directly aVect SEG neurons with ipsilateral axons. Comer 1996). To minimize movement, the proximal Despite the drawbacks of this surgical procedure, for two-thirds of the antenna were conWned in a plastic the sake of clarity, we will refer to the experimental tube, with only the tip exposed to the stimulus. We group with their SEG split as SEG-less cockroaches. recorded responses from the right neck connective, To produce “headless” cockroaches, we opened a Xap which contains axons of DMIs, with a pair of extracel- in the ventral neck cuticle to expose the neck (cervical) lular silver hook electrodes. After recording several connectives. By cutting both connectives with Wne scis- tactile-evoked responses, a mid-sagittal section was sors we disabled all interneurons descending from the performed in the SEG in situ and stimuli were again head ganglia. applied to the antenna. Next, as a negative control, the CirC were cut with Wne scissors. Recordings were con- Staining ducted for 5 min in each case, and the cockroach was given 5 min to recover between surgical procedures. To verify that splitting the SEG impairs SEG-DINs while leaving the B-DINs projecting to the thorax ana- Behavior tomically intact, we performed backWlls of the axons running through the right neck connective in control Behavioral tasks were conducted at room temperature and SEG-less cockroaches. Cockroaches were anesthe- 1 day after the surgical procedure and occasionally tized with carbon dioxide and placed ventral side up on videotaped for later observation and analysis. Three a Petri dish covered with Sylgard. The neck connec- groups of control cockroaches were used: (1) untreated tives were exposed by removing a Xap of ventral cuticle cockroaches; (2) controls for brainless cockroaches, between the thorax and SEG. In SEG-less prepara- in which a Xap was opened in the dorsal head cuticle, tions, a mid-sagittal slit through the SEG was per- the antennae were severed at the base and the ocelli formed at this time. The right neck connective was and compound eyes were covered with paint; (3) con- then cut and the stump placed inside a Vaseline well trols for SEG-less and headless cockroaches, in which Wlled with 4% cobalt hexamine chloride solution (East- a Xap was opened in the ventral head cuticle. Since man Kodak Company, Rochester, NY, USA). To allow there was no signiWcant diVerence between these diVusion of the dye to the head ganglia, the cockroach groups (data not shown), they were all averaged was placed in a moist chamber at 4°C. We determined together. empirically the optimal diVusion time for best quality staining of cell bodies to be 40–48 h for brain prepara- Walking tions and 20–24 h for SEG preparations. After diVu- sion of the dye, the nervous system was dissected out A cockroach was tethered with a metal clamp waxed to and a few drops of ammonium sulWde were applied to the dorsal pronotum and connected to a micromanipu- the dish and quickly washed out. The preparations were lator. The cockroach was held in a position that then Wxed in 4% paraformaldehyde in cockroach saline, allowed it to assume a normal walking posture on a dehydrated with a graded series of ethanol, cleared in Petri dish lightly covered with vegetable oil to allow methyl salicylate and observed as whole-mounts under friction-free walking. It was given 5 min to adjust to the light microscope. Somata of the stained neurons were novel environment and then the total duration of spon- drawn with a commercial 3D reconstruction system taneous walking was measured with a stopwatch during (Neurolucida, MicroBrightField, Colchester, VT, a 10-min period. 123 1006 J Comp Physiol A (2006) 192:1003–1020

Escaping bursts in the cycle of the right hindwing. Overall, phases were calculated for roughly 1,000 Xight cycles A cockroach was placed in the center of a round arena from eight diVerent cockroaches. Low frequency (30 cm radius) with a plastic cup. The cup was then movement artifacts were digitally Wltered. removed and the distal part of the wings was lightly touched with a small brush. Normal cockroaches usu- Electromyograms in freely behaving cockroaches ally ran in a straight line to the arena wall in response to such stimulus and therefore we measured the distance Cockroaches were mildly cool-anesthetized and their of the escape response, with the maximum distance wings removed at the base. They were then immobi- being the radius of the arena. The procedure was lized ventral side up on a cold plate (4°C) by restrain- repeated three times for each cockroach, at 1-min inter- ing all legs with insect pin staples. Three pairs of EMG vals, and the average escape distance was calculated. electrodes—50 m steel wires insulated to the tip— were implanted in the coxal depressor (extensor) mus- Righting cles of the right mesothoracic leg (muscle 135d) and the two metathoracic legs (muscles 177d) [notations Cockroaches were positioned on their backs on coarse according to Carbonell (1947); electrode implantation sandpaper using Wne forceps, and their ability to right, according to Watson and Ritzmann (1998)]. A ground that is Xip over and stand upright was noted. Cock- electrode was implanted in the lateral, non-vascular roaches that could not right in less than 30 s were con- pronotum. All electrodes were fastened to the cuticle sidered lacking the ability to right. We tested each with dental wax. The cockroach was then turned dorsal cockroach three times at 1-min intervals, and calcu- side up. On the dorsal side of the thorax, the electrodes lated the percentage of successful righting. The dura- were twisted to form a cable, which was surrounded tion of struggling (i.e. leg movements) while the with a small-diameter rubber tube to protect the wires cockroach was overturned was also noted. during handling. This cable, connected to a micromanipulator, also served as a tether. Since it was waxed Wind-evoked Xying dorsally over the cockroach’s center of gravity, the cable allowed the cockroach to stay horizontal when We recorded Xight with three pairs of electromyogram lifted or manipulated. (EMG) electrodes—50 m copper wires insulated to The cockroach was held by the cable tether at a con- the tip—which we implanted in the subalar depressor stant height above an oiled Plexiglas surface in a posi- muscles of the right forewing and two hindwings, as tion that allowed it to assume normal walking posture. described in Libersat (1994). BrieXy, a cockroach was The walking surface was secured above a swimming pinned dorsal side up to a recording platform after area: an opaque, electrically grounded container ablating the legs above the trochanter and the wings (30 £ 22 £ 15 cm) Wlled with 10 cm of water at 25°C. back to stumps. A brief wind puV lasting roughly Before testing began, the tethered cockroach was left 150 ms was delivered at the cerci from above and in undisturbed for 30 min on the walking surface to fully front of the animal (i.e., head-to-tail direction) from a recover from anesthesia and electrode implantation. distance of 2 cm by a wind stimulator. Wind velocities During walking tests, cockroaches walked spontane- were recorded separately with a hot wire anemometer ously and, from time to time, were induced to escape (HPA 98, Strata Mechanics Research Institute, Cra- by touching the tip of the abdomen with a Wne paint- cow, Poland). For all Xight experiments, room temper- brush. After testing walking for approximately 10 min, ature was kept at 27–28°C. We began by applying low the cockroach was induced to swim by removing the velocity wind puVs, at 1-min intervals, and increased walking surface, allowing 4–6 s during which the cock- the velocity until Xight was initiated. We considered roach was held horizontal in the air by the cable tether, the Xight threshold to be the lowest wind velocity at and then lowering the cockroach to the swimming area which Xight was initiated in at least two out of three tri- below. To avoid muscle fatigue, cockroaches were als. After the threshold wind velocity was reached, a allowed to swim for a maximum of 1–2 min, then raised series of Wve supra-threshold stimuli (0.5 m/s) were with the micromanipulator and left undisturbed for at applied at 1-min intervals to examine Xight duration, least 20 min on the walking surface. The procedure was measured as the duration between the Wrst and last usually repeated twice for each animal. We found the EMG spikes of the right hindwing. We also evaluated recovery periods after electrode implantation and after interwing coordination during Xight episodes by calcu- a swim trial, as well as the 2-min trial limit, to be critical lating the phase of the right forewing and left hindwing for the success of swimming experiments. Allowing 123 J Comp Physiol A (2006) 192:1003–1020 1007 cockroaches to swim for longer periods of time or not reaching the thorax, was similar in SEG-less prepara- allowing enough time between swimming sessions usu- tions and control (95 § 13 and 106 § 13, respectively; ally proved fatal to the animal. In these incidents the P = 0.273, t test). In contrast, the average number of immersed cockroach stopped swimming for a short stained SEG-DINs was signiWcantly smaller in SEG-less period, which was followed by strong, usually uncoor- preparations compared with controls (23 § 5 and dinated movements of all legs. These movements were 88 § 13, respectively; P < 0.001, t test). As expected, in often accompanied with fast depressor potentials (Df) SEG-less preparations, all stained SEG-DINs had soma and lasted about 10 s, following which the cockroach ipsilateral to the Wlled connective (Fig. 1b). stopped moving altogether and then died. Those inci- To verify the physiological integrity of B-DINs in dents were omitted from the results. SEG-less cockroaches, we applied tactile stimuli to the During swimming, ‘stepping’ frequency (cycles/sec- left antenna while recording action potentials from the ond) was calculated for the left metathoracic leg based right neck connective (Fig. 2a). The responses were on the muscle activity of the coxal depressor muscle, comparable before and after splitting the SEG, but and the occurrence of slow depressor potentials (Ds) completely disappeared after cutting the CirC allowed us to identify leg retractions (‘stance’). To (Fig. 2b). evaluate interleg coordination, we calculated the phase These results suggest that splitting the SEG removes of the right meso- and metathoracic coxal depressor over 70% of the SEG-DINs’ input to the thorax, with bursts in the cycle of the left metathoracic depressor little structural or functional damage to B-DINs pro- burst, i.e. for contralateral legs in successive segments jecting through the margins of the SEG directly to the and for contralateral legs in the same segment. Such thorax. recordings allowed the identiWcation of tripod-gait coordination, where contralateral legs in successive The eVect of head ganglia lesions on motor behaviors segments move together and contralateral legs in the same segment move alternatively. Overall, phases were To examine the speciWc eVects of B-DINs and SEG- calculated from roughly 1,000 swimming cycles in 12 DINs on thoracic motor behaviors, we tested control diVerent cockroaches. and lesioned cockroaches in a battery of behavioral tasks. Since the behavioral deWcits of headless cock- Data analysis roaches have already been examined by others (see Zill 1986; Schaefer and Ritzmann 2001; Ridgel and All electrophysiological recordings were stored on a Ritzmann 2005 and references therein), we focus our videotape with Data Neuro-corder DR-890 (Neuro analysis on control, brainless and SEG-less cock- Data Instruments, New York, NY, USA), digitized roaches. with Micro 1401 MKII analog to digital board and pro- cessed with Spike2 (CED, Cambridge, UK) data acqui- Flying sition software. To test for diVerences among groups, we used t test for normally distributed data and Mann– To characterize the eVect of descending inputs on Xight Whitney Rank Sum Test for non-normally distributed behavior, we removed the cockroach legs and tested data. Unless otherwise stated, all numerical data in this the eVect of a wind gust applied to the cerci (see Liber- work are presented as means § standard deviation, sat 1994). We Wrst tested the eVect of head ganglia and n represents the number of animals tested. lesions on the threshold wind velocity required to elicit Xight (Fig. 3a). The average threshold velocity was not signiWcantly diVerent between control (n = 25), SEG- Results less (n = 16) and headless (n = 6) cockroaches (0.10 § 0.09 m/s, 0.08 § 0.07 m/s and 0.10 § 0.13 m/s, The SEG-less preparation respectively; t test). In sharp contrast, the threshold velocity of brainless cockroaches was markedly ele- To verify that splitting the SEG in cockroaches disables vated, as none of the tested brainless cockroaches SEG-DINs while leaving intact the through running (n = 12) showed Xight behavior when the wind was at axons of B-DINs, we backWlled the right neck connec- 0.5 m/s (maximal wind velocity tested). tive with a cobalt solution in control and SEG-less cock- To examine the eVect of head ganglia lesions on the roaches and counted the number of cells in the brain and duration of wind-evoked Xight, we applied a wind puV SEG (Fig. 1). The average number of stained somata in of 0.5 m/s to the cerci and measured the duration of the the brain, i.e. the number of B-DINs with intact axons resulting Xight episode (Fig. 3b). On average, this 123 1008 J Comp Physiol A (2006) 192:1003–1020

Fig. 1 The split-SEG proce- a dure. a Micrographs are superimposed on a schematic drawing of the brain (Br) and Br SEG, in control (left) and SEG-less (right) cockroaches. BackWlls were performed from the right neck connec- 1mm tive (asterisk). Micrographs show one focal plane of a whole-mount preparation. b Camera Lucida reconstruction of stained cells in the SEG head ganglia. In the SEG, cells in white were stained in both control and SEG-less prepa- * rations. Cells in black were * stained only in control, but b c Control SEG-less not in SEG-less preparations, 120 demonstrating the fraction of *** SEG-DINs that was eliminated by splitting the SEG. As 100 expected, almost all somata in black are contralateral to the 80 site of backWll (asterisk). c The number of stained somata of 60 B-DINs was similar in control and SEG-less cockroaches. In 40 contrast, splitting the SEG sig-

niWcantly reduced the number ofNumber somata stained 20 of stained somata of SEG- DINs (***P <0.001) 0 * B-DINs SEG-DINs stimulus evoked a Xight burst of 1.6 § 1.3 s in control on interwing coordination could not be determined since cockroaches (n = 11) and failed to evoke Xight in brain- these cockroaches did not show Xight behavior. less cockroaches (n = 12). In contrast, SEG-less cockroaches (n = 6) seemed unable to terminate Xight, and Walking Xew signiWcantly longer than controls (35.3 § 32.2 s; P <0.05, t test). Also, in some cases, Xight bursts The total duration of spontaneous walking during a 10- occurred spontaneously in these cockroaches with no min trial was observed in 14 tethered cockroaches of apparent stimulus. Headless cockroaches (n =5) each group (Fig. 5a). Control cockroaches spontane- responded to the wind stimulus with a very short Xight ously walked, on average, a total of 1.4 § 1.2 min during burst (0.3 § 0.2 s). the trial period. Brainless cockroaches spontaneously To examine whether removing the SEG alters the walked for signiWcantly longer durations than controls interwing coordination during Xight, we calculated the (7.5 § 1.8 min; P < 0.001, t test), and seemed relatively phase of the EMG spike of the depressor muscles of the unable to terminate walking bouts. In contrast, SEG- right forewing (R1) and left hindwing (L2) in the cycle less, as well as headless cockroaches, were quite hypoki- of the right hindwing (R2) in control and SEG-less cock- netic. Walking and walking attempts were almost roaches (n = 4 for each group) (Fig. 4). The average completely absent in SEG-less cockroaches (walking phase of the ipsilateral wings (R1–R2 phase) was similar duration was 0.02 § 0.04 min, signiWcantly less than con- in control and SEG-less cockroaches (¡0.17 and ¡0.10, trols; P < 0.001, Mann–Whitney Rank Sum Test), and respectively), although the distribution around the aver- headless cockroaches never walked spontaneously. age was somewhat wider in controls. Likewise, the average phase of the contralateral hindwings (L2–R2 phase) Escaping was the same (0.01) in control and SEG-less cockroaches. This suggests that the basic interwing coordina- Twenty-Wve cockroaches of each group were tested for tion is independent of the SEG. The eVect of the brain escape behavior elicited by a tactile stimulus to the 123 J Comp Physiol A (2006) 192:1003–1020 1009

a b i.e. Xip over and stand upright, was tested during a 30-s trial period (Table 1). When overturned, all control 25ms cockroaches showed strong struggling movements and Ant were able to successfully right themselves, usually within less than 3 s. None of the operated cockroaches, except for one SEG-less cockroach that righted after Control 21 s, was able to right successfully during the 30 s trial. Brain In fact, these cockroaches stayed in the overturned posture for much longer periods, as long as 24 h after CirC 2 the beginning of the experiments. Nevertheless, all operated cockroaches did struggle by moving their legs SEG 1 continuously throughout the 30-s trial period. The SEG split struggling always ceased if the tarsi established contact NeckC with a suitable foothold. For example, as soon as we presented a piece of sandpaper to the ventral side of a T1 struggling cockroach, in a way that the legs could grab the sandpaper surface, struggling movements termi- Cut CirC nated. Instead, and similarly to their typical behavior, brainless cockroaches began walking on the inverted Fig. 2 Physiology of B-DINs in SEG-less cockroaches. a Experi- surface and SEG-less and headless cockroaches mental setup. Tactile stimuli (arrow) were applied to the left antenna (Ant) to evoke responses of speciWc B-DINs. These were stopped moving altogether, grabbing the sandpaper recorded from the right neck connective (NeckC) with a pair of surface. extracellular hook electrodes. Broken line represents antennal An interesting observation is that in 60% of the tri- V sensory a erents. Also shown is a descending mechanosensory als, the overturned SEG-less cockroaches (n =15) interneuron. After recording from the intact animal, the SEG was X split mid-sagittaly (dotted line 1), and the responses to antennal showed a ‘behavioral switch’ and initiated ight rather stimuli were again recorded. Next, as a negative control, the input than righting. This Xight sequence, during which strug- of B-DINs to the thorax was eliminated by cutting the circum- gling movements ceased, typically continued more esophageal connectives (CirC) (dotted line 2). T1: prothoracic than 30 s non-stop, but terminated if the legs were pre- ganglion. b Representative extracellular recordings from the neck connective of a cockroach before (control) and after splitting the sented with a foothold as described above. Wingbeats SEG, and after cutting the CirC. The responses before and after never occurred in overturned brainless or headless splitting the SEG were comparable. In contrast, no response was cockroaches, and only rarely occurred in control cock- recorded after the CirC were cut. Arrows mark the onset and roaches. In such cases, however, wingbeats persisted oVset of the tactile stimulus less than 1 s, during which struggling movements with abdomen (Fig. 5b). All control cockroaches responded the legs continued. Thus, unlike SEG-less cockroaches, to the stimulus with a startle followed by running from control cockroaches seem to beat their wings to con- the middle of the arena to the arena wall, a distance of tribute to righting rather than for Xying. 30 cm. Likewise, all brainless cockroaches responded In an overturned cockroach, leg-substrate contact is to the stimulus and reached the arena wall, albeit their absent and loading on the legs is consequently reduced. locomotion appeared slower than controls. The escape This is similar to Xying, during which the legs also do response of SEG-less and of headless cockroaches, not support the cockroach’s weight. We explored the however, was markedly aVected by the lesion. Ninety- mechanism that allows overturned intact cockroaches six percent of the SEG-less and 44% of the headless to initiate righting rather than Xight. When intact cock- cockroaches were able to initiate a form of ‘startle’ roaches (n = 10) were placed on their backs on a pol- response, but all failed to express the subsequent run. ished surface that prevents successful righting due to The average escape distance of SEG-less and of head- lack of foothold, righting was always initiated sponta- less cockroaches was 2.8 § 1.6 cm and 0.5 § 0.9 cm, neously and the cockroach struggled with all legs for respectively, which was signiWcantly shorter than con- long periods of time (> 20 s). Then, leg movements trols (P < 0.001, Mann–Whitney Rank Sum Test). ceased but could easily be evoked again by a wind puV applied to the cerci. Righting Using Wne forceps to hold the pronotum, we lifted the overturned cockroach approximately 10 cm in the Fifteen cockroaches of each group were placed on their air while maintaining its upside-down posture. When backs on coarse sandpaper, and their ability to right, doing so, all cockroaches stopped struggling and 123 1010 J Comp Physiol A (2006) 192:1003–1020

V Fig. 3 E ect of descending ac inputs on Xight. a The distribution of the wind velocity required to elicit Xight is similar in control (empty bars), SEG- less (-SEG; diagonal lines) and headless (horizontal lines) cockroaches. None of the brainless cockroaches (-Br; solid bars) Xew when stimulated at these wind velocities. b A stronger wind stimulus (0.5 m/s) was delivered to the cerci. While brainless animals did not Xy at this wind velocity, SEG-less Xew signiWcantly longer and headless cockroaches signiWcantly less than controls. *P < 0.05; **P < 0.01; ***P < 0.001 rela- b tive to controls. c Representa- tive EMG traces from a wing depressor muscle, showing the response of control, brainless, SEG-less and headless cockroaches to a 0.5 m/s wind stimulus (arrow). Scale bar applies for all traces

X Control initiated ight instead, either spontaneously or after 60 -0.15 0.01 applying a wind puV to the cerci. The same results were R1 ) 50 % 40 obtained for cockroaches in which the eyes and ocelli were covered with paint before testing (n =4). Hence, R2 30 20 the initiation of righting when the cockroach is over-

Occurance ( Occurance 10 L2 turned on the ground does not seem to depend upon 0 the cockroach’s position relative to the gravitation vec- -SEG tor or upon visual or ocellar sensory inputs. We thus 40 -0.10 0.01 R1 ) hypothesized that the wing-substrate contact, which is % 30 maintained in overturned but not in Xying cock- R2 20 roaches, contributes to the selection of righting behav- 10 ior rather than Xight. L2 ( Occurance 0 To test this hypothesis, we removed the fore- and -0.5 0.0 0.5 50ms hindwings back to stumps. Such wingless cockroaches Phase (n = 10) were behaviorally indistinguishable from nor- Fig. 4 Interwing coordination during Xight. EMG traces and mal cockroaches when standing upright. However, a phase histograms describing the coordination between the right ‘behavioral switch’ occurred when the cockroach was forewing and right hindwing (R1–R2; black histograms), and placed on its back, as almost all overturned wingless between the left and right hind wings (L2–R2; gray histograms). X The median phases are also indicated. Wing coordination is typi- cockroaches (nine out of ten) initiated ight rather cal of cockroach Xight and is not impaired in SEG-less (-SEG) than righting (Table 1). Typically, overturned Xight cockroaches episodes were preceded by 1–5 s of righting attempts

123 J Comp Physiol A (2006) 192:1003–1020 1011

a 10 Table 1 Behavior of overturned control, brainless (-Br), SEG- *** less (-SEG), headless (-head) and wingless (-wings) cockroaches, 9 with (right column) and without (left column) leg-substrate con- 8 tact 7 Cockroach position 6 5 4 3 Control Righting Walking 2 -Br Struggling Walking -SEG Flying (60%) or Motionless 1 *** *** Walking duration (min/10min) struggling (40%) 0 -Head Struggling Motionless Control -Br -SEG Headless -Wings Flying Walking b 30 See text for details

25 Terrestrial versus immersed locomotion

20 Tethered cockroaches were allowed to walk on a fric-

15 tion-free surface while muscle activity from the coxae was recorded with EMG electrodes. As described (Fig. 5), on the walking surface, control and brainless 10 cockroaches walked readily while SEG-less and head-

Escape distance (cm) *** *** less cockroaches moved only when stimulated, in 5 which case they usually performed a jump forward characterized by the simultaneous extension of all legs, 0 Control -Br -SEG Headless as can be seen in the EMG recording (Fig. 6a). After testing walking for several minutes, the walk- Fig. 5 EVect of descending inputs on terrestrial locomotion: W ing surface was removed and the cockroach was low- a brainless cockroaches (-Br) spontaneously walked signi cantly W more, and SEG-less (-SEG), as well as headless cockroaches, sig- ered into a water- lled container positioned below. niWcantly less than controls. b Control and brainless cockroaches Water immersion evoked swimming or swimming-like responded to tactile stimuli by moving to the arena wall (30 cm). locomotion in all cockroaches, and we characterized In contrast, SEG-less and headless cockroaches initiated a re- the rhythmicity and interleg coordination of this loco- sponse, but moved a signiWcantly shorter distance. ***P <0.001 compared with controls motion in control, brainless and SEG-less cockroaches. Immersed control cockroaches (n = 4) expressed and then the cockroach initiated Xight, which persisted rhythmic leg movements with an average ‘stepping’ for 6–30 s. In three cockroaches, Xight was not initiated frequency of 3.4 § 0.5 cycles/s (measured for the left spontaneously, but it could be readily evoked with a metathoracic leg), equivalent to slow or moderate ter- wind puV applied to the cerci. Although we did not restrial walking. The ‘stepping’ frequency tended to be analyze overturned Xight in detail, such Xight episodes higher at the beginning of the locomotory response, seemed comparable to normal upright Xight; they were but almost never exceeded 5 cycles/s. Potentials in fast characterized by full Xexion of the legs, postural muscle were more abundant during faster locomotion change of the cerci and high-frequency Xapping of the (see below). Immersed brainless cockroaches (n =4) wing-stumps. These observations suggest that sensilla showed rhythmic locomotion with a ‘stepping’ fre- on the wings are involved in the selection of righting quency of 3.3 § 0.5 cycles/s, which was not statistically rather than Xight. Moreover, in all the tested intact diVerent from controls (P = 0.659, t test). (n = 14) or wingless (n = 10) overturned cockroaches, In contrast, leg movements of immersed SEG-less righting attempts or Xight (respectively) could easily be cockroaches (n = 4) seemed relatively arrhythmic. terminated by presenting the overturned cockroach Immediately upon water immersion these cockroaches with a foothold as described above. These results showed rhythmic activity for several seconds, but suggest that the contact between the legs and substrate rhythmicity was then interrupted with periods of inhibits righting as well as Xight, regardless of whether immobility or sudden extensions of the legs, with no or not the wings are in contact with the substrate. apparent interleg coordination. In some instances, 123 1012 J Comp Physiol A (2006) 192:1003–1020

Terrestrial Immersed a b c Control Df R2 Ds

-Br R2

R3 -SEG

250ms 100ms

Fig. 6 Characteristics of terrestrial versus immersed locomotion: nation. The locomotion of SEG-less cockroaches was evoked by EMG traces from the coxal depressors, demonstrating the coor- tactile stimuli (arrow), and consists of simultaneous extension of dination between the right mesothoracic leg (R2) and the right all legs with no subsequent locomotion. b In water, only SEG-less and left metathoracic legs (R3 and L3) in control, brainless (-Br) cockroaches fail to show spontaneous tripod-gait coordinated and SEG-less (-SEG) cockroaches. Within each group, traces locomotion. Although initiated spontaneously, the responses of were obtained from the same individual during terrestrial (a) and SEG-less cockroaches to water immersion resemble their re- immersed (b, c) locomotion. In a and b, the Df spikes have been sponses to tactile stimuli on land. c Full-scale EMG traces show- trimmed to show only the bursting of the slow depressors (Ds). ing fast depressors (Df) potentials during fast immersed a On land, the spontaneously initiated locomotion of control and locomotion. Ds and Df spikes are indicated with arrows brainless cockroaches is characterized by the tripod-gait coordi- however, the cockroach regained rhythmicity sponta- right and left metathoracic legs (R3 and L3) alternated, neously, though the rhythmic bouts typically persisted and the average R3–L3 phase was 0.51. The phase his- for no more than 5–6 ‘stepping’ cycles (Fig. 6c shows togram (Fig. 7) showed clear peaks at these phases with such rhythmic activity). A few cycles of rhythmic loco- very little distribution around them, indicating fully motion could also be evoked by touching the wings or coordinated tripod-gait while swimming. Similarly to abdomen with a small paintbrush. The average ‘step- controls, brainless cockroaches (n = 4) also showed tri- ping’ frequency during the rhythmic episodes was pod-gait coordination in water, with an average R2–L3 somewhat lower than control and brainless cock- phase of ¡0.01 and an R3–L3 phase of 0.42. However, roaches and the distribution around it much wider the distribution around the peak phases was somewhat (3.0 § 1.3 cycles/s). Nevertheless, since rhythmic epi- wider in brainless compared with control cockroaches. sodes did occur, the water immersion seems to evoke In contrast, the leg movements of immersed SEG-less some motor pattern in SEG-less cockroaches as well. cockroaches (n = 4) were mostly uncoordinated. The dominant interleg coordination pattern during Although the locomotion of SEG-less cockroaches was immersed locomotion in control cockroaches (n =4) generally less rhythmic than that of controls, as men- was the ‘alternating tripod-gait’ typical of cockroach tioned above, the lack of interleg coordination appeared terrestrial and immersed locomotion (Hoyle 1976; also where rhythmic locomotion did occur. For exam- Cocatre-Zilgien and Delcomyn 1990) (Figs. 6b, 7). The ple, a few cycles of rhythmic activity often appeared in right mesothoracic (R2) and left metathoracic (L3) legs one or more legs, but the rhythm in one leg was seldom moved synchronously, and the average phase of the temporally correlated with the others. In several cases, right mesothoracic depressor burst in the left metatho- however, two legs moved together or alternatively for a racic cycle (R2–L3 phase) was ¡0.01, the negative sign few cycles, and in rare cases tripod-gait locomotion was indicating that L3 started moving before R2 did. The also observed. The average R2–L3 phase in SEG-less

123 J Comp Physiol A (2006) 192:1003–1020 1013

Control 30 -0.01 0.51 slow coxal depressors (Ds) potentials. Typically, during fast terrestrial walking or running (above 10 cycles/s), 25 these slow potentials are accompanied by fast depressor 20 potentials (Df) which are indicative of strong leg excur- 15 sions. Df potentials appeared readily during terrestrial running in control cockroaches, and, when induced to 10 escape with a tactile stimulus, sometimes also in brain- Occurrence ( % ) 5 less cockroaches. In contrast, Df potentials rarely 0 appeared in SEG-less cockroaches on land. Interest- ingly however, in all cockroaches, Df potentials often appeared at the end of the Ds burst during immersed -Br 12 0.03 0.41 locomotion (Fig. 6c). This was in spite of the relatively 10 slow stepping frequencies (below 5 cycles/s), at which 8 Df potentials never occur during terrestrial locomotion.

4 Discussion

Occurrence ( % ) 2 Research in vertebrates and insects suggests that 0 inputs from “higher-order” centers are crucial for the selection, coordination and modulation of motor patterns generated in the “lower” portions of the central -SEG 12 -0.29 0.29 nervous system (Kien et al. 1992; Deliagina et al. 2002; 10 Strauss 2002; Briggman et al. 2005; Grillner et al. 2005; Ridgel and Ritzmann 2005). In insects, speciWc neuro- 8 nal lesions are often employed to study the eVect of 6 higher neuronal centers (brain and SEG) on motor behaviors generated in the thorax. Thus far, lesion 4 experiments have consisted of either cutting the CirC

Occurrence ( % ) 2 between the brain and SEG to remove descending input from the brain, or cutting the neck connectives 0 –1.0 –0.5 0.0 0.5 1.0 between the SEG and thorax to remove inputs from Phase both head ganglia and produce “headless” animals (see, for example, Ridgel and Ritzmann 2005 and ref- Fig. 7 Interleg coordination during immersed locomotion: relative occurrence of phases of the right mesothoracic (black histo- erences therein). Owing to the anatomical location of grams) and metathoracic (gray histograms) Ds bursts in the cycle the SEG between the brain and thoracic ganglia, its of the left metathoracic Ds burst. The median phases are also role in modulating thoracic PGs was interpreted from indicated. Note the lack of tripod-gait coordination in SEG-less the diVerential examination of behavioral deWcits after cockroaches (-SEG). -Br: brainless cockroaches CirC lesions and after neck lesions. Here, we have employed an experimental procedure (Matsumoto cockroaches was ¡0.41 and the average R3–L3 phase and Sakai 2000) that can remove descending input was 0.35. There was a rather large distribution around from the SEG without disturbing descending input the average phases, indicative of the uncoordinated from the brain. The investigations presented here of locomotion. The histogram for SEG-less coordination the interactions of each of the two “higher-order” neu- revealed small peaks around 0.0 for both R2–L3 and ronal centers with thoracic circuits and sensory feed- R3–L3 phase relationships (Fig. 7). This is due to fre- back provide a functional model for the integration of quent simultaneous extensions of two or more legs, or these functional units to generate the appropriate ‘jumps’, which were also observed visually. This move- motor output. The proposed model is given in Fig. 8 ment resembled the typical response of SEG-less cock- and some of the evidence contributing to this model roaches to a tactile stimulus on land (Fig. 6a), although is summarized in Table 2. We have found that (a) in water it was initiated spontaneously (Fig. 6b). the SEG and the brain provide antagonistic and In our experiments, EMG recordings indicated fem- opposite input to the Xight and walking PGs; (b) the oral extensions during stance as tonic activity of the SEG is necessary for generating the pattern of 123 1014 J Comp Physiol A (2006) 192:1003–1020

Table 2 Summary of the relative contribution of the brain and The SEG-less preparation SEG to the initiation and maintenance of thoracic motor behaviors in the intact cockroach To conWrm that a mid-sagittal slit in the SEG elimi- Motor behavior Brain SEG nates most SEG-DINs while leaving intact the B- DINs projecting through the SEG to the thorax, we Spontaneous walking ¡ + Walking coordination 0 + performed anatomical and physiological studies in Flight initiation 0 0 control and split-SEG cockroaches. Using backWlls Flight maintenance + ¡ from the neck connective in control cockroaches, we Flight coordination ? 0 counted roughly 200 somata in the brain and SEG Escape initiation (‘startle’) 0 0 Escape maintenance (run) 0 + which project to the thoracic ganglia. This is higher Swimming initiation 0 0 than the number of descending interneurons esti- Swimming maintenance 0 + mated by Burdohan and Comer (1996) and lower Righting initiation 0 0 than that of Okada et al. (2003), but these discrepan- Righting maintenance + + cies are almost certainly due to diVerent staining + excitatory or permissive eVect, ¡ inhibitory eVect, 0 no eVect methods and bear little consequence on the rationale of our experiments. We demonstrate that a mid-sagittal section through walking-related PGs and (c) emergency behaviors are the SEG does not signiWcantly damage, anatomically initiated in the thorax, but require the head ganglia for or physiologically, axons of B-DINs that project to successful completion. the thorax (Figs. 1, 2). In contrast, we consider the

a SENS Water Tactile/Wind Legs in the air Wings on the ground

124 6 3 5

Swimming Escape Righting Flight

THX

7 Walking PG Flight PG

8 9 10 11 12

SEG HEAD Brain

Fig. 8 A simpliWed functional model for the initiation (a) and the interactions between diVerent sensory modalities locally pro- maintenance (b) of selected motor behaviors in the cockroach. mote the initiation of a speciWc motor behavior while inhibiting Arrowheads indicate a net excitatory eVect; circles indicate a net contradicting behaviors. Furthermore, walking and Xight, two inhibitory eVect; diamond indicates a permissive eVect. a DiVer- incompatible behaviors, are reciprocally inhibited at the thoracic ent sensory cues (SENS) excite or inhibit diVerent motor patterns level (7). b After a motor behavior has been initiated, the head in the thorax (THX) reXexively, to initiate an appropriate motor ganglia (HEAD) modulate the maintenance of the appropriate behavior. When the cockroach is immersed in water, swimming is pattern generators (PG). Moreover, the inXuence of the brain initiated due to water contact (1); when a tactile or wind stimulus and SEG on walking and Xight is antagonistic, as the brain inhib- is applied to the cerci, both escape (2) and Xight (3) mechanisms its walking (8) and excites Xight (12), whereas the SEG inhibits are excited; when the cockroach legs are in the air, both righting Xight (11), excites walking (9) and contributes to interleg coordi- (4) and Xight (5) mechanisms are excited; when sensilla on the nation during walking-related behaviors (10). See text for details wings are in contact with a substrate, Xight is inhibited (6). Thus, 123 J Comp Physiol A (2006) 192:1003–1020 1015 removal of SEG-DINs’ input to the thorax by the pattern. Since cockroaches are chieXy terrestrial insects, mid-sagittal slit to be quite substantial. This is they are capable of expressing a relatively wide reper- because over 70% of the SEG-DINs cross over at the toire of terrestrial locomotory patterns. However, it has SEG midline to send their axons to the thoracic gan- previously been demonstrated that many of these pat- glia (Fig. 1). Moreover, most SEG-DINs have wide- terns are essentially variations of the same PG, to which spread dendritic arborizations in the SEG neuropile we will henceforth refer as the “walking PG”. For exam- ipsilateral to the soma or proximal to the SEG mid- ple, although running involves fast motor-neurons (e.g., line (Altman and Kien 1987; Kien et al. 1990). These the fast coxal depressors, Df) that produce the strong and arborizations must be, in large part, disrupted in split- fast muscle contractions required for running, it is consid- SEG preparations. Hence, the structural integrity of ered a modiWcation of the walking pattern although it is most of the remaining ipsilateral SEG-DINs, as well somewhat more rigid and less compliant to sensory feed- as the physiological integrity of the contralateral back (see Delcomyn 1991). Likewise, a similar coordina- SEG-DINs, is also largely impaired. We are therefore tion pattern is expressed, and similar muscles and motor- conWdent that a mid-sagittal slit in the SEG is a fairly neurons are incorporated during swimming and walking reliable procedure to generate “SEG-less” cock- behaviors (Figs. 6, 7; see also Cocatre-Zilgien and Delc- roaches. This provides us with a tool to study the indi- omyn 1990). Another example of a walking-related vidual inXuence of the SEG on motor behaviors behavior is righting behavior, which is expressed as sen- generated in the thoracic ganglia. sory inputs are altered when the cockroach is overturned In the locust, approximately 30% of all B-DINs ter- (Reingold and Camhi 1977; Sherman et al. 1977; Zill minate in the SEG (Heinrich 2002). In our experi- 1986). The notion that running, swimming and righting ments, the indirect eVect of such interneurons on are essentially variations of the walking pattern is also thoracic circuitries is similarly eliminated in brainless supported by our results: this is because similar manipu- and SEG-less cockroaches. This is the case also for lations of the cockroach CNS aVect the initiation and interneurons ascending from the SEG to the brain. maintenance of these three motor outputs in a similar Moreover, it is only when the behavioral changes of manner. The Xight behavior, for comparison, is aVected brainless are similar to those of SEG-less cockroaches quite diVerently. that such deWcits might be attributed to the removal of this brain-SEG loop per se. It is thus parsimonious to The initiation of ‘emergency’ behaviors assume that this loop has a limited eVect compared with that of the true descending projections. By removing descending inputs from the brain or SEG or both, we show that ‘emergency’ behaviors are initi- Thoracic pattern generators in the cockroach ated in the thoracic circuitry, even in the absence of the head ganglia. We refer to ‘emergency’ behaviors as DiVerent motor patterns that involve similar neuronal those crucial for the survival of the cockroach in its nat- networks and muscles are generally thought to be con- ural environment, and require fast and strong reac- trolled by a common central PG (Reingold and Camhi tions. The ‘emergency’ behaviors tested in this work 1977; Sherman et al. 1977; Marder et al. 2005). Thus, a are escape, Xight, swimming and righting. central PG can act as a multifunctional unit, the appro- We show that all these ‘emergency’ behaviors are priate output of which is expressed due to diVerential initiated at the thoracic level regardless of head ganglia neuromodulatory central and peripheral inputs (Marder input, and that the ‘emergency’ stimuli locally increase and Bucher 2001; Marder et al. 2005). These inputs are in the excitatory drive to the appropriate PG. By being large part the result of descending modulation by initiated at the thoracic level, the ‘emergency’ behavior “higher-order” neuronal centers (Kien and Altman can begin more quickly and the PG is already primed 1992a) or sensory modulation (Büschges 2005) or, most when input arrives from the head ganglia to organize commonly, of both. In insects, many of the sensory struc- and maintain the behavior. However, the head ganglia tures that are known to extensively modulate the expres- are necessary for successful completion of the emer- sion pattern of a given locomotory central PG reside on gency behavior. First, all operated cockroaches show a the legs and wings (Frye and Dickinson 2004; Noah form of ‘startle’ (see Camhi 1984; Schaefer et al. 1994) et al. 2004; Pearson 2004; Zill et al. 2004; Büschges in response to a tactile stimulus applied to the abdo- 2005), while the “higher-order” centers reside in the men (Fig. 5b). Our results are consistent with previous head ganglia. In the present work, we examined the eVect studies that also investigated the spatio-temporal orga- of “higher-order” centers on two distinct behavioral pat- nization of the startle response in headless cockroaches terns in the cockroach: the Xight pattern and the walking (Schaefer and Ritzmann 2001), and suggest that tactile 123 1016 J Comp Physiol A (2006) 192:1003–1020 or wind stimuli initiate escape at the thoracic level response. An alternative to this explanation is that dur- (Fig. 8: 2). Similarly, we demonstrate that if a wind ing leg retraction, the high viscosity of the water stimulus is applied to the cerci of a headless cockroach medium relative to air enlarges an error signal from leg when its legs are not in contact with a substrate, a short sensors within each moving leg. This might, in turn, Xight sequence is initiated (Fig. 3b, c). The threshold result in stronger muscle contraction to decrease the wind velocity required to elicit Xight is unchanged after ‘error signal’ (see Cruse et al. 2004). An interesting removing descending input from the SEG or from both observation, however, suggests that the former expla- head ganglia, although it was markedly elevated after nation is more probable than the latter: allowing cock- removing descending inputs from the brain (Fig. 3a). roaches to swim more than a few minutes typically This suggests that the head ganglia are not necessary results in a period of uncoordinated leg movements, for the initiation of Xight (Fig. 8: 3), although the brain after which the cockroach usually dies (see Materials seems to exert an excitatory eVect on the Xight PG, and methods). These leg movements are characterized which will be discussed separately. The neuronal path- by high frequency ‘stepping’ cycles and short leg exten- way by which a tactile or wind stimulus is able to initi- sions. Nevertheless, Df potentials often accompany ate escape or Xight is well documented (Camhi and these movements in control or operated cockroaches, Levi 1988). Moreover, since in all operated cock- although the legs are usually stretched to only one- roaches Xight and leg movements never occur simulta- third of their typical extent. This supports our neously, an inhibitory pathway between the Xight and hypothesis, and suggests that the extreme emergency walking PGs must exist at the thoracic level and is raises neuronal excitability in the thorax to allow Df independent of head ganglia input (Fig. 8: 7). recruitment. Second, operated cockroaches spontaneously initi- Third, we show that when overturned, control, head- ate swimming when they are immersed in water less, brainless and 40% of the SEG-less cockroaches (Fig. 6), suggesting that the head ganglia are not neces- initiate righting, i.e. struggle by producing Xailing move- sary to initiate swimming (Fig. 8: 1). We consider swim- ments with the legs (Table 1). Thus, righting is another ming as an ‘emergency’ behavior since if an immersed ‘emergency’ behavior that is initiated reXexively, i.e. in cockroach is allowed to move freely, it will try to climb the absence of head ganglia descending inputs. Previous out of the water and will not willingly return to water. studies of headless cockroaches support this conclusion Furthermore, cockroaches are usually unable to sur- (Sherman et al. 1977; Zill 1986). What are the signals vive in water for more than a few minutes, probably required to initiate righting and where are the sensory because water Xoods the spiracles and tracheae. The structures located? We show that it is the integration of stimulus which evokes swimming is yet to be identiWed, two sensory cues which evokes righting. When the but the decrease in load characteristic of water immer- cockroach is overturned with the legs in the air, sensory sion or the stimulation of water-sensitive sensilla are feedback from sensilla on the legs is modiWed as a result likely candidates. A prominent diVerence between of a load decrease. However, when an overturned cock- swimming and terrestrial locomotion is that during ter- roach clings to an inverted surface, a natural behavior restrial locomotion the fast coxal depressor motor- of cockroaches, righting is not evoked (see, for exam- neurons (Df) are recruited only when the stepping fre- ple, Larsen et al. 1995). In our study, all control and quency exceeds 10 cycles/s (Pearson 1972; Delcomyn operated cockroaches terminated righting when a con- and Usherwood 1973), but these potentials regularly tact between the legs and substrate could be established occur during immersed locomotion at a low ‘stepping’ (Table 1). This observation, together with previous frequency of about 4 cycles/s (Fig. 6c; Cocatre-Zilgien studies (Reingold and Camhi 1977; Sherman et al. 1977; and Delcomyn 1990). This occurs not only in control, Zill 1986), demonstrate that “legs in the air” is one cue but also in brainless and in SEG-less cockroaches, that can initiate righting, and consequently modify the which suggests that the head ganglia are not necessary output of the walking PG to produce righting move- for the recruitment of fast motor neurons during loco- ments. Also, this mechanism is obviously independent motion. Supporting this notion, the study of Zill (1986) of head ganglia descending inputs (Fig. 8: 4). However, demonstrated that Df potentials are recruited during the “legs in the air” cue is also known to elicit Xight ‘walking attempts’ in headless cockroaches. Why are behavior (Ritzmann et al. 1980) (Fig. 8: 5). Hence, in an Df recruited during low frequency locomotion in water overturned cockroach, a second mechanism must act in but not on land? We suggest that water immersion is an concert with the “legs in the air” cue to favor righting ‘emergency’ stimulus that, similar to other such stimuli, over Xight. Since removing the wings promotes the increases thoracic excitability to allow the recruitment selection of Xight rather than righting in overturned of fast motor-neurons and thereby a faster, stronger cockroaches (Table 1), it seems that sensilla on the 123 J Comp Physiol A (2006) 192:1003–1020 1017 wings participate in reporting the overturned posture (i.e., the pathway described in Fig. 8: 6), thus producing (Fig. 8: 6). Insects’ wings bear mechanoreceptive a ‘behavioral switch’ in SEG-less cockroaches. sensilla (Dickinson et al. 1997; Page and Matheson 2004), which are likely candidates to participate in this Descending inXuence on walking-related behaviors process. Since, as discussed above, an inhibitory pathway apparently exists between the Xight PG and Removing descending inputs from the brain prolongs the walking PG, wing sensilla could either inhibit Xight spontaneous walking (Fig. 5a). This is in agreement (as illustrated in Fig. 8: 6) or excite the walking PG with previous studies (Roeder 1937; Pearson and Iles (not shown) to favor righting over Xight. Further 1973; Graham 1979; Bässler 1983; Ridgel and Ritzmann research is required to discriminate between these two 2005 and references therein) and suggests the brain to possibilities. be a continuous source of inhibition on the walking PG Taken together, we demonstrate that speciWc stimuli (Fig. 8: 8). Thus, spontaneous walking requires release can increase the excitatory drive to the appropriate from brain inhibition, a neuronal mechanism that has thoracic PG locally, to initiate a reXexive ‘emergency’ been proposed for numerous behaviors in insects (Roe- response (Fig. 8a). This mechanism probably evolved der et al. 1960; Dominick and Truman 1986; Heinrich in order to produce faster motor responses in the case et al. 1998; Matsumoto and Sakai 2000), as well as ver- of an immediate danger. Nevertheless, our results also tebrates (Grillner 2003). Furthermore, our results sug- demonstrate that descending inputs from the head gan- gest that the brain has a weak role in the control of glia are crucial for the proper maintenance of rhythmic interleg coordination. This is because brain removal has motor patterns that follow the initial reXexive little, if any, eVect on tripod-gait coordination in all responses. walking-related behaviors tested (Figs. 5b, 6, 7), though it mildly increases the variability in interleg phase rela- The eVect of the brain and SEG on thoracic motor tionship, suggesting that the ‘Wne tuning’ of the interleg patterns coordination may be controlled by the brain which integrates visual information. In contrast to other walking- Descending inXuence on Xight related behaviors, the righting ability seems to be impaired in brainless, as well as in all other operated By applying wind puVs of diVerent velocities to wind- cockroaches. A possible explanation for this deWcit is sensitive structures on the cerci (Ritzmann et al. 1982), that the overall force generated by the legs may have we show that Xight is abolished in brainless cock- been reduced as a result of the ganglionic lesion. It has roaches (Fig. 3). This suggests a permissive eVect of the previously been demonstrated that the force that each brain on the thoracic Xight PG (Fig. 8: 12). Conversely, single leg expresses is signiWcantly larger during righting removing the SEG markedly prolong Xight duration than, for example, during running (Full et al. 1995). It is without aVecting inter-wing coordination (Figs. 3, 4). therefore possible that although brainless or SEG-less Thus, in the intact cockroach, the SEG appears to exert a cockroaches can recruit fast potentials necessary for continuous inhibitory eVect on the Xight PG (Fig. 8: 11). righting, they are unable to recruit enough motor units This notion is further supported by two observations: to complete righting since removing the brain, the SEG (1) headless cockroaches, in which descending input or the brain-SEG relay decreases the overall excitatory from both the brain and SEG is removed, can initiate input to the thorax. Xight but express very brief Xight episodes (Fig. 3); (2) In marked contrast to brain removal, removing when SEG-less cockroaches are overturned, Xight is descending inputs from the SEG produces hypokinetic initiated in 60% of the trials, spontaneously or follow- cockroaches and almost completely abolishes the initi- ing a wind stimulus, instead of righting (Table 1). ation of spontaneous locomotion (Fig. 5a). Further- Hence, it appears that the Xight PG is additively inhib- more, when ‘emergency’ stimuli that normally evoke a ited by at least two neuronal pathways: descending walking-related behavior are applied to SEG-less cock- inputs from the SEG, and sensory inputs from sensilla roaches, a motor response is initiated but the cock- on the legs which have been shown to inhibit Xight roaches fail to express the subsequent motor pattern. when contact with a substrate is maintained (Ritzmann For example, when an SEG-less cockroach is provoked et al. 1980). We suggest that in an overturned SEG-less to escape with a tactile stimulus or water immersion, an cockroach, the Xight PG is disinhibited due to SEG atypical motor response is evoked instead of the nor- removal together with the absence of leg-substrate mal tripod-gait coordination (Figs. 6, 7). Thus, apart contact. This disinhibition is evidently stronger than from exerting an overall continuous excitatory inXu- the inhibition of Xight exerted by sensilla on the wings ence on the walking PG (Fig. 8: 9), descending inputs 123 1018 J Comp Physiol A (2006) 192:1003–1020 from the SEG appear to be required also for the generating centers in the “lower” parts of the CNS expression of appropriate interleg coordination (Fig. 8: have been selected more than once during evolution. 10). Similar results have been reported for headless Nevertheless, accumulating evidence suggests that animals (Roeder 1937, 1963; Hughes 1965; Reingold these are widespread motifs throughout the animal and Camhi 1977; Graham 1979; Bässler 1983; Zill 1986; kingdom, from worms to mammals. Kien 1990a, b; Bohm and Schildberger 1992; Keegan and Comer 1993; Strauss and Heisenberg 1993; Whelan Acknowledgements We thank Gustavo Glusman for his valu- 1996; Libersat et al. 1999; Schaefer and Ritzmann 2001; able technical assistance. We also thank Aviva Weisel-Eichler, Lior Ann Rosenberg, Yael Lavi, Gal Haspel and two anonymous Ridgel and Ritzmann 2005; Cornford et al. 2006). A reviewers for valuable comments. This work was supported by likely explanation for the crucial role of the SEG in Grant 2001044 from the United States-Israel Binational Science coordinating walking was proposed by Kien and Foundation (BSF). These experiments comply with the “Princi- Altman (1984): by recording from SEG neurons, they ples of animal care”, publication No. 86–23, revised 1985 of the National Institute of Health, and also with the current laws of the demonstrate that the SEG is possibly involved in a State of Israel. positive feedback loop required to maintain walking. In our study, by observing cockroaches in which SEG descending inputs were removed selectively, we provide direct evidence to support this notion. References

Altman JS, Kien J (1987) Functional organization of the subesophageal ganglion in arthropods. In: Gupta A (ed) Conclusion Arthropod brain: its evolution, development, structure, and functions. Wiley, New York, pp 265–301 In the present work, we have examined the net eVect of Arshavsky YI, Deliagina TG, Orlovsky GN (1997) Pattern gener- V ation. Curr Opin Neurobiol 7(6):781–789 descending inputs from the brain and SEG on di erent Bässler U (1983) Neural basis of elementary behavior in stick in- thoracic motor pattern generators. Our results suggest sects (studies of brain function). Springer, Berlin Heidelberg that descending control over diVerent motor patterns is New York anatomically and functionally compartmentalized. Yet, Bässler U, Büschges A (1998) Pattern generation for stick insect walking movements—multisensory control of a locomotor our results do not imply that the head ganglia are program. Brain Res Rev 27:65–88 merely ‘on/oV switches’ for thoracic PGs where each Berkowitz A, Laurent G (1996) Local control of leg movements head ganglion provides pure excitatory or inhibitory and motor patterns during grooming in locusts. J Neurosci descending input to a given thoracic PG. It is therefore 16:8067–8078 Bohm H, Schildberger K (1992) Brain neurons involved in the important to clarify that our experiments reveal only control of walking in the cricket Gryllus bimaculatus. J Exp the last link in a long and highly complex chain of neu- Biol 166:113–130 ronal events that involves sensory integration, memory Briggman KL, Abarbanel HDI, Kristan WB (2005) Optical imag- processes, arousal levels and other processes occurring ing of neuronal populations during decision-making. Science 307(5711):896–901 in the head ganglia. How these important factors con- Bullock TH, Horridge GA (1965) Structure and function in the tribute to the selection, initiation and modulation of nervous systems of invertebrates. W. H. Freeman and Co, thoracic PGs is beyond the scope of the present study San Francisco and needs to be addressed with complementary experi- Burdohan JA, Comer CM (1996) Cellular organization of an antennal mechanosensory pathway in the cockroach, Peri- mental approaches such as single and multi-unit planeta americana. J Neurosci 16(18):5830–5843 recordings of neurons in the head ganglia combined Büschges A (2005) Sensory control and organization of neural with detailed kinematic analysis of motor behaviors. networks mediating coordination of multisegmental organs The anatomical and functional partition of the for locomotion. J Neurophysiol 93:1127–1135 Camhi JM (1984) Neuroethology: nerve cells and natural behav- “higher order” neuronal centers that we report here is ior of animals. Chapter 4: A case study in neuroethology: the not an exclusive feature of insects. For instance, Bull- escape system of the cockroach. Sinauer Associates Inc., ock and Horridge (1965) and, more recently Cornford Sunderland, pp 79–105 et al. (2006), have shown that, in leeches, removing the Camhi JM, Levi A (1988) Organization of a complex movement: Wxed and variable components of the cockroach escape brain causes increased activity, whereas removing the behavior. J Comp Physiol A 163(3):317–328 SEG depresses spontaneous activity, although the Carbonell CS (1947) The thoracic muscles of the cockroach Peri- leeches could still crawl and swim if prodded suY- planeta americana. Smithson Misc Collect 107:1–23 ciently. It is unclear whether the anatomical compart- Cocatre-Zilgien JH, Delcomyn F (1990) Fast axon activity and the motor pattern in cockroach legs during swimming. Physiol mentalization of the centers responsible for Entomol 15(4):385–392 maintaining behaviors into antagonistic “higher-order” Cornford A, Kristan WB, Malnove S, Kristan WBJ, French KA units in the head, and their partition from the rhythm- (2006) Functions of the subesophageal ganglion in the 123 J Comp Physiol A (2006) 192:1003–1020 1019

medicinal leech revealed by ablation of neuromeres in em- Keegan AP, Comer CM (1993) The wind-elicited escape re- bryos. J Exp Biol 209:493–503 sponse of cockroaches (Periplaneta americana) is inXuenced Cruse H, Kühn S, Park S, Schmitz J (2004) Adaptive control for by lesions rostral to the escape circuit. Brain Res 620:310– insect leg position: controller properties depend on substrate 316 compliance. J Comp Physiol A 190:983–991 Kien J (1983) The initiation and maintenance of walking in the lo- Delcomyn F (1991) Perturbation of the motor system in freely cust: an alternative to the command concept. Proc R Soc walking cockroaches. II. The timing of motor activity in leg Lond B Biol 219:137–174 muscles after amputation of a middle leg. J Exp Biol Kien J (1990a) Neuronal activity during spontaneous walking. I. 156:503–517 Starting and stopping. Comp Biochem Physiol A 95:607–621 Delcomyn F, Usherwood PNR (1973) Motor activity during Kien J (1990b) Neuronal activity during spontaneous walking. II. walking in the cockroach Periplaneta americana. I. Free Correlation with stepping. Comp Biochem Physiol A 95:623– walking. J Exp Biol 59:629–942 638 Deliagina TG, Zelenin PV, Grigori N, Orlovsky GN (2002) Kien J, Altman JS (1984) Descending interneurones from the Encoding and decoding of reticulospinal commands. Brain brain and suboesophageal ganglia and their role in control of Res Rev 40:166–177 locust behaviour. J Insect Physiol 30(1):59–72 Dickinson MH, Hannaford S, Palka J (1997) The evolution of in- Kien J, Altman JS (1992a) Decision-making in the insect nervous sect wings and their sensory apparatus. Brain Behav Evol system: a model for selection and maintenance of motor pro- 50(1):13–24 grammes. In: Kien J, McCrohan CR, Winlow W (eds) Neu- Dickinson MH, Farley CT, Full RJ, Koehl MAR, Kram R, robiology of motor programme selection. Pergamon Press, Lehman S (2000) How animals move: an integrative view. Oxford. Pergamon studies in neuroscience, no. 4, pp 147–169 Science 288(5463):100–106 Kien J, Altman JS (1992b) Preparation and execution of move- Dietz V (2003) Spinal cord pattern generators for locomotion. ment: parallels between insect and mammalian motor sys- Clin Neurophysiol 114(8):1379–1389 tems. Comp Biochem Physiol A 103(1):15–24 Dominick OS, Truman JW (1986) The physiology of wandering Kien J, Fletcher JS, Altman J, Ramirez M, Roth U (1990) Orga- behavior in Manduca sexta. III. Organization of wandering nization of intersegmental interneurons in the suboesopha- behavior in the larval nervous system. J Exp Biol 121:115–132 geal ganglion of Schistocerca gregaria (Forskal) and Locusta Duysens J, Clarac F, Cruse H (2000) Load-regulating mecha- migratoria migratorioides (Reiche & Fairmaire) (Acrididae, nisms in gait and posture: comparative aspects. Physiol Rev Orthoptera). Int J Insect Morphol Embryol 19(1):35–60 80(1):83–133 Kien J, McCrohan CR, Winlow W (eds) (1992) Neurobiology of Eaton RC, Farley RD (1969) The neural control of cercal groom- motor programme selection. Pergamon Press, Oxford. Perg- ing behaviour in the cockroach, Periplaneta americana. amon studies in neuroscience, no. 4 J Insect Physiol 15:1047–1065 Knop G, Denzer L, Buschages A (2001) A central pattern-gener- Frye MA, Dickinson MH (2004) Closing the loop between neuro- ating network contributes to “reXex-reversal”-like leg moto- biology and Xight behavior in Drosophila. Curr Opin Neuro- neuron activity in the locust. . J Neurophysiol 86(6):3065– biol 14(6):729–736 3068 Full RJ, Yamauchi A, Jindrich DL (1995) Maximum single leg Larsen GS, Frazier SF, Fish SE, Zill SN (1995) EVects of load force production: cockroaches righting on photoelastic gela- inversion in cockroach walking. J Comp Physiol A tin. J Exp Biol 198:2441–2452 176(2):229–238 Graham D (1979) EVects of circum-oesophageal lesion on the Libersat F (1994) The dorsal giant interneurons mediate evasive behavior of the stick insect Carausius morosus. II. Changes behavior in Xying cockroaches. J Exp Biol 197:405–411 in walking co-ordination. Biol Cybern 32:147–152 Libersat F, PXueger HJ (2004) Monoamines and the orchestra- Grillner S (2003) The motor infrastructure: from ion channels to tion of behavior. Bioscience 54(1):17–25 neuronal networks. Nat Rev Neurosci 4:573–586 Libersat F, Haspel G, Casagrand J, Fouad K (1999) Localization Grillner S, Wallen P (2002) Cellular bases of a vertebrate locomo- of the site of eVect of a wasp’s venom in the cockroach escape tor system-steering, intersegmental and segmental co-ordi- circuitry. J Comp Physiol A 184:333–345 nation and sensory control. Brain Res Rev 40:92–106 Marder E, Bucher D (2001) Central pattern generators and the Grillner S, Hellgren J, Ménard A, Saitoh K, Wikström MA (2005) control of rhythmic movements. Curr Biol 11:986–996 Mechanisms for selection of basic motor programs—roles Marder E, Bucher D, Schulz DJ, Taylor AL (2005) Invertebrate for the striatum and pallidum, Trends Neurosci 28:364–370 central pattern generation moves along. Curr Biol Heinrich R (2002) Impact of descending brain neurons on the 15(17):685–99 control of stridulation, walking, and Xight in orthoptera. Matheson T (1997) Hindleg targeting during scratching in the Microsc Res Tech 56(4):292–301 locust. J Exp Biol 200(1):93–100 Heinrich R, Rozwod K, Elsner N (1998) Neuropharmacological Matsumoto Y, Sakai M (2000) Brain control of mating behavior in evidence for inhibitory cephalic control mechanisms of strid- the male cricket Gryllus bimaculatus DeGeer: the center for ulatory behaviour in grasshoppers. J Comp Physiol A inhibition of copulation actions. J Insect Physiol 46(4):527–538 183:389–399 Noah JA, Quimby L, Frazier SF, Zill SN (2004) Walking on a ‘peg Hoyle G (1976) Arthropod walking. In: Herman RM, Grillner S, leg’: extensor muscle activities and sensory feedback after Stein PSG, Stuart DG (eds) Neural control of locomotion. distal leg denervation in cockroaches. J Comp Physiol A Plenum Press, New York, pp 137–179 190:217–231 Hughes GM (1965) Neuronal pathways in the insect central ner- Okada R, Sakura M, Mizunami M (2003) Distribution of den- vous system. In: Treherne JE, Beaments JWL (eds) The drites of descending neurons and its implications for the ba- physiology of the insect central nervous system. Academic, sic organization of the cockroach brain. J Comp Neurol New York, pp 79–112 458(2):158–174 Johnston RM, Consoulas C, PXueger H, Levine RB (1999) Pat- Page KL, Matheson T (2004) Wing hair sensilla underlying aimed terned activation of unpaired median neurons during Wctive hindleg scratching of the locust. J Exp Biol 207(15):2691– crawling in Manduca sexta larvae. J Exp Biol 202(2):103–113 2703

123 1020 J Comp Physiol A (2006) 192:1003–1020

Pearson KG (1972) Central programming and reXex control of Schaefer P, Ritzmann RE (2001) Descending inXuences on es- walking in the cockroach. J Exp Biol 56:173–193 cape behavior and motor pattern in the cockroach. J Neuro- Pearson KG (2004) Generating the walking gait: role of sensory biol 49(1):9–28 feedback. Prog Brain Res 143:123–129 Schaefer PL, Kondagunta GV, Ritzmann RE (1994) Motion anal- Pearson KG, Iles JF (1970) Discharge patterns of coxal levator ysis of escape movements evoked by tactile stimulation in the and depressor motoneurones of the cockroach, Periplaneta cockroach Periplaneta americana. J Exp Biol 190:287–294 americana. J Exp Biol 52:139–165 Sherman E, Novotny M, Camhi J (1977) A modiWed walking Pearson KG, Iles JF (1973) Nervous mechanisms underlying in- rhythm employed during righting behavior in the cockroach tersegmental co-ordination of leg movements during walk- Gromphadorhina portentosa. J Comp Physiol A 113:303–316 ing in the cockroach. J Exp Biol 58:725–744 Strausfeld NJ (1999) A brain region in insects that supervises Reingold SC, Camhi JM (1977) A quantitative analysis of rhythmic walking. Prog Brain Res 123:273–84 leg movements during three diVerent behaviors in the cock- Strauss R (2002) The central complex and the genetic dissection roach, Periplaneta americana. J Insect Physiol 23:1407–1420 of locomotor behaviour. Curr Opin Neurobiol 12(6):633–638 Ridgel AL, Ritzmann RE (2005) EVects of neck and circumo- Strauss R, Heisenberg M (1993) A higher control center of loco- esophageal connective lesions on posture and locomotion in motor behavior in the Drosophila brain. J Neurosci 13:1852– the cockroach. J Comp Physiol A 191(6):559–573 1861 Ritzmann RE, Tobias ML, Fourtner CR (1980) Flight activity ini- Svidersky VL, Plotnikova SI (2002) Insects and vertebrates: anal- tiated via giant interneurons of the cockroach: evidence for ogous structures in higher integrative centers of the brain. bifunctional trigger interneurons. Science 210:443–445 J Evol Biochem Phys 38(5):627–639 Ritzmann RE, Pollack AJ, Tobias ML (1982) Flight activity med- Watson JT, Ritzmann RE (1998) Leg kinematics and muscle iated by intracellular stimulation of dorsal giant interneurons activity during treadmill running in the cockroach, Blaberus of the cockroach Periplaneta americana. J Comp Physiol A discoidalis: I. Slow running. J Comp Physiol A 182(1):11–22 147:313–322 Whelan PJ (1996) Control of locomotion in the decerebrate cat. Roeder KD (1937) The control of tonus and locomotor activity in Prog Neurobiol 49:481–515 the praying mantis, Mantis religiosa L. J Exp Zool 76:353–374 Wilson DM (1961) The central nervous control of Xight in a lo- Roeder KD (1948) Organization of the ascending giant Wber sys- cust. J Exp Biol 38:471–490 tem in the cockroach (Periplaneta americana). J Exp Zool Wilson DM (1967) Central nervous mechanisms for the genera- 108:243–261 tion of rhythmic behavior in arthropods. Symp Soc Exp Biol Roeder KD (1963) Nerve cells and insect behavior. Harvard Uni- 47:133–151 versity Press, Cambridge, p 238 Zill SN (1986) A model of pattern generation of cockroach walk- Roeder KD, Tozian L, Weiant E (1960) Endogenous nerve activity ing reconsidered. J Neurobiol 17(4):317–328 and behaviour in the mantis and cockroach. J Insect Physiol Zill SN, Schmitz J, Büschges A (2004) Load sensing and control of 4:45–62 posture and locomotion. Arthropod Struct Dev 33:273–286

Ram Gal1,2 and Frederic Libersat3,* a forward jump immediately followed by one or two steps 1Zlotowski Center for Neuroscience backward, i.e., in the direction of the stimulus (data not 2Department of Life Sciences shown). Such behavior was never observed in the control Ben-Gurion University of the Negev group. None of the observed behavioral deficits could be P.O. Box 653 explained by direct sensory or motor deficits (Figure S1 and Be’er Sheva 84105 see also [2]). Israel 3Institut de Neurobiologie de la Me´ diterrane´ e Wasp Venom Elevates the Threshold for Walking Initiation Parc Scientifique de Luminy Are stung cockroaches unable to initiate walking, or is their 163 route de Luminy, BP13 behavioral threshold for walking elevated? To answer this 13273 Marseille cedex 09 question, we applied either controlled noxious or repetitive France stimuli (Figure 1). For noxious stimuli, we used a modified Shuttle Box [8] in which we administered escapable foot shocks to the cockroach’s legs, without previous training. Summary The box comprised a shocking chamber and a neutral chamber. We recorded the threshold voltage required to elicit (1) The parasitoid wasp A. compressa hunts cockroaches as leg contractions without subsequent locomotion (i.e., the a live food supply for its offspring. The wasp selectively in- ‘‘nociceptive’’ threshold) and (2) walking to the neutral cham- jects venom into the cerebral ganglia of the prey to induce ber, a distance of 5 cm (based on [8, 9]). We recorded these long-term hypokinesia [1–5], during which the stung cock- two values before (time point 0 in Figure 1A) and at different roach, although not paralyzed, does not initiate spontane- time points after the sting (n = 33) or, in control individuals ous walking and fails to escape aversive stimuli. This allows (n = 15), after stressing the cockroaches by handling. For the wasp to grab the cockroach by the antenna and walk it to each individual cockroach, we considered the arithmetic dif- a nest much like a dog on a leash. There, the wasp lays an ference between the nociceptive and walking threshold values egg on the prey, seals the nest, and leaves. The stung cock- to be the ‘‘escape threshold’’ of the individual (Figure 1A). Prior roach, however, does not fight to escape its tomb but rather to the sting, the average escape threshold was 3 6 2 V, dem- awaits its fate, being consumed alive by the hatching larva onstrating that cockroaches escaped the shock soon after it over several days. We investigated whether the venom- became noxious. In marked contrast, after the sting but not af- induced hypokinesia is a result of an overall decrease in ter the control treatment, none of the cockroaches walked to arousal or, alternatively, a specific decrease in the drive to the neutral chamber in response to noxious stimuli of initial- initiate or maintain walking. We found that the venom specif- level intensities. By increasing the applied voltage in 3 V incre- ically affects both the threshold for the initiation and the ments at 30 s intervals, we found that the escape threshold in maintenance of walking-related behaviors. Nevertheless, stung cockroaches increased gradually after the sting, reach- the walking pattern generator itself appears to be intact. ing peak levels at 2–4 hr, staying constant for at least 20 hr, and We thus report that the venom, rather than decreasing over- returning to baseline 72 hr after the sting (Figure 1A). In fact, at all arousal, manipulates neuronal centers within the cerebral the peak of the venom’s effect, stung cockroaches endured ganglia that are specifically involved in the initiation and voltages 2- to 3-fold stronger than their nociceptive threshold maintenance of walking. without moving to the neutral chamber, although leg contractions appeared normal. Cockroaches of the control group Results and Discussion showed no significant change in the escape threshold voltage throughout the study. Wasp Venom Decreases Spontaneous Next, we tested whether repetitive stimuli of identical inten- and Evoked Walking sities could evoke walking in stung cockroaches (Figures 1B Stung cockroaches are hypokinetic and rarely, if ever, initiate and 1C). This was achieved by applying trains of 1–5 brief cali- walking under natural conditions [1]. We first confirmed these brated tactile stimuli (<0.5 s, spaced 1.5 s apart) to the abdo- results (see Figure S1 available online) and found that although men of tethered cockroaches standing on a slippery surface. >90% of the control group explored a novel area (n = 14) and In control individuals, a single stimulus was sufficient to evoke effectively escaped aversive stimuli (n = 26), stung cock- a stereotypical escape response (28 6 16 steps; n = 5; data not roaches showed no exploratory behavior (n = 14) and in only shown). In stung cockroaches, by contrast, trains of one or two <15% of the trials (n = 33) escaped different types of stimuli, stimuli rarely initiated a synchronized walking response (one although each stimulus recruits different sensory and central stimulus, 1 6 1 steps; two stimuli, 5 6 5 steps; n = 6; premotor pathways [6, 7]. Nevertheless, stung cockroaches Figure 1B). Trains of three or more stimuli, however, typically were not completely motionless: antennae occasionally evoked walking that outlasted the length of the actual stimulus moved in an exploratory manner. In 8% of the stung cock- train (roughly 15 steps). Increasing the number of stimuli in roaches, a tactile stimulus directed at the abdomen evoked a train more reliably evoked walking. Taken together, our results demonstrate that stung cockroaches have an elevated behavioral threshold for the initiation *Correspondence: [email protected] of walking. The venom is injected into the cerebral ganglia and Current Biology Vol 18 No 12 878

Figure 1. Stung Cockroaches Show Elevated Behavioral Thresholds for Walking (A) Cockroaches were subjected to escapable foot shocks in a modified Shuttle Box before (time point 0) and at different time points after the sting or after handling (control). The threshold voltage required to elicit escape responses gradually and reversibly increases in stung but not in control individuals. Each data point represents the mean 6 SD; data points labeled with the same letter are not significantly different, whereas different letters represent significant differences at p < 0.001. Data points in the control group are not significantly different. (B and C) Stung cockroaches were positioned in a walking posture on a slippery surface, and walking was recorded with a photoresistor placed beneath the mesothoracic leg. Trains composed of 1–5 brief (0.5 s) tactile stimuli of identical intensities were applied to the abdomen. (B) Although trains composed of 1–2 stimuli fail to evoke walking, trains of 3 or more stimuli evoke walking that outlasts the duration of the stimulus train. Data points represent mean 6 SD; asterisks represent significant differences (*p < 0.05; **p < 0.001) relative to the number of steps elicited by a single stimulus. (C) Motion traces of the mesothoracic leg showing the response to trains of tactile stimuli (arrowheads) in the same cockroach before and 180 min after a sting. A single stimulus evokes walking before (top) but not after (middle) the sting. Four repetitive stimuli applied to the stung cockroach (bottom), however, evoke a walking episode that outlasted the stimuli train. The walking rhythm is significantly slower after the sting (note the different time scales used in the top and bottom panels). does not affect action potential propagation [10], so it seems occasionally attempting to climb out (Figure 2A). Overall, con- that stung cockroaches have a deficit in ‘‘reaching the deci- trol subjects spent 53 6 11 s of the 1 min trial period actively sion’’ to walk. This deficit could, however, result from (1) swimming (data not shown). Swimming behavior typically a change in the drive for walking initiation or (2) a change in ceased toward the end of the trial period, although the cock- the drive for walking maintenance, despite successful initia- roaches continued to move their antennae in an exploratory tion [11]. manner while floating passively on the water surface. In this To discriminate between these two alternatives, we tested state, applying wind or tactile stimuli reliably reinitiated swim- whether the maintenance of walking, once initiated by a supra- ming. threshold stimulus, is impaired in stung cockroaches. Like their control counterparts, 93% of the stung cockroaches initiated rigorous wall-oriented swimming immedi- Wasp Venom Affects Walking Maintenance ately upon immersion (Figure 2A). However, in marked Walking episodes initiated by repetitive tactile stimuli contrast to members of the control group, swimming in stung (Figure 1C) were significantly slower in stung as compared individuals ceased soon afterwards although antennal move- with control individuals (2 6 1 and 6 6 1 steps/s, respectively; ments, as in controls, persisted. The mean duration of active p < 0.05). To further analyze walking maintenance, we adopted swimming in the stung group was 10 6 8 s, significantly shorter a behavioral paradigm commonly used in the study of motiva- than observed with the control group (p < 0.001). Three obser- tion in mammals, namely the forced swimming test [12]. For vations indicated that the reduction in swimming duration was cockroaches, which are terrestrial insects, water immersion not due to direct motor deficits or muscle fatigue. First, as in for periods longer than several minutes can prove fatal. Water the control group, external stimuli applied during immobile immersion thus provides a reliable and continuous stressful periods tended to re-evoke swimming, although the resulting stimulus that typically produces strong, stereotypic aversion swimming episodes were shorter in stung cockroaches. Sec- responses [13, 14]. We can thus investigate whether stung ond, in rare cases (such as presented in Figure 2A), short cockroaches initiate and maintain normal aversion responses bursts of active swimming were spontaneously initiated after when exposed to continuous stress. Stung and control individ- long periods of immobility. Finally, after removal from the uals (n = 20 and n = 34, respectively) were immersed in a water cylinder, cockroaches were placed on their backs and water-filled cylinder and the duration of spontaneous active their ability to right themselves, i.e., to turn over using their swimming within a 1 min trial period was recorded. All control legs and stand upright, was tested. All stung cockroaches individuals initiated swimming upon immersion, quickly reach- showed vigorous leg movements during righting attempts ing the wall and continuing to swim rigorously close to the rim, and successfully turned themselves over, similar to members Wasp Manipulates Cockroach Drive for Walking 879

Figure 2. Stung Cockroaches Are Less Active in the Forced Swimming Test but Show Normal Motor Pattern during Active Swimming (A) Motion traces of one control (left) and one stung (right) cockroach in the forced swimming test. Arrows represent the position at the beginning of the trial, asterisks represent the position at the end of the trial, bold segments represent periods of inactivity. Trial duration is 1 min. (B) Simultaneous EMG recordings from the coxal depressor muscles of the right mesothoracic (R2) and left and right metathoracic legs (L3 and R3, respectively) during active swimming by a stung cockroach. The rhythmic Ds potentials signify a normal swimming pattern. A phase histogram (right) of the slow coxal depressor bursts during active swimming by stung cockroaches demonstrates that ‘‘alternating-tripod’’ interleg coordination is maintained. Values above the dashed lines represent the mean phases of the right mesothoracic and metathoracic legs in the cycle of the left metathoracic leg (R2/L3 and R3/L3 phase, respectively).

considered as ‘‘walking on water.’’ Thus, investigating changes in swimming patterns can provide insight into the walking deficits of stung cockroaches. of the control group. Thus, to summarize, it appears that The characteristic correlation between cycle period, Ds dis- venom injection into the cerebral ganglia affects the drive for charge rate, and stance duration persisted in stung individuals both initiation and maintenance of walking. during active swimming and was similar to the correlation noted with control individuals (n = 4 and n = 5, respectively; Wasp Venom Does Not Affect Pattern Generator Properties p < 0.001) (Figure 3). However, the average Ds discharge rates The motor pattern observed during active swimming in stung within the stance phase were significantly lower, and the cockroaches was comparable to that presented by control stance duration significantly longer throughout the range of individuals, as revealed by EMG electrodes implanted in the cycle periods in stung insects, as compared with the control coxal depressor muscles of the legs. First, stung cockroaches group (p < 0.001) (Figure 3). These differences imply a decrease (n = 4) presented a normal gait during active swimming in the excitation level of the pattern generator responsible for (Figure 2B), namely the ‘‘alternating-tripod’’ gait that is also swimming. Because swimming and walking appear to be gov- observed in untreated cockroaches during walking [13, 14]. erned by the same pattern generator, it is reasonable to as- Furthermore, and similar to control individuals, fast coxal de- sume that walking deficits in stung cockroaches are also the pressor potentials (Df) were sometimes recruited during active result of decreased excitability in thoracic circuits. Such a de- swimming bursts in stung cockroaches (data not shown). This crease in excitability could be due to a decrease in the activity phenomenon was never observed in stung cockroaches of thoracic dorsal unpaired median (DUM) neurons [18], a pop- standing on a solid surface. Thus, under potentially fatal envi- ulation of excitatory octopamine-releasing neurons known to ronmental conditions, stung cockroaches are able to express affect the walking [19] and escape pattern generators [20]. a normal walking-like motor pattern for short periods. Next, to characterize the intrinsic properties and activity of Wasp Venom Has Little Effect on Other the thoracic pattern generator in control and stung cock- Locomotory Behaviors roaches, we focused on the occurrence of Ds potentials in Finally, we considered whether the wasp’s venom specifically the metathoracic leg (Figure 3). The motor neurons producing affects walking, or motor behavior in general. We assumed these potentials are active during the retraction (stance) phase that if the venom specifically affects the drive to walk, rather of swimming or walking to provide forward propulsion. An in- than the overall arousal state, then the execution and mainte- trinsic property of the walking pattern generator in Periplaneta nance of behavioral patterns other than walking should not be is that with increasing cycle period, the average Ds discharge significantly affected. We investigated this claim by testing rate decreases while the stance duration increases [15–17]. two distinct motor behaviors: righting, which involves motor We began our investigation by validating that in untreated structures similar to walking, and flying, which involves differ- cockroaches (n = 5), swimming and walking are similar at the ent motor structures (Figure 4). Stung cockroaches did not level of the pattern generator output, with correlation between show any deficiency in righting behavior, with all cockroaches cycle period, average Ds discharge rate, and duration of the being able to turn themselves upright in less than 3 s (n = 15; stance phase all being comparable for swimming and walking data not shown). By placing cockroaches on their dorsal sides (p < 0.001; Figure S2). These results support previous findings on a smooth surface, we were able to record ongoing muscle [13, 14] suggesting that walking and swimming in Periplaneta potentials from the mesothoracic leg during righting attempts. are highly similar behaviors and that swimming may be Here again, and similarly to what was observed in control Current Biology Vol 18 No 12 880

Figure 3. Stung Cockroaches Show Normal Pattern Generator Properties but Decreased Activity of Leg Muscle Potentials Pooled data of Ds muscle potentials from the metathoracic leg, as recorded with EMG electrodes during active swimming. The graphs demonstrate the correlation between the average discharge rates of Ds potentials (A), the duration of the stance phase (B), and the cycle period. In both the stung (open circles) and control (ﬁlled circles) group, the Ds discharge rate decreases, while the stance duration increases with increasing cycle period. However, for each cycle period, the Ds discharge rate is lower in stung cockroaches and the overall distribution of Ds discharge rates is shifted to lower values, as compared with the control group (A, insert). Similarly, the stance duration is longer for stung cockroaches, as compared with controls (B, insert).

In conclusion, we provide here evidence that the venom injected by A. compressa into the head of its cockroach prey specifically manipulates the drive to initiate and maintain walking, rather than the general arousal state of the prey. The sum of intrinsic forces pro- moting defined sets of behaviors in response to relevant stimuli defines the ‘‘motivational state’’ of an animal to initiate or maintain a specific behavior [21– 23]. We therefore suggest that the venom manipulates a precursor form of motivation in the cockroach prey. This could be achieved, for instance, if the venom targets neurotransmission in the cerebral ganglia. Monoaminergic systems, and in particular the octopaminergic and dopaminergic systems, are individuals, stung cockroaches showed rhythmic as well as known to profoundly affect motivation and locomotion in sporadic Ds bursts, occasionally accompanied by Df poten- insects [24–27] and are thus possible targets of the venom. tials (Figure 4A). Accordingly, recent evidence suggests that the monoamine In a different experiment, we removed the legs of cock- pair dopamine/octopamine plays an important role in cerebral roaches to eliminate leg-ground contact and then applied cali- circuits in the induction of hypokinesia in the wasp’s cock- brated wind stimuli to wind-sensitive hairs on the cerci. This roach prey [28, 29]. Such manipulation would have to affect setup allowed us to investigate fictive flying episodes in teth- specific pathways converging, directly or indirectly (for ered cockroaches via EMG electrodes implanted in the de- example, via thoracic DUM neurons [18]), onto thoracic pat- pressor muscles of three wings, simultaneously. Overall, stung tern-generating circuits to specifically reduce the propensity cockroaches (n = 11) showed no deficits in initiating flying be- of walking-related behaviors. Further investigation of these havior or in the flight motor pattern, as compared with control pathways, which represent the link between decisions made individuals (n = 9). First, the distribution of the threshold wind in the cerebral ganglia and their execution in the thoracic velocity required to elicit flying overlapped in the two groups of ganglia, might lead to further understanding of the neuronal cockroaches (p = 0.415; data not shown). Second, wingbeat underpinnings of motivation and goal-directed actions in frequency was not significantly different between the control insects. and stung groups (24.4 6 1.5 Hz and 26.1 6 1.6 Hz, respectively; p = 0.159). Third, EMG recordings and successive phase Experimental Procedures analysis (Figure 4B) showed that contralateral wings on the same body segment moved synchronously and that the fore- Animals wing burst preceded the hindwing burst by approximately Parasitoid wasps, Ampulex compressa Fabricius (Hymenoptera: Ampulici- 7 ms. Both parameters are characteristic of cockroach flying dae), were taken 2–6 weeks after metamorphosis. Male Periplaneta americana cockroaches were taken 2–4 months after adult molt. All animals [14]. Taken together, our results show that the wasp venom were reared under laboratory conditions [29]. A wasp was allowed to com- appears to specifically affect walking, rather than motor plete a full stinging sequence and the cockroach was immediately removed behavior in general. to prevent further manipulation by the wasp. Wasp Manipulates Cockroach Drive for Walking 881

Figure 4. Stung Cockroaches Show Righting and Flying Behaviors Comparable to Those of Normal Cockroaches (A) EMG recordings from the coxal depressor muscles of the metathoracic leg during righting attempts are similar in control (left) and stung (right) individuals. Slow (Ds) and fast (Df) potentials occur regularly during righting attempts in control as well as in stung cockroaches. The scale bar applies to both traces. (B) Simultaneous EMG recordings from depressor muscles of the right forewing (R1) and the right and left hindwings (R2 and L2, respectively) during a flying episode in a stung cockroach. A phase histogram (right) of wing depressor potentials demonstrates a normal flying pattern, in which the burst of the forewings slightly pre- cedes the synchronous burst of the hindwings. Dashed lines represent the average phases, quantified by the values above.

Behavioral Assays We focused on swimming and flying because these are two rhythmic behav- A detailed description of some of the assays performed in the present work iors that could be elicited in stung cockroaches. For swimming, we ana- can be found in [14]. In brief, assays were performed on freely moving cock- lyzed, in detail, >800 retraction cycles of active swimming, based on [14, roaches in an open-field arena (radius = 30 cm) or, for more detailed behav- 16]. For flying, we analyzed, in detail, >4000 wingbeat cycles of active flight, ioral analyses, on tethered individuals walking on an oiled surface. Explor- based on [14]. All electrophysiological recordings were digitized and pro- atory behavior was measured as the total duration of exploration and cessed with Spike2 data acquisition software (CED, Cambridge, UK). climbing attempts in an open-field arena during a 10 min trial. Positive ex- Low-frequency movement artifacts were filtered out digitally. ploratory behavior was defined as walking >1 min in the novel environment. Escape responses were measured as the distance traveled after receiving Statistical Analysis a tactile stimulus to the antenna or abdomen or a wind stimulus to the cerci. We used Student’s t test for analysis of normally distributed data or the A positive escape response was defined by walking >5 cm. Each stimulus Mann-Whitney Rank Sum Test for testing non-normally distributed data. was given three times for each cockroach with 1 min intervals. Escapable Correlations were tested with the Pearson Product-Moment Correlation foot shocks were administered in a custom-built Perspex box (35 3 7 3 test. Data in this work are presented as means 6 standard deviation (SD), 15 cm), divided into a shocking chamber and a neutral chamber. The shock- with n representing the number of animals tested. ing chamber was lined with a 6-mm-thick antistatic foam sheet (SJM Euro- stat, Cheshire, UK), through which electric shocks (10 s pulse trains, pulse duration = 1 s, pulse period = 1.5 s, amplitude range = 30–100 V) were ap- Supplemental Data plied to the tarsi of standing cockroaches. Repetitive stimuli of identical intensity were applied to the abdomen of tethered cockroaches via a glass rod Two figures and one movie are available at http://www.current-biology. connected to a step motor. A digitized sequencer (Spike2, CED, Cambridge, com/cgi/content/full/18/12/877/DC1/. UK) supplied a train of 1–5 brief (<0.5 s) stimuli, spaced 1.5 s apart. Different trains, consisting of a different number of stimuli, were applied at random Acknowledgments order, and each cockroach received at least three trains of each type. In some instances, the same cockroach was tested before and after the sting. We thank Gustavo Glusman for technical assistance, Benjamin Libersat for A photoresistor was placed directly beneath the mesothoracic leg that was performing some experiments, and Gal Haspel, Antonia Delago, Francois illuminated from above, allowing us to record voltage deflections resulting Clarac, and Jerry Eichler for valuable comments. This work was supported from leg movements evoked by the tactile stimuli. Swimming in a modified by the Israel Academy of Sciences and Humanities (848/07). The experi- forced swimming test [12] was induced by placing a cockroach in an opaque ments performed comply with Principles of Animal Care, NIH publication pool (25 cm in diameter) filled with water to a height of 10 cm, maintained at no. 86-23, revised in 1985, and also with the current laws of the State of 25C. The duration of swimming during a 1 min period was measured, and Israel. swimming behavior was recorded with a camera (Cohu, San Diego, CA) for later analysis and motion tracking. Flying was tested by excising the Received: March 4, 2008 legs and applying a wind puff to the cerci, as described previously [14]. Revised: April 18, 2008 Righting was tested by placing a cockroach ventral side up on coarse sand- Accepted: April 29, 2008 paper and measuring the duration it took for the cockroach to right itself, Published online: June 5, 2008 i.e., to turn over and stand upright. A righting attempt was considered successful if the cockroach turned over in less than 3 s. References

Electrophysiology 1. Fouad, K., Libersat, F., and Rathmayer, W. (1994). The venom of the Cockroaches used for electrophysiological studies were not pooled with cockroach-hunting wasp Ampulex compressa changes motor thresh- cockroaches used for behavioral studies. Behavioral parameters measured olds: a novel tool for studying the neural control of arousal? Zoology in cockroaches used for electrophysiology, however, were not signiﬁcantly 98, 23–34. different from data obtained in behavioral studies (data not shown). EMG 2. Fouad, K., Libersat, F., and Rathmayer, W. (1996). Neuromodulation of electrodes used were 50 mm steel wires insulated to the tip, implanted the escape behavior of the cockroach Periplaneta americana by the into the coxal depressor muscles of the legs or the depressor muscles of venom of the parasitic wasp Ampulex compressa. J. Comp. Physiol. the wings, and held in place with dental wax (for further details, see [14]). [A] 178, 91–100. Current Biology Vol 18 No 12 882

3. Libersat, F. (2003). Wasp uses venom cocktail to manipulate the behav- 28. Weisel-Eichler, A., and Libersat, F. (2002). Are monoaminergic systems ior of its cockroach prey. J. Comp. Physiol. [A] 189, 497–508. involved in the lethargy induced by a parasitoid wasp in the cockroach 4. Haspel, G., Rosenberg, L.A., and Libersat, F. (2003). Direct injection of prey? J. Comp. Physiol. [A] 188, 315–324. venom by a predatory wasp into cockroach brain. J. Neurobiol. 56, 29. Rosenberg, L.A., Glusman, J.G., and Libersat, F. (2007). Octopamine 287–292. partially restores walking in hypokinetic cockroaches stung by the par- 5. Gal, R., Rosenberg, L.A., and Libersat, F. (2005). Parasitoid wasp uses asitoid wasp Ampulex compressa. J. Exp. Biol. 210, 4411–4417. a venom cocktail injected into the brain to manipulate the behavior and metabolism of its cockroach prey. Arch. Insect. Biochem. 60, 198–208. 6. Pollack, A.J., Ritzmann, R.E., and Watson, J.T. (1995). Dual pathways for tactile sensory information to thoracic interneurons in the cockroach. J. Neurobiol. 26, 33–46. 7. Ye, S., Leung, V., Khan, A., Baba, Y., and Comer, C.M. (2003). The antennal system and cockroach evasive behavior. I. Roles for visual and mechanosensory cues in the response. J. Comp. Physiol. [A] 189, 89–96. 8. Abramson, C.I., Morris, A.W., Michaluk, L.M., and Squire, J. (2004). An- tistatic foam as a shocking surface for behavioral studies with honey bees (Hymenoptera: Apidae) and American cockroaches (Orthoptera: Blattelidae). J. Entomol. Sci. 39, 562–566. 9. Brown, G.E., Anderson, C.L., and Scruggs, J.L. (1994). Shock-induced analgesia in the cockroach (Periplaneta americana). Psychol. Rep. 74, 1051–1057. 10. Haspel, G., and Libersat, F. (2003). Wasp venom blocks central cholinergic synapses to induce transient paralysis in cockroach prey. J. Neu- robiol. 54, 628–637. 11. Kien, J., and Altman, J.S. (1992). Decision-making in the insect nervous system: a model for selection and maintenance of motor programmes. In Neurobiology of Motor Programme Selection, Pergamon Studies in Neuroscience, no. 4, J. Kien, C.R. McCrohan, and W. Winlow, eds. (Oxford: Pergamon Press), pp. 147–169. 12. McArthur, R., and Borsini, F. (2006). Animal models of depression in drug discovery: a historical perspective. Pharmacol. Biochem. Behav. 84, 436–452. 13. Cocatre-Ziligien, J.H., and Delcomyn, F. (1990). Fast axon activity and the motor pattern in cockroach legs during swimming. Physiol. Entomol. 15, 385–392. 14. Gal, R., and Libersat, F. (2006). New vistas on the initiation and maintenance of insect motor behaviors revealed by specific lesions of the head ganglia. J. Comp. Physiol. [A] 192, 1003–1020. 15. Delcomyn, F. (1971). Locomotion of cockroach Periplaneta-americana. J. Exp. Biol. 54, 443–452. 16. Pearson, K.G. (1972). Central programming and reflex control of walking in cockroach. J. Exp. Biol. 56, 173–193. 17. Sherman, E., Novotny, E., and Camhi, J.M. (1977). Modified walking rhythm employed during righting behavior in cockroach Gromphado- rhina-portentosa. J. Comp. Physiol. [A] 113, 303–316. 18. Rosenberg, L.A., Pfluger, H.J., Wegener, G., and Libersat, F. (2006). Wasp venom injected into the prey’s brain modulates thoracic identified monoaminergic neurons. J. Neurobiol. 66, 155–168. 19. Ridgel, A.L., and Ritzmann, R.E. (2005). Effects of neck and circumoeso- phageal connective lesions on posture and locomotion in the cockroach. J. Comp. Physiol. [A] 191, 559–573. 20. Goldstein, R.S., and Camhi, J.M. (1991). Different effects of the biogenic amines dopamine, serotonin and octopamine on the thoracic and abdominal portions of the escape circuit in the cockroach. J. Comp. Phys- iol. [A] 68, 103–112. 21. Mogenson, G.J., Jones, D.L., and Yim, C.Y. (1980). From motivation to action: functional interface between the limbic system and the motor system. Prog. Neurobiol. 14, 69–97. 22. Dickinson, A., and Balleine, B. (1994). Motivational control of goal- directed action. Anim. Learn. Behav. 22, 1–18. 23. Huber, R. (2005). Amines and motivated behaviors: a simpler systems approach to complex behavioral phenomena. J. Comp. Physiol. [A] 191, 231–239. 24. Roeder, T. (1999). Octopamine in invertebrates. Prog. Neurobiol. 59, 533–561. 25. Libersat, F., and Pflueger, H.J. (2004). Monoamines and the orchestration of behavior. Bioscience 54, 17–25. 26. Saraswati, S., Fox, L.E., Soll, D.R., and Wu, C.F. (2004). Tyramine and octopamine have opposite effects on the locomotion of Drosophila larvae. J. Neurobiol. 58, 425–441. 27. Jordan, K.W., Carbone, M.A., Yamamoto, A., Morgan, T.J., and Mackay, T.F. (2007). Quantitative genomics of locomotor behavior in Drosophila melanogaster. Genome Biol. 8, R172. Supplemental Data S1 A Parasitoid Wasp Manipulates the Drive for Walking of Its Cockroach Prey

Ram Gal and Frederic Libersat

Figure S1. Stung Cockroaches Show Decreased Spontaneous and Evoked Walking but Are Not Paralyzed (A) Stung cockroaches explore a novel environment and escape different types of stimuli (wind to the cerci, tactile stimuli to the abdomen or antenna) signiﬁcantly less well than do control individuals. Figure S2. Untreated Cockroaches Show a Similar Motor Pattern during (B) Traces showing EMG recordings from coxal depressor muscles of the Swimming and Walking metathoracic leg. In a control individual (top), a tactile stimulus (arrowhead) Pooled data of Ds muscle potentials from the metathoracic leg during walk- to the abdomen evokes a rhythmic running response composed of fast (Df) ing (ﬁlled circles) and swimming (open circles) in untreated cockroaches, as and slow (Ds) depressor potentials. The same stimulus in a stung cockroach recorded with EMGs. The average Ds discharge rate (A) decreases, while (bottom) evokes a transient increase in the Ds discharge rate without the the stance duration (B) increases with increasing cycle period to a similar occurrence of Df potentials. The scale bar applies to both traces. degree during swimming and walking. A Wasp Manipulates Neuronal Activity in the Sub-Esophageal Ganglion to Decrease the Drive for Walking in Its Cockroach Prey

Ram Gal1*, Frederic Libersat1,2* 1 Department of Life Sciences, Ben-Gurion University of the Negev, Be’er-Sheva, Israel, 2 Institut de Neurobiologie de la Me´diterrane´e INSERM U901, Universite´ de la Me´diterrane´e, Parc Scientifique de Luminy, Marseille, France

Abstract

Background: The parasitoid Jewel Wasp hunts cockroaches to serve as a live food supply for its offspring. The wasp stings the cockroach in the head and delivers a cocktail of neurotoxins directly inside the prey’s cerebral ganglia. Although not paralyzed, the stung cockroach becomes a living yet docile ‘zombie’, incapable of self-initiating spontaneous or evoked walking. We show here that such neuro-chemical manipulation can be attributed to decreased neuronal activity in a small region of the cockroach cerebral nervous system, the sub-esophageal ganglion (SEG). A decrease in descending permissive inputs from this ganglion to thoracic central pattern generators decreases the propensity for walking-related behaviors.

Methodology and Principal Findings: We have used behavioral, neuro-pharmacological and electrophysiological methods to show that: (1) Surgically removing the cockroach SEG prior to wasp stinging prolongs the duration of the sting 5-fold, suggesting that the wasp actively targets the SEG during the stinging sequence; (2) injecting a sodium channel blocker, procaine, into the SEG of non-stung cockroaches reversibly decreases spontaneous and evoked walking, suggesting that the SEG plays an important role in the up-regulation of locomotion; (3) artificial focal injection of crude milked venom into the SEG of non-stung cockroaches decreases spontaneous and evoked walking, as seen with naturally-stung cockroaches; and (4) spontaneous and evoked neuronal spiking activity in the SEG, recorded with an extracellular bipolar microelectrode, is markedly decreased in stung cockroaches versus non-stung controls.

Conclusions and Significance: We have identified the neuronal substrate responsible for the venom-induced manipulation of the cockroach’s drive for walking. Our data strongly support previous findings suggesting a critical and permissive role for the SEG in the regulation of locomotion in insects. By injecting a venom cocktail directly into the SEG, the parasitoid Jewel Wasp selectively manipulates the cockroach’s motivation to initiate walking without interfering with other non-related behaviors.

Citation: Gal R, Libersat F (2010) A Wasp Manipulates Neuronal Activity in the Sub-Esophageal Ganglion to Decrease the Drive for Walking in Its Cockroach Prey. PLoS ONE 5(4): e10019. doi:10.1371/journal.pone.0010019 Editor: Vladimir Brezina, Mount Sinai School of Medicine, United States of America Received January 4, 2010; Accepted March 11, 2010; Published April 7, 2010 Copyright: ß 2010 Gal, Libersat. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by a grant from the Israeli Science Foundation (848/07). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] (RG); [email protected] (FL)

Introduction adult female wasp hunts cockroaches (Periplaneta americana) for use as a live food supply for its offspring. Since development of the Animals are not automatons that react identically every time wasp’s larva requires feeding on live cockroaches for several days they encounter the same stimulus [1]. Changes in the internal [11], the adult wasp does not kill or paralyze the cockroach prey physiological state of an animal alter its responsiveness to stimuli but instead uses neurotoxins to selectively ‘hijack the free will’ of and consequently affect its motivation to engage in a given the prey. A cockroach stung by a Jewel Wasp first grooms itself behavior. Such processes and their underlying neuronal substrates excessively for 30 minutes, and then becomes hypokinetic for 3–7 have been the subject of extensive study for decades [2–4]. These days, during which time it loses the ability to self-initiate and efforts have undoubtedly benefited from studies on animals with maintain walking-related behaviors [12,13]. The stung cockroach relatively simple nervous systems, controlling stereotyped behav- is not, however, paralyzed, allowing the wasp to grab its prey by iors [1,5–8]. the antenna and lead it to a nest, with the cockroach all the while Through millions of years of co-evolution, a few animal species following in a docile manner, much like a submissive dog on a have evolved unique strategies to control the motivation of their leash (movie available online [10]). The wasp then lays an egg on prey to engage in specific behaviors, thereby manipulating the the cockroach, seals the nest and leaves the docile prey inside. The prey in most exceptional ways [9]. One such example is the wasp larva hatches two days later and feeds on the cockroach for parasitoid Jewel Wasp (Ampulex compressa) which specifically another three days. The prey, although still alive throughout this depresses the drive of its prey to engage in locomotion [10]. The process, does not put up a fight nor try to escape its tomb. The

PLoS ONE | www.plosone.org 1 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation larva then pupates inside the cockroach’s abdomen and hatches a month later as an adult, ready to continue its life cycle [11]. To render the cockroach hypokinetic, the wasp stings it twice, first in the thorax and then in the head [14]. The thoracic sting is brief and transiently paralyses the cockroach’s front legs [15,16] to facilitate the second and more precise sting into the head. The head sting is longer in duration and is responsible for the later behavioral alterations observed in stung cockroaches, i.e., excessive grooming, long-term hypokinesia and changes in the cockroach’s metabolism designed to preserve nutrients for the developing larva [11]. To investigate where in the cockroach head does the wasp inject its venom, Haspel et al. [17] injected wasps with radiolabeled amino acids and traced the radioactive venom using autoradiography. In cockroaches stung in the head by such ‘hot’ wasps, venom was traced by and large inside the cockroach’s cerebral ganglia, namely the supra-esophageal ganglion (SupEG) and sub-esophageal ganglion (SEG). Furthermore, Gal et al. [18] demonstrated that the wasp actively searches for, at least, the SupEG inside the cockroach’s head capsule during the head sting. These findings suggest that the behavioral changes observed in stung cockroaches result from the neurotoxic effects of the venom on the SupEG, the SEG, or both. However, the role of each ganglion in inducing these behavioral changes is still unclear. In insects, the cerebral ganglia are known to comprise the ‘higher-order’ neuronal centers implicated in modulating the thoracic Central Pattern Generators responsible for the spatio- temporal pattern of locomotion [19–25]. We have recently demonstrated that in stung hypokinetic cockroaches, the thoracic Central Pattern Generators are not directly affected by injected wasp venom. Rather, it is the drive to initiate and maintain walking-related behaviors that is selectively depressed in stung Figure 1. Effect of lesions in the cockroach CNS on the wasp’s stinging duration. A. Left: A Jewel Wasp stinging a cockroach in the cockroaches, with minimal or no interference to other behaviors head. Right: Schematic lateral view of the cockroach head cavity (CNS is [10]. Since the SEG has been suggested to tonically up-regulate in yellow) and the wasp’s stinger (St), shown as a scanning electron walking-related behaviors, while the SupEG appears to be micrograph superimposed and drawn to scale, penetrating through the generally inhibitory [24], we hypothesized in the present study head cuticle to reach the cerebral ganglia (supra-esophageal ganglion, that the SEG is the primary neuronal substrate responsible for the SupEG, and sub-esophageal ganglion, SEG). Locations of experimental neuro-chemical manipulation of the cockroach’s drive to initiate CNS lesions are marked with scissors: In SEG-ablated cockroaches, the connectives rostral and caudal to the SEG were severed and the walking. To test this hypothesis directly, we have employed ganglion physically removed from the head cavity. In neck connectives- behavioral, neuro-pharmacological and electrophysiological tools cut cockroaches, by contrast, only the neck connectives (NC) caudal to to investigate whether specific modulation of neuronal activity in the SEG were severed, with the ganglion itself left intact. Es: esophagus. the SEG can account for the hypokinetic state of stung Ant: antenna. Scale bar: 0.5 mm. Modified from [17]. B. Wasp stinging cockroaches. behavior after specific experimental lesions of the cockroach’s cerebral CNS. Cerebral lesions do not affect the duration of the first sting directed at the thorax (black bars). In contrast, physically removing the Results SEG from the cockroach head cavity prior to the sting (SEG ablated), but not cutting the cockroach neck connectives (NC-cut), significantly The wasp actively targets the cockroach’s SEG during the increases the duration of the second sting directed at the head (red head sting bars). ***p,0.001, as compared with sham-operated and NC-cut If the cockroach’s SEG plays a crucial role in the venom- cockroaches. doi:10.1371/journal.pone.0010019.g001 induced hypokinesia, then one would expect the wasp to actively target not only the SupEG [18] but also the SEG during the head sting. To test this hypothesis, we quantified the stinging behavior 1566 sec; NC-cut: 1563 sec; sham-operated: 1665 sec), demon- of wasps to which three groups of cockroaches were presented strating that cerebral lesions did not interfere with the wasp’s (n = 8 cockroaches in each group; Fig. 1A): (1) SEG-ablated motivation to sting or with its initial stinging behavior. Indeed, cockroaches, namely cockroaches from which the SEG had been following the typical thoracic sting, the wasp readily pulled out its surgically removed prior to the sting; (2) Neck-connectives (NC)- stinger and aimed it at the cockroach’s head. The duration of the cut cockroaches, in which the neck connectives between the thorax head sting, in marked contrast with the thoracic sting, was and the SEG were cut prior to the sting. These cockroaches, significantly longer when the wasps stung SEG-ablated cock- similar to SEG-ablated cockroaches, had no descending cerebral roaches (196688 sec, p,0.001), as compared with NC-cut or inputs reaching thoracic motor centers. Unlike SEG-ablated sham-operated cockroaches (3968 sec and 39612 sec, respec- cockroaches, however, no neuronal tissue was physically removed tively) (Fig. 1B). There was no significant difference between from the head cavity of NC-cut cockroaches. Finally, as a control, stinging durations of NC-cut and sham-operated cockroaches wasps were also presented with (3) sham-operated cockroaches. (p = 0.941), showing that elimination of descending cerebral inputs The duration of the first sting into the thorax was not to the thorax is not sufficient, by itself, to increase the head sting significantly different for any group of cockroaches (SEG-ablated: duration. Given these initial observations, A. compressa appears to

PLoS ONE | www.plosone.org 2 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation actively search for the SEG inside the head capsule of its not in the SupEG) is sufficient to decrease the drive for walking in cockroach prey while delivering the head sting. otherwise normally behaving cockroaches.

Neuro-pharmacological inhibition of the SEG decreases Neuronal activity in the SEG is decreased in stung walking in non-stung cockroaches cockroaches If the wasp’s venom inhibits neuronal activity in the SEG, thereby To directly test whether the sting manipulates neuronal activity depressing the cockroach’s drive for walking, one would expect that in the SEG, we compared the SEG activity of stung cockroaches to focal neuro-pharmacological inactivation of SEG activity would that of non-stung controls (n = 6 in each group). In these similarly depress the drive for walking. To test this hypothesis, we experiments, we used an extracellular bipolar electrode to record used the reversible sodium-channel blocker, procaine, which has spontaneous and evoked spiking activity within the SEG (Fig. 4, 5). been shown to reversibly inhibit neuronal activity in the insect central We focused our investigation on the central and middle (150–200 nervous system [26,27]. When we first tested the effect of procaine on micrometers deep) portion of this small ganglion, corresponding to a cerebral ganglion by applying it onto the SEG, the local anesthetic the natural venom injection site [17]. reversibly inhibited all neuronal activity in this ganglion (Fig. 2A). On average, spontaneous spiking activity in the core of the SEG Accordingly, we injected procaine or saline (n = 14 for each group) was decreased two-fold in stung cockroaches, as compared with focally into the SEG of non-stung cockroaches and assessed the controls (56.360.6 spikes/sec and 109.768.7 spikes/sec, respec- behavioral outcome. As a control, we also assessed the behavior of tively; p,0.05) (Fig. 4). We further characterized this difference by cockroaches after focally injecting procaine into the SupEG. applying wind stimuli to the cerci or tactile stimuli to the antenna Similar to a wasp’s sting, procaine injection into the SEG (Fig. 5), both known to be poorly effective in eliciting escape in dramatically depressed spontaneous and evoked locomotion stung versus normal cockroaches [12,14]. The number of stimulus- (Fig. 2B). The anesthetic treatment significantly decreased sponta- evoked spikes during the first 200 ms after stimulus application neous walking duration from 4.460.6 min to 0.460.7 min during a was significantly lower (p,0.05) in stung cockroaches, as 10-min trial, with the distance of escape responses dropping from compared with control cockroaches (wind stimuli: 45614 spikes 30 cm before the injection to 264 cm afterwards. In contrast, and 9368 spikes, respectively; tactile stimuli: 3068 spikes and procaine injected into the SupEG slightly increased spontaneous 89619 spikes, respectively) (Fig. 5B, C). Such a decrease in the walking (4.260.9 min before injection and 5.962.8 min afterwards, SEG neuronal response is unlikely to be caused by changes in p = 0.21) and did not affect the distance of escape responses (30 cm ascending sensory inputs, since the latency between stimulus onset before injection and 30 cm afterwards). The inhibitory effects of and maximal neuronal response in the SEG was similar in stung procaine were specific to the anesthetic, as saline injected into any of and control cockroaches (wind stimuli: 123614 ms and the head ganglia did not significantly affect spontaneous or evoked 128629 ms, respectively; p = 0.878; tactile stimuli: 4569 ms and walking (spontaneous walking duration: 4.461 min before injection 4666 ms, respectively; p = 0.288) (Fig. 5D). and 3.861 min afterwards; escape distance: 30 cm before injection and 2964 cm afterwards). Furthermore, the inhibitory effects of Discussion procaine were reversible, as 1 h after injection, no significant Cockroaches stung by the parasitoid Jewel Wasp (A. compressa), differences between procaine-injected and saline-injected cockroach- although not paralyzed, loose the ability to self-initiate locomotion es were noted in terms of walking duration (p = 0.245) or escape for several days [18]. This deficit cannot be attributed to an overall distance (p = 0.934). Thus, focal inhibition of neuronal activity in the sleep-like state for three main reasons. First, the deficit is highly SEG (but not in the SupEG) is sufficient to decrease the drive for specific, in that the threshold for initiation of other motor walking in otherwise normally behaving cockroaches. behaviors (such as righting, swimming, flight, etc.) is little affected [10]. Second, stung cockroaches do not assume a typical Crude venom injected in the SEG depresses walking in ‘quiescent’ position [28] and occasionally move their antennae non-stung cockroaches in an exploratory manner. Third, when startled by a supra- Is the injection of wasp venom into the cockroach’s SEG threshold stimulus, stung cockroaches respond by jumping in place sufficient to decrease the drive for walking, in ways similar to but do not perform the stereotypic subsequent run [14]. Thus, and procaine injection or a natural sting? To answer this question, we since the sensory and motor systems per se are fully functional in milked wasps and used a nano-injector to apply crude venom (or stung cockroaches [12,14], the wasp venom appears to specifically saline, in controls) directly and focally into the SEG or the SupEG decrease the cockroach’s drive for walking. The fact that the wasp of non-stung cockroaches (n = 5 in each group). injects its neurotoxic venom directly into the cockroach’s cerebral Crude venom injected into the SEG of non-stung cockroaches ganglia to ‘hijack the cockroach’s free will’ allows us to explore the dramatically depressed walking (Fig. 3). Injected cockroaches spent neuronal substrate responsible for this unique behavioral manip- very little time spontaneously exploring a novel arena ulation. (0.160.2 min, as compared with 7.062.2 min in controls; Insect attention and arousal states, and their correlation with p,0.001) and failed to escape tactile stimuli (escape distance: mammalian equivalents, have been thoroughly investigated during 1.560.6 cm, as compared with 2768 cm in controls; p,0.001). the last few years [29–33]. However, despite their obvious Overall, the behavior of SEG-injected cockroaches was highly implications on the regulation of behavior, the neuronal similar to that of their naturally-stung counterparts (p = 0.19), who underpinnings of motivation, or the drive to initiate specific motor displayed a walking duration of 0.160.2 min and an escape behaviors, have received relatively little attention. As such, our distance of 0.860.1 cm. In contrast with venom injected into the present study aimed at employing the wasp-cockroach parasitic SEG, venom injected into the SupEG tended to increase walking interaction to define the neural substrate responsible for the drive duration (11.167.1 min, as compared with 3.262.8 min in the to initiate walking in insects. We show that the neuro-chemical appropriate control; p = 0.07) and did not impair escape responses manipulation performed by the wasp is achieved, at the least, by (3062.5 cm, as compared with 3060.9 cm in the appropriate inhibition of neuronal activity in a small region of the cerebral control). Thus, the presence of crude wasp venom in the SEG (but ganglia of insects, namely the sub-esophageal ganglion (SEG). We

PLoS ONE | www.plosone.org 3 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation

PLoS ONE | www.plosone.org 4 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation

Figure 2. Procaine-induced inhibition of SEG neural activity depresses cockroach locomotion. A. Extracellular bipolar recording of spiking activity in the SEG of a non-stung cockroach. Transient (1 min, red bar) application of the sodium-channel blocker, procaine, completely and reversibly abolishes neuronal activity in the ganglion. B. Behavioral analysis of non-stung cockroaches injected with saline (black bars) or procaine into the SEG (red bars) or the SupEG (green bars). Spontaneous walking (top) and evoked escape responses (bottom) are reversibly suppressed by the inhibition of neuronal activity in the SEG but not of the SupEG. ***p,0.001, as compared with controls. doi:10.1371/journal.pone.0010019.g002 propose this to be the case for the following reasons: (1) The wasp the stimuli used here do not involve a preparatory phase and the has been previously shown to use sensory feedback from its stinger sequential recruitment of SEG and SupEG interneurons does not to locate the SupEG within the cockroach head cavity when take place. Taken together, the results presented in this report inflicting the head sting [18]. Hence, due to the anatomical strongly support such ‘rest state’ neuronal organization of higher location of the SEG on the trajectory to the SupEG, venom found motor control and suggest that selective inhibition of neuronal in the SEG [17] could be, in principle, an incidental by-product of activity in the SEG is sufficient to decrease the drive for walking, the sting prime target, the SupEG. In the present work, we show without interfering with other behaviors or with the thoracic that this is not the case, since surgically removing the cockroach Central Pattern Generators directly. SEG prior to stinging, but not cutting the neck connectives The exact role of the SupEG in the venom-induced manipu- without SEG removal, significantly increases the duration of the lation of the cockroach motor behavior remains, as yet, rather head sting. This suggests that during the head sting the wasp elusive. Several possibilities can be offered, such as a role in actively targets not only the SupEG, as previously reported [18], evoking the excessive grooming behavior seen in stung cockroaches but also the SEG, unraveling the potential role of this ganglion in [13], or importance for venom-induced changes in cockroach venom-induced hypokinesia; (2) in vivo pharmacological inhibition metabolism [11]. It is also possible that the SupEG, in concert with of SEG neuronal activity by procaine, a reversible sodium-channel the SEG, plays a role in inducing certain aspects of venom-induced blocker, reversibly decreases the propensity for spontaneous and hypokinesia either directly, by affecting specific circuitries in this evoked walking in non-stung cockroaches. Inhibition of the ganglion, or indirectly, by affecting ascending SEG interneurons SupEG with procaine, in contrast, has little effect on spontaneous and evoked walking. These results are in agreement with previous studies (see [24] and references therein) which have suggested, using lesion experiments, that the SEG exerts a net tonic permissive effect on thoracic motor centers; (3) micro-injection of crude wasp venom into the SEG (but not into the SupEG) of non-stung cockroaches is sufficient to decrease the propensity for spontaneous and evoked walking, similar to what is seen in stung cockroaches; (4) spontaneous and evoked electrophysiological activity in the SEG is decreased in stung cockroaches, as compared with controls. Thus, our data unequivocally demonstrate the role of the SEG in the venom-induced inhibition of the drive for walking in cockroaches stung by A. compressa. To the best of our knowledge, these results provide the first direct evidence to support this long-standing hypothesis [14,17]. Although the role of the SEG in the regulation of insect locomotion is, to date, still unclear, some previous evidence suggests that it exerts a permissive descending tonic effect on thoracic motor centers (see [24] and references therein). This is in contrast with the descending influence of the SupEG, where some neuronal structures (e.g. the Central Body Complex) seem to up- regulate, while others (e.g. the Mushroom Bodies) apparently down-regulate thoracic motor centers [22,25,34,35]. In locusts, decision-making with respect to the selection and maintenance of walking, has also been examined using intracellular recordings of neurons in the SEG and the SupEG [36]. The spontaneous initiation of walking is accompanied by changes in the firing pattern of several SEG and SupEG descending interneurons. However, while SEG and SupEG interneurons both fire during walking, and are thus both involved in walking maintenance (see also [21,37]), predominantly SEG interneurons fire during the preparatory phase of walking. This observation suggests a prime role for SEG neuronal circuits in determining the motivational level or ‘rest state’ of the animal [7] to engage into walking. Figure 3. Behavioral analysis of stung and non-stung cock- Inhibition of SEG neuronal activity could therefore, in principle, roaches injected with crude milked venom into different decrease the propensity for expression of spontaneously initiated regions of the cerebral ganglia. Venom injected into the SEG, walking-related behaviors. Similarly, since the SEG sends similar to a natural wasp sting, significantly depresses spontaneous walking (top) and escape responses (bottom). In contrast, venom permissive tonic inputs to thoracic pattern generators [24], injected into the SupEG has an opposite, albeit not significant effect. inhibition of SEG activity could also depress walking in response ***p,0.001 compared with the respective controls. to sensory stimuli, such as those used in the present study, although doi:10.1371/journal.pone.0010019.g003

PLoS ONE | www.plosone.org 5 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation

Figure 4. Spontaneous neural activity in the SEG of control and stung cockroaches. A, B. Extracellular bipolar microelectrode recordings of spontaneous spiking activity in the core of the SEG in one non-stung (‘control’) and one stung cockroach. The dashed region in A is enlarged in B. C. The spontaneous firing rate in the core of the SEG is significantly decreased in stung cockroaches. **p,0.05. doi:10.1371/journal.pone.0010019.g004 which, in turn, modulate SupEG circuitries that control motor shown that certain structures within the SupEG, and especially the behavior. A direct effect of the venom on the SupEG apparently Central Body Complex, affect some finer aspects of locomotion, contradicts previous studies which showed that decerebrated including the frequency, duration and coordination of walking, insects (i.e., those without a SupEG) tend to walk uninhibitedly turning behavior and obstacle climbing [19,22,23,35,42–46]. The (see, for instance, [23,24,38–41]), suggesting a generally inhibitory venom could thus, in principle, specifically manipulate these effect of this ganglion on locomotion. However, it has also been SupEG structures, in addition to manipulating SEG activity, to

PLoS ONE | www.plosone.org 6 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation

Figure 5. Evoked neural activity in the SEG of control and stung cockroaches. A. Extracellular bipolar microelectrode recordings of evoked activity in the core of the SEG in one non-stung (‘control’) and one stung cockroach (left: wind stimulus to the cerci; right: tactile stimulus to the antenna; arrows: stimulus onset). B. Peri-stimulus time histograms for wind (left) and tactile (right) stimuli. Each data point represents the mean (6 SEM) number of spikes within a 20 ms time bin. The response to stimuli is decreased in stung cockroaches, especially during the first 200 ms after stimulus onset (represented by a bar above the histograms). C. The number of spikes during the first 200 ms immediately after the stimulus onset. Stung cockroaches show decreased responses to wind and tactile stimuli. **p,0.05. D. The latency between stimulus onset and maximal spiking response is similar in stung and control cockroaches. doi:10.1371/journal.pone.0010019.g005

PLoS ONE | www.plosone.org 7 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation further control the initiation of locomotory behavior in the interference with neuronal tissue. Micro-injections: ANano- cockroach prey. Volumetric Injector (NVI-570A/V, Medical Systems, Greenvale, The specific neurons within the SEG that are targeted by the NY) was used to deliver solutions directly into the cerebral ganglia venom to induce hypokinesia are currently under investigation. (approximately 40 nl to the SEG or 100 nl to the SupEG). For Prime candidates are neuromodulatory interneurons, in particular SEG-injections, the cockroach was immobilized ventral side up, a monoaminergic interneurons, descending from the SEG to flap was opened in the ventral head cuticle and the neck muscle thoracic motor centers and/or ascending from the SEG to the gently moved aside to expose the ganglion. Injections were aimed SupEG. One such population comprises the octopaminergic (OA) and guided stereotactically, using ganglionic fiducials, to the middle unpaired median neurons of the SEG, the axons of some of which and centre of the SEG, approximately 150–200 mm deep. For innervate segmental ganglia, while others innervate major SupEG injections, the cockroach was immobilized dorsal side up neuropiles in the SupEG [47–56]. Recently, activity in SEG-OA and a small flap was opened between the compound eyes. neurons of Manduca larvae has been correlated with fictive Injections were directed between the two hemispheres of the locomotion [57], further highlighting these neurons as major protocerebrum, between the Mushroom Bodies and in the vicinity candidates for the venom-induced hypokinesia observed in of the Central Body Complex, concomitant with the location of a cockroaches. Furthermore, we have recently shown that in stung natural wasp sting [17]. All injected solutions were added with an cockroaches, the octopamine receptor agonist, chlordimeform, inert viable tracer (0.1% Janus Green) to allow tracing of the induces a significant increase in spontaneous walking when injection site post mortem. injected into the SupEG [53]. This suggests that the wasp’s venom interferes with octopaminergic modulation of walking Pharmacology initiation in central structures of the cockroach SupEG. Procaine was freshly prepared and dissolved to a concentration To summarize, we have shown here that the wasp actively of 500 mg/ml in vehicle containing cockroach saline and 0.1% searches for the SEG of its host into which to inject its venom. Janus Green. Venom was freshly milked from 10 wasps as Having previously shown that the wasp injects venom directly into described previously [16] and dissolved approximately 1:10 in a the two cerebral ganglia [17] and that the venom’s major effect is vehicle containing 0.1% Janus Green, 10 mM HEPES buffer and to decrease the drive for walking initiation [10], the novelty of the 0.1 mM PMSF. Controls were injected with the respective vehicle present study is the experimental verification that the wasp alone. decreases the neuronal activity in the SEG to specifically down- regulate the drive for walking in its host. Given these facts, one can Behavioral assays only wonder how, through millions of years of co-evolution, A detailed description of some of the assays performed here can Ampulex has evolved this exquisite strategy to chemically control be found in [24]. Briefly, all cockroach behavioral assays were such ‘higher’ behavioral function in its host. By further identifying performed on freely-moving cockroaches in an open-field arena the neuronal basis of these parasite-induced alterations of host (radius = 30 cm). Spontaneous walking was measured as the total behavior, we hope to increase our understanding of the duration of exploration of the arena during a 10-min (after neurobiology of the selection and initiation of behaviors and the procaine injections) or a 30-min (after venom injections) trial associated neural mechanisms underlying changes in responsive- period. Walking episodes that occurred simultaneously with ness, both prime issues in the study of motivation. grooming were considered as walking episodes. Escape responses were measured as the distance the cockroach ran after receiving a Materials and Methods tactile stimulus to the abdomen. The procedure was repeated three times with 1 min intervals and the results averaged for each Animals cockroach and then pooled with the results of the entire group. Ampulex compressa Fabricius (Hymenoptera: Ampulicidae) wasps and Periplaneta americana cockroaches were reared in crowded Electrophysiology colonies under laboratory conditions of 40–60% humidity, 30uC Setup. Cockroaches were anesthetized with carbon dioxide, and a 12L:12D cycle. All animals were supplied with water and immobilized ventral side up after removing the legs and wings to food (cat chow for cockroaches and honey for wasps) ad libitum. For stumps, and covered with a sheet of modeling clay to limit stinging, an adult female wasp was introduced into a terrarium hemolymph loss and to prevent spontaneous righting or flight-like together with an adult male cockroach and allowed to afflict the movements (see [24]). Next, the head was fixed to the recording full stinging sequence, namely a thoracic sting followed by a head platform with insect-pin staples and beeswax to prevent sting. After stinging, the cockroach was immediately removed and movement. The SEG was exposed by removing the mouthparts isolated to prevent further manipulation by the wasp. and neck muscle, and care was taken to minimize damage to the trachea. The ganglion was desheathed and perfused with isotonic Surgical procedures cockroach saline [58] throughout the experiment. In General. Prior to all surgical procedures, cockroaches were electrophysiological experiments where procaine was used, we anesthetized with carbon dioxide and immobilized with modeling recorded ongoing activity for 30 min, then applied procaine clay on a wax platform. A staple-shaped insect pin was softly (500 mg/ml in cockroach saline) onto the SEG for 1 min, and pressed against the neck to regulate hemolymph flow to the head then washed the ganglion thoroughly with saline. during the procedure [24]. All cuticular incisions were allowed to Stimulation. Cockroaches were allowed 15 min to recover seal by hemolymph coagulation. Cockroach CNS lesions (see Fig. 1A): from the surgical procedure and were then subjected to a To cut the neck connectives (NCs), a U-shaped incision was stimulation protocol composed of wind and tactile stimuli (6–12 performed to open a small flap in the ventral head cuticle and the stimuli of each type) applied alternatively at 30 sec intervals. Wind NCs were cut with fine micro-scissors. In SEG-ablated cockroaches, stimuli were directed at the cerci in the tail-to-head direction with the circumesophageal connectives were subsequently severed and a custom-built wind generator [59] which delivered wind puffs of the SEG was physically removed from the head. In sham-operated roughly 150 ms in duration. To apply tactile stimuli to the control cockroaches, a flap was opened for 10 min with no antenna, we first prevented spontaneous antennal movements by

PLoS ONE | www.plosone.org 8 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation holding the antenna in place using a staple-shaped insect pin. The the total number of spikes in each bin was counted and averaged pin was pushed into the wax-coated recording platform so that it across repeats to yield the individual average response of a very lightly pressed the base of the antenna against the platform. cockroach. This response was then averaged across different This confined the antennal flagellum to movements of ,0.1 cm in cockroaches to produce the pooled peri-stimulus time histogram the lateral plane and ,0.1 cm in the dorso-ventral plane. The presented in Fig. 5B. Stimulus-response latency was calculated as stimuli were presented to the middle of the antennal flagellum, the time between the onset of the stimulus and the peak neuronal approximately 3 cm from the scape, using a steel rod which briefly response, defined as the maximum number of spikes within a (,60 cm/sec) deflected the antenna 1.5 cm medio-laterally. Such 20 ms time bin. The stability of the recording quality throughout a stimulus induced a lateral bending of the base of the antenna the experiment was controlled by calculating the percent change against the confining pin, which we empirically determined to in wind-evoked spikes between the first and last stimuli applied evoke the most spiking activity in the core of the SEG. during the stimulation protocol. In this study, we only included Furthermore, the effect of this stimulus approximates the natural experiments in which this change did not exceed 20%, which we condition [60], where the wasp bends and then cuts the cockroach consider as acceptable variability for extracellular in vivo antennae after the sting. While such bending of the antenna recordings. Spikes were acquired, sorted and analyzed with reliably evokes a rapid escape response in normal cockroaches, it Spike2 data acquisition software (CED, Cambridge, UK) on a fails to evoke such a response in stung cockroaches. Recording:We personal computer. All pooled electrophysiological data are recorded spiking activity from the center and middle of the SEG, presented as mean 6 SEM. 150–180 mm deep in the ganglion, with an extracellular bipolar tungsten microelectrode (1 MV,1mm tip diameter and 125 mm Statistical analysis tip spacing; World Precision Instruments, Sarasota, FL). We chose We used Student’s t-test to analyze normally distributed data or this region for three main reasons: First, during a sting, the wasp the Mann-Whitney Rank Sum Test for non-normally distributed injects its venom mainly into the middle and center of the data. Except for the electrophysiological data described above, all ganglion, in and around the middle neuromere [17]. Second, in data in this work are presented as mean 6 standard deviation, control cockroaches, we found this region to be the most with n representing the number of animals considered. spontaneously and reliably active, as well as demonstrating the widest variety of ongoing spike shapes and sizes. Third, we found this region to be the most responsive to tactile and wind stimuli, as Acknowledgments described above. The electrode was guided stereotactically using a We thank Gustavo Glusman, Ze’ev Itsekson and Cesar Echavarria for their finely scaled micromanipulator and trachea as ganglionic fiducials. help with preliminary experiments performed in this study. The In preliminary experiments, the tip of the electrode was dipped in experiments performed comply with Principles of Animal Care, NIH a solution of fluorescent dye (DiI, Biotium, Hayward, CA) prior to publication no. 86–23, revised in 1985, and with the current laws of the recording, and the ganglion observed as a whole mount post mortem State of Israel. to evaluate the exact recording site. Analysis. We analyzed spiking activity one second before a Author Contributions stimulus (‘spontaneous activity’) and two seconds afterwards Conceived and designed the experiments: RG FL. Performed the (‘evoked activity’). For each cockroach and for each stimulus experiments: RG. Analyzed the data: RG FL. Contributed reagents/ type, such 3-sec recordings were divided into 20 ms time bins and materials/analysis tools: FL. Wrote the paper: RG FL.

References 1. Maye A, Hsieh CH, Sugihara G, Brembs B (2007) Order in spontaneous 14. Fouad K, Libersat F, Rathmayer W (1994) The venom of the cockroach-hunting behavior. PLoS ONE 2(5): e443. wasp Ampulex compressa changes motor thresholds: a novel tool for studying 2. Mogenson GJ, Jones DL, Yim CY (1980) From motivation to action: functional the neural control of arousal? Zoology 98: 23–34. interface between the limbic system and the motor system. Prog Neurobiol 14: 15. Haspel G, Libersat F (2003) Wasp venom blocks central cholinergic synapses to 69–97. induce transient paralysis in cockroach prey. J Neurobiol 54: 628–637. 3. Dickinson A, Balleine B (1994) Motivational control of goal directed action. 16. Moore EL, Haspel G, Libersat F, Adams ME (2006) Parasitoid wasp sting: a Anim Learn Behav 22: 1–18. cocktail of GABA, taurine, and beta-alanine opens chloride channels for central 4. Palmiter RD (2008) Dopamine signaling in the dorsal striatum is essential for synaptic block and transient paralysis of a cockroach host. J Neurobiol 66: motivated behaviors: lessons from dopamine-deficient mice. Ann N Y Acad Sci 811–820. 1129: 35–46. 17. Haspel G, Rosenberg LA, Libersat F (2003) Direct injection of venom by a 5. Fujiwara M, Sengupta PL, McIntire SL (2002) Regulation of body size and predatory wasp into cockroach brain. J Neurobiol 56: 287–292. behavioral state of C. elegans by sensory perception and the EGL-4 cGMP 18. Gal R, Rosenberg LA, Libersat F (2005) Parasitoid wasp uses a venom cocktail dependent protein kinase. Neuron 36: 1091–1102. injected into the brain to manipulate the behavior and metabolism of its 6. Lipton J, Kleemann G, Ghosh R, Lints R, Emmons SW (2004) Mate searching cockroach prey. Arch Insect Biochem Physiol 60: 198–208. in Caenorhabditis elegans: A genetic model for sex drive in a simple 19. Kien J, Altman JS (1992) Decision-making in the insect nervous system: a model invertebrate. J Neurosci 24: 7427–7434. for selection and maintenance of motor programmes. In: Kien J, McCrohan CR, 7. Briggman KL, Abarbanel HDI, Kristan WB (2005) Optical imaging of neuronal Winlow W, eds. Neurobiology of motor programme selection, Pergamon Press, populations during decision-making. Science 307: 896–901. Oxford. Pergamon studies in neuroscience, no. 4. pp 147–169. 8. Huber R (2005) Amines and motivated behaviors: a simpler systems approach to 20. Strauss R, Heisenberg M (1993) A higher control center of locomotor behaviour complex behavioral phenomena. J Comp Physiol A 191: 231–239. in the Drosophila brain. J Neurosci 13: 1852–1861. 9. Libersat F, Delago A, Gal R (2009) Manipulation of host behavior by parasitic 21. Schaefer P, Ritzmann RE (2001) Descending influences on escape behavior and insects and insect parasites. Annu Rev Entomol 54: 189–207. motor pattern in the cockroach. J Neurobiol 49: 9–28. 10. Gal R, Libersat F (2008) A parasitoid wasp manipulates the drive for walking of 22. Strauss R (2002) The central complex and the genetic dissection of locomotor its cockroach prey. Curr Biol 18: 877–882. behaviour. Curr Opin Neurobiol 12: 633–638. 11. Haspel G, Gefen E, Ar A, Glusman JG, Libersat F (2005) Parasitoid wasp affects 23. Ridgel AL, Ritzmann RE (2005) Effects of neck and circumesophageal metabolism of cockroach host to favour food preservation for its offspring. connective lesions on posture and locomotion in the cockroach. J Comp J Comp Physiol A 191: 529–534. Physiol A 191: 559–573. 12. Fouad F, Libersat F, Rathmayer W (1996) Neuromodulation of the escape 24. Gal R, Libersat F (2006) New vistas on the initiation and maintenance of insect behavior in the cockroach Periplaneta americana by the venom of the parasitic motor behaviors revealed by specific lesions of the head ganglia. J Comp wasp Ampulex compressa. J Comp Physiol A 178: 91–100. Physiol A 192: 1003–1020. 13. Libersat F (2003) Wasp uses venom cocktail to manipulate the behavior of its 25. Wessnitzer J, Webb B (2006) Multimodal sensory integration in insects - towards cockroach prey. J Comp Physiol A 189: 497–508. insect brain control architectures. Bioinspir Biomim 1: 63–75.

PLoS ONE | www.plosone.org 9 April 2010 | Volume 5 | Issue 4 | e10019 Venom Decreases Motivation

26. Mu¨ller D, Staffelt D, Fiala A, Menzel R (2003) Procaine impairs learning and 44. Pielage J, SteVes G, Lau DC, Parente BA, Crews ST, et al. (2002) Novel memory consolidation in the honeybee. Brain Res 977: 124–127. behavioral and developmental defects associated with Drosophila single-minded. 27. Devaud JM, Blunk A, Podufall J, Giurfa M, Gru¨newald B (2007) Using local Dev Biol 249: 283–299. anaesthetics to block neuronal activity and map specific learning tasks to the 45. Ritzmann RE, Pollack AJ, Archinal J, Ridgel AL, Quinn RD (2005) Descending mushroom bodies of an insect brain. Eur J Neurosci 26: 3193–3206. control of body attitude in the cockroach, Blaberus discoidalis and its role in 28. Watson JT, Ritzmann RE (1994) The escape response versus the quiescent incline climbing. J Comp Physiol A 191: 253–264. response of the American cockroach: behavioural choice mediated by 46. Ridgel AL, Alexander BE, Ritzmann RE (2007) Descending control of turning physiological state. Anim Behav 48: 476–478. behavior in the cockroach, Blaberus discoidalis. J Comp Physiol A 193: 385–402. 29. van Swinderen B (2005) The remote roots of consciousness in fruit-fly selective 47. Monastirioti M, Gorczyca M, Rapus J, Eckert M, White K, et al. (1995) attention? Bioessays 27: 321–330. Octopamine immunoreactivity in the fruit fly Drosophila melanogaster. J Comp 30. van Swinderen B, Andretic R (2003) Arousal in Drosophila. Behav Processes 64: Neurol 356: 275–287. 133–144. 48. Duch C, Mentel T, Pflu¨ger HJ (1999) Distribution and activation of different 31. Cirelli C (2009) The genetic and molecular regulation of sleep: from fruit flies to types of octopaminergic DUM neurons in the locust. J Comp Neurol 403: 119–134. humans. Nat Rev Neurosci 10: 549–560. 49. Bra¨unig P, Burrows M (2004) Projection patterns of posterior dorsal unpaired 32. Cirelli C, Bushey D (2008) Sleep and wakefulness in Drosophila melanogaster. median neurons of the locust subesophageal ganglion. J Comp Neurol 478: Ann NY Acad Sci 1129: 323–329. 164–175. 33. Greenspan RJ, Tononi G, Cirelli C, Shaw PJ (2001) Sleep and the fruit fly. 50. Sinakevitch I, Niwa M, Strausfeld NJ (2005) Octopamine-like immunoreactivity Trends Neurosci 24: 142–145. in the honey bee and cockroach: comparable organization in the brain and 34. Martin J, Ernst R, Heisenberg M (1998) Mushroom bodies suppress locomotor subesophageal ganglion. J Comp Neurol 488: 233–254. activity in Drosophila melanogaster. Learning Memory 5: 179–191. 51. Crisp KM, Mesce KA (2006) Beyond the central pattern generator: amine 35. Martin JR, Raabe T, Heisenberg M (1999) Central complex substructures are modulation of decision-making neural pathways descending from the brain of required for the maintenance of locomotor activity in Drosophila melanogaster. the medicinal leech. J Exp Biol 209: 1746–1756. J Comp Physiol A 185: 277–288. 52. Busch S, Selcho M, Ito K, Tanimoto H (2009) A map of octopaminergic 36. Kien J, Altman JS (1992) Preparation and execution of movement: parallels neurons in the Drosophila brain. J Comp Neurol 513: 643–667. between insect and mammalian motor systems. Comp Biochem Physiol A 103: 53. Rosenberg LA, Glusman JG, Libersat F (2007) Octopamine partially restores 15–24. walking in hypokinetic cockroaches stung by the parasitoid wasp Ampulex 37. Delcomyn F (1999) Walking robots and the central and peripheral control of compressa. J Exp Biol 210: 4411–4417. locomotion in insects. Auton Robots 7: 259–270. 54. Bra¨unig P (1991) Suboesophageal DUM neurones innervate the principal 38. Roeder KD (1937) The control of tonus and locomotor activity in the praying neuropiles of the locust brain. Phil Trans R Soc Lond B 322: 221–240. mantis, Mantis religiosa L. J Exp Zool 76: 353–374. 55. Eckert M, Rapus J, Nu¨rnberger A, Penzlin H (1992) A new specific antibody 39. Pearson KG, Iles JF (1973) Nervous mechanisms underlying intersegmental co- reveals octopamine-like immunoreactivity in cockroach ventral nerve cord. ordination of leg movements during walking in the cockroach. J Exp Biol 58: J Comp Neurol 322: 1–15. 725–744. 56. Dacks AM, Christensen TA, Agricola H-J, Wollweber L, Hildebrand JG (2005) 40. Graham D (1979) Effects of circum-oesophageal lesion on the behavior of the Octopamine-immunoreactive neurons in the brain and subesophageal ganglion stick insect Carausius morosus. II. Changes in walking co-ordination. Biol of the hawkmoth Manduca sexta. J Comp Neurol 488: 255–268. Cybern 32: 147–152. 57. Cholewa J, Pflu¨ger HJ (2009) Descending unpaired median neurons with 41. Bassler U (1986) Afferent control of walking movements in the stick insect bilaterally symmetrical axons in the suboesophageal ganglion of Manduca sexta Cuniculina impigra. I. Decerebrated animals on a treadband. J Comp Physiol A larvae. Zoology 112: 251–262. 158: 345–349. 58. Blagburn JM, Sattelle DB (1987) Nicotinic acetylcholine receptors on a 42. Huber F (1960) Untersuchungen uber die funtion des entralnervensystems und cholinergic nerve terminal in the cockroach Periplaneta americana. J Comp insbesondere des gehirnes bei der fortbewegung und der lauterzeugung der Physiol A 161: 215–225. Grillen. Zeit Vergleich Physiol 44: 60–132. 59. Libersat F (1994) The dorsal giant interneurons mediate evasive behavior in flying cockroaches. J Exp Biol 197: 405–411. 43. Strauss R, Hanesch U, Kinkelin M, Wolf R, Heisenberg M (1992) No-bridge of 60. Keasar T, Sheffer N, Glusman G, Libersat F (2006) Host handling: an innate Drosophila melanogaster: portrait of a structural brain mutant of the central component of foraging behavior in the parasitoid wasp Ampulex compressa. complex. J Neurogenet 8: 125–155. Ethology 112: 699–706.

PLoS ONE | www.plosone.org 10 April 2010 | Volume 5 | Issue 4 | e10019 198 Gal et al. Archives of Insect Biochemistry and Physiology 60:198–208 (2005) Parasitoid Wasp Uses a Venom Cocktail Injected Into the Brain to Manipulate the Behavior and Metabolism of Its Cockroach Prey

Ram Gal, Lior Ann Rosenberg, and Frederic Libersat*

Unlike other venomous predators, the parasitoid wasp Ampulex compressa incapacitates its prey, the cockroach Periplaneta americana, to provide a fresh food supply for its offspring. We first established that the wasp larval development, from egg laying to pupation, lasts about 8 days during which the cockroach must remain alive but immobile. To this end, the wasp injects a cocktail of neurotoxins to manipulate the behavior of the cockroach. The cocktail is injected directly into the head ganglia using biosensors located on the stinger. The head sting induces first 30 min of intense grooming followed by hypokinesia during which the cockroach is unable to generate an escape response. In addition, stung cockroaches survive longer, lose less water, and consume less oxygen. Dopamine contained in the venom appears to be responsible for inducing grooming behavior. For the hypokinesia, our hypothesis is that the injected venom affects neurons located in the head ganglia, which send descending tonic input to bioaminergic neurons. These, in turn, control the thoracic premotor circuitry for locomotion. We show that the activity of identified octopaminergic neurons from the thoracic ganglia is altered in stung animals. The alteration in the octopaminergic neurons’ activity could be one of the mechanisms by which the venom modulates the escape circuit in the cockroach’s central nervous system and metabolism in the peripheral system. Arch. Insect Biochem. Physiol. 60:198–208, 2005. © 2005 Wiley-Liss, Inc.

KEYWORDS: parasitoid wasp; cockroach; Ampulex compressa; Periplaneta americana; venom; neurotoxins; paralysis; grooming; biogenic amines

INTRODUCTION lyzed host. Wasps of the family Pompilidae, for example, paralyze a spider with multiple variable Most exoparasitoid wasps belong to the fami- stings, then drag their victim to a prepared burrow lies Pompilidae, Sphecidae, Mutillidae, and Bethyl- and deposit one egg on the spider’s abdomen idae (Rathmayer, 1978). These venomous wasps (Steiner, 1986). use terrestrial arthropods as food supply for their In those species of Sphecidae where the para- offspring. Most wasps paralyze their prey, drag it lyzing venom is injected through the stinger into to a burrow, and lay one egg on the prey item. the hemolymph of the prey, as in the beewolf When the wasp larva hatches, it feeds on the para- (Philanthus triangulum), the venom, which contains

Department of Life Sciences, Ben-Gurion University, Beer-Sheva, Israel Presented at the 2004 International Congress of Entomology Symposium entitled “Physiological and Behavioral Host-Parasite Interactions,” Brisbane, Australia. Abbreviation used: SEG = subesophageal ganglion. Contract grant sponsor: United States-Israel Bi-National Science Foundation (BSF); Contract grant numbers: 96/00472, 2001044. *Correspondence to: Dr. F. Libersat, Dept. of Life Sciences, Ben-Gurion University of the Negev, P.O. Box 653, Beer Sheva, Israel 84105. E-mail: [email protected]

© 2005 Wiley-Liss, Inc. Archives of Insect Biochemistry and Physiology DOI: 10.1002/arch.20092 Published online in Wiley InterScience (www.interscience.wiley.com) Wasp’s Venom Changes Cockroach Physiology 199 philanthotoxins, interferes presynaptically with the wasp from egg to pupation requires the preserva- release of glutamate and blocks the postsynaptic tion of the vital functions of the envenomated host glutamate receptors at the nerve-muscle junction for at least 7 days. To achieve this, we show that (Rathmayer, 1962; Piek and Spanjer, 1986; Elde- the wasp injects its venom directly with precision frawi et al., 1988). The impairment of neuromus- into the head ganglia of the host using cuticular cular transmission results in muscular flaccid sensors on the stinger for guidance. Then, we as- paralysis of the insect prey. sess the effect of the sting on the behavior of the Within the Sphecidae family, a few species do host and unravel some of its neuronal correlates. not paralyze but manipulate the behavior of their victims in most interesting ways. Of particular in- WASP’S LARVA LIFE CYCLE terest are parasitoid wasps that prey on insects from the Orthoptera and Dictyoptera families (cock- The feeding of the wasp’s larva lasts several days roaches, crickets, mantids, and grasshoppers; before the host dies and the larva pupates. Thus, Gnatzy, 2001; Fouad et al., 1994; Piek, 1990). The unlike many other venomous predators that con- wasp stings its prey, inducing various behavioral sume prey immediately, the developmental time- states, and pulls it to a burrow where it lays one table of A. compressa offspring requires that the host egg on the cuticle surface. The larva develops out- be preserved after envenomization. To determine side of the prey, feeding on the hemolymph after this timetable, the development of Ampulex compressa perforating the cuticle. It later moves inside the larvae was followed visually from egg to pupa on prey’s body to feed and pupate for completing its parasitized cockroaches (Haspel et al., 2005). Mor- development. phological and positional changes were recorded The Sphecid wasp Ampulex compressa applies a daily and imaged with a digital camera. After sting- unique strategy of behavioral modulation of its ing the cockroach, the wasp lays an egg and affixes cockroach prey (Williams, 1942; Piek et al., 1984; it on the cuticle of the coxal segment of the met- Fouad et al., 1994). This parasitoid solitary wasp athoracic leg (Fig. 1A,B). A larva hatches from the hunts cockroaches (Periplaneta americana) by sting- egg within 2 to 3 (mean ± SD; 2.3 ± 0.5, n = 14) ing them first in the thorax and then in the head. days and perforates the cuticle of the cockroach’s The stung cockroach exhibits three consecutive coxa to feed on the hemolymph for the next few phases of envenomation. First, the cockroach shows days (Fig. 1C). About five (5.5 ± 0.5, n = 14) days a transient paralysis of its front legs lasting only 1– after the egg was laid (Fig. 1D), the larva moves to 2 minutes (Fouad et al., 1994; Haspel and Libersat, the thoraco-coxal junction of the metathoracic leg 2003). Then it grooms extensively, after which it be- and bites a large hole along the soft cuticular joint comes sluggish and is not responsive to various until it penetrates the cockroach. It then feeds on stimuli (Fouad et al., 1994; Weisel-Eichler et al., the internal organs of the host for about 3 days and 1999; Weisel-Eichler and Libersat, 2002). The wasp pupates inside the cockroach abdomen (Fig. 1E). grabs the cockroach’s antenna and walks it to a suit- Thus, pupation occurs roughly eight (8.1 ± 0.4, n = able oviposition location. The cockroach follows the 14) days after the egg has been laid. Roughly 5 weeks wasp in a docile manner like a dog on a leash (Wil- later (40 ± 6 days, n = 33), a mature wasp hatches liams, 1942; Fouad et al., 1994). A few days later, from the cocoon (Fig. 1F), exits the cockroach host’s the cockroach serves as an immobilized and fresh abdomen, and leaves the burrow. food source for the wasp’s offspring. The aim of the present review is to provide in- WASP INJECTS VENOM DIRECTLY sights into the neuronal mechanisms by which the INTO THE PREY’S HEAD GANGLIA wasp Ampulex compressa manipulates the behavior of its host, the cockroach Periplaneta americana. The unique effects of Ampulex’s venom on prey First, we show that the developmental time of the behavior and the site of venom injection both sug-

December 2005 200 Gal et al.

Fig. 1. Development of Ampulex compressa larva. A: A egg was laid, the larva is ready to penetrate the cockroach. single egg (arrow) laid by A. compressa on a stung cock- E: The larva consumes the internal organs of the cockroach. B: The egg is glued to the middle coxa of a cockroach and pupates inside the cockroach, which dies at this roach (higher magnification of A), which is not reachable stage. The pupa fills most of the cockroach’s abdomen. F: by the cockroach. C: A larva hatches within about two Roughly five weeks later, a mature wasp hatches from the days. D: The larva bites a hole in the cockroach’s coxa cockroach abdomen (Modified from Haspel et al., 2005). and feeds on hemolymph, and about six days after the gest that the venom targets the insect’s central ner- giques, Jean Henri Fabre suggested that specific vous system (Fig. 2A). Until recently, the mecha- wasps sting directly into target ganglia (Fabre, nism by which behavior-modifying compounds 1879) because these wasps sting in a pattern that in the venom reach the central nervous system, corresponds to the location and arrangement of given the protective ganglionic sheath around the nerve centers in the prey. Others challenged ganglia, was unknown. Hence, for more than a Fabre’s idea and claimed that the wasps sting in century, entomologists have agued over whether the vicinity but not into the ganglion (Ferton, or not several species of parasitoid wasps deliver 1902; Roubaud, 1917). To resolve this controversy, their venom by stinging directly into the prey’s we have produced so-called “hot” wasps after in- nervous system. In his book Souvenirs Entomolo- jecting them with a mixture of C14 radiolabeled

Archives of Insect Biochemistry and Physiology Wasp’s Venom Changes Cockroach Physiology 201

Fig. 2. Ampulex compressa stings directly into the head ally injected into the head cavity of control cockroaches ganglia of cockroaches. A: The wasp stings a cockroach in (open bars, n = 15), the levels of radiolabeled venom are the head. B: A micrograph of the stinger (st) is shown significantly higher in non-neuronal head tissue than in- over a schematic lateral view of a cockroach head drawn side the head ganglia (P < 0.01). The measurements are to scale. The wasp stinger is long enough (2.5 ± 0.2 mm; represented as the percentile fraction of the total CPM n = 5) to reach the brain (br), which lies 1 to 2 mm deep (counts per minute) of a specimen. D,E: Two sections of in the head capsule. s: subesophageal ganglion. Scale bar: representative head ganglia (brain and SEG) preparations 0.5 mm. C: Localization of radiolabeled venom by liquid of a cockroach stung by a radiolabeled wasp. Radiolabeled scintillation. In stung cockroaches (solid bars, n = 16), venom is located posterior to the central complex and the levels of radiolabeled venom are significantly higher around the mushroom bodies of the brain (D) and around in the head ganglia (brain and SEG) than in non-neu- the center of the SEG (E). Scale bar = 0.25 mm (modified ronal head tissue. When radioactive amino acids are manu- from Haspel et al., 2003). amino acids, which were incorporated into the 2003). Moreover, when examining the head gan- venom. In cockroaches stung by radiolabeled glia, we found a high concentration of radioac- wasps, we found most of the radioactive signal in tive signal in the central part of the brain (Fig. the thoracic ganglion and inside the head gan- 2D), posterior to the central complex and around glia (Fig. 2C): the brain and subesophageal gan- the mushroom bodies, as well as around the mid- glion (SEG). Only a small amount of radioactivity line of the SEG (Fig. 2E). The wasp stinger, which was detected in the surrounding, non-neuronal is about 2.5 mm in length, is long enough to tissue in the head and thorax (Haspel et al., reach the brain, which lies 1 to 2 mm deep in

December 2005 202 Gal et al. the head capsule (Fig. 2B). The precise anatomi- stinger (Fig. 3A; Gal et al., 2003). To investigate cal targeting of the wasp stinger through the body whether the wasp is able to discriminate nervous wall and ganglionic sheath and into specific ar- from non-nervous tissue, we compared the stinging eas of the brain is akin to the most advanced ste- durations for wasps stinging sham-operated cock- reotactic delivery of drugs (Haspel et al., 2003). roaches with no neuronal lesions to those of brain- We hypothesized that the wasp locates the brain less cockroaches, in which the brain was removed with specialized sensors on its stinger. These prior to the sting. The stinging duration increased would be used to discriminate nervous tissue from 15-fold when the wasps stung brainless cockroaches non-nervous tissue inside the cockroach head cap- (Fig. 3B). Furthermore, using radiolabeled wasps, sule. We identified a few types of sensilla-like cu- we showed that little or no venom is injected into ticular structures on the wasp’s stinger, one of the head cavity of brainless cockroaches compared which was mostly concentrated on the tip of the to sham-operated cockroaches (Gal et al., 2003).

Fig. 3. A: Scanning Electron Micrograph showing sensilla- like cuticular structures (arrow), located mainly on the distal third of the stinger. Scale bars (in µm, left to right): 500, 50, 10, 2. B: When the cockroach brain is removed prior to the sting, the wasp searches for its nervous target inside the cockroach head cavity significantly longer, compared to sham-operated cockroaches (shaded bars, P < 0.001; Mann- Whitney Rank Sum Test). In contrast, there is no significant change in the duration of the sting directed at the thorax, where nervous tissue remained intact (solid bars, P = 0.529, t- test). Bars represent mean ± S.D; n = 8 for each group.

Archives of Insect Biochemistry and Physiology Wasp’s Venom Changes Cockroach Physiology 203

Thus, while stinging the cockroach, A. compressa neurons. This antibody has been shown to label appears to use sensory feedback from specialized specifically dopaminergic neurons (for example, cuticular sense organs located on its stinger to iden- see Nassel and Elekes 1992). It was found that, tify the brain of its prey. indeed, the SEG houses a rather large population of dopaminergic neurons, some of which produce WASP VENOM INJECTION extensive terminal branches within this ganglion INDUCES GROOMING (Fig. 4C). Flupenthixol, an antagonist of dopamine receptors in the cockroach brain (Notman and Cockroaches stung by the wasp into the head Downer, 1987; Orr et al., 1992), greatly reduced ganglia groom almost continuously during the 30 venom-induced grooming when injected into the min following the recovery from the transient pa- cockroach’s hemolymph prior to the wasp sting ralysis of the front legs (Fig. 4A; Weisel-Eichler et (Fig 4D). In contrast, mianserin, which has been al. 1999). Grooming is evoked only when venom found to be an effective octopamine receptor an- is injected into the head ganglia, and cannot be tagonist in locust brain and cockroach nerve cord attributed to the stress of the attack, contact with (Roeder, 1992; Orr et al., 1992), did not reduce the wasp, mechanical irritation, or venom injec- venom-induced grooming at all (Fig. 4D). In fact, tion into a location other than the head (Fig. 4B). using Gas Chromatography-Mass Spectrometry, we The association of neuromodulatory systems identified a dopamine-like substance in the venom with specific behaviors has been well established (Fig. 4E; Weisel-Eichler et al., 1999). Thus, dopam- in invertebrates, in particular for the monoamines, ine in the venom is likely to be the component which have been found to modulate the release of that induces prolonged grooming. To the best of well-defined behaviors (Bicker and Menzel, 1989; our knowledge, this is the first report in the litera- Libersat and Pflüeger, 2004). For instance, mono- ture of a venom injection via a sting into the host amines have been found to activate behaviors, such central nervous system that elicits a specific behav- as stridulation in the grasshopper (Ocker et al., ioral pattern in the host. 1995) or flight in the moth and locust (Sombati and Hoyle, 1984; Claasen and Kammer, 1986; WASP VENOM INJECTION INDUCES LONG-TERM Stevenson and Kutsch, 1987). Because monoam- HYPOKINESIA AND CHANGES IN METABOLISM ines are known to initiate specific behaviors in insects, we explored the possibility that the venom Besides grooming behavior, the sting in the acts via a monoaminergic system to evoke groom- head ganglia causes hypokinesia, which com- ing behavior. When a single dose of 10 µl of reser- mences within 30 min and lasts, if the cockroach pine (1 mM), which elevates the concentration of is not parasitized by a wasp larva, for about 3 monoamines, was injected into the SEG, cock- weeks. Stung cockroaches show very little sponta- roaches started to groom intensively (Weisel-Eichler neous or provoked activity such as escape (Fouad and Libersat, 2002). Moreover, injection of dopam- et al., 1994; Libersat et al., 1999). In contrast, our ine or a dopamine agonist (SKF 82958: a verte- results did not show any differences between stung brate D1 agonist that has been successfully used and control cockroaches in spontaneous or pro- on invertebrates (for example, see Keating and Or- voked grooming, righting behavior, or ability to chard, 2004)) induced prolonged grooming, thus fly in a wind tunnel (Weisel-Eichler and Libersat, demonstrating that dopamine alone is sufficient 2002). Hence, the head sting affects specific mo- to induce grooming (Weisel-Eichler et al., 1999). tor behaviors while leaving others untouched. As To check for the presence of a natural source of has been mentioned above, such specificity could dopamine in the SEG, a monoclonal fluorescent be achieved by targeting a specific neuromod- antibody specific for an enzyme in the dopamine ulatory system that regulates the initiation and/or synthesis pathway was used to label dopaminergic execution of a single behavior or a specific subset

December 2005 204 Gal et al.

Fig. 4. Venom-induced grooming in cockroaches. A: which have axons branching extensively in the SEG while Stung cockroach performing two frequent components of others send their axons to the brain or the thorax. D: Cock- grooming behavior: grooming an antenna (left) and roaches that received flupenthixol, a dopamine (DA) re- grooming a foreleg with the mouthparts (middle and ceptor antagonist, before a sting groom significantly less right). B: Cockroaches that received a full stinging se- (***P < 0.001) than cockroaches that received saline or quence by the wasp groom for 23.0 ± 2.3 min during the mianserin, an octopamine (OA) antagonist, before the 30 min following the sting. This grooming duration is sig- sting. E: Mass spectrogram of the large venom peak eluted nificantly longer (**P < 0.01) than that observed in cock- at 12.68 min of the Gas Chromatography (not shown); roaches stung only in the thorax and then punctured in this spectrum is comparable to the mass spectrograms (in- the SEG with a pin (7.8 ± 5.4 min). C: Dorsal and lateral set) of dopamine. X-axes indicate mass/charge (modified views of an SEG stained with Tyrosine Hydroxylase anti- from Weisel-Eichler et al., 1999). body reveal a group of dopaminergic neurons, some of of behaviors. Consistent with this hypothesis, re- impairment in the ability of cockroaches and crick- serpine, a plant alkaloid, which depletes the syn- ets to generate escape behavior (Weisel-Eichler and aptic content of monoaminergic neurons, induces Libersat, 2002). Moreover, a similar impairment in

Archives of Insect Biochemistry and Physiology Wasp’s Venom Changes Cockroach Physiology 205 their escape behavior is observed in crickets de- ines (Casagrand and Ritzmann, 1992). Our hy- pleted of octopamine and dopamine (Stevenson pothesis is that the injected venom affects neurons et al., 2000). This suggests a role of dopamine or located in the brain, which send descending tonic octopamine as chemical modulators of escape be- input to monoaminergic neurons (Fig. 5). These, havior in cockroaches. In insects, locomotion is in turn, control the thoracic premotor circuitry. generated in the thoracic ganglia and modulated Octopamine is a monoamine that is secreted by by the head ganglia (Kien and Altman, 1992). The octopaminergic neurons in the prey’s central ner- wasp injects its cocktail of neurotoxins into the vous system. It is known to regulate the expres- cockroach’s brain to induce a long-lasting change sion of specific motor patterns by modulating the in the initiation and execution of locomotory be- excitability of specific neurons (Bräunig and haviors. For instance, the thoracic escape-running Pflüger, 2001). neuronal circuit of a stung animal is impaired at We have examined the descending control ex- the level of central synapses between thoracic in- erted on the octopaminergic neurons (Rosenberg terneurons to motoneurons (Fig. 5). These central and Libersat, 2004). This was approached by re- synapses are known to be modulated by monoam- cording the activity of octopaminergic neurons af-

Fig. 5. Neuromodulation of cockroach escape circuitry during long-term hypokinesia. A model of wasp venom induced long term-hypokinesia. Schematic drawing of the cockroach’s nervous system and escape neuronal circuit. Sensory mechanoreceptors of the cerci (1) recruit ascending giant interneurons (2); these giant interneurons together with brain descending interneurons (3) converge onto the thoracic interneurons (4). The thoracic interneurons excite directly or via local interneurons (5) the leg motoneurons (6) involved in fast leg movements. Neurons (7) in the head ganglia provide descending permissive input to neuromodulatory neurons (8) located in the thoracic ganglia. These modulate the synapses between the thoracic interneurons and specific motoneurons. The venom represses the activity of head ganglia neurons, thereby removing the descending excitatory drive to the neuromodulatory neurons.

December 2005 206 Gal et al. ter a wasp sting into the head. We show that the manipulates these neuromodulatory systems in the activity of octopaminergic neurons from a thoracic head ganglia to modify the host’s metabolism and ganglion is altered in stung animals. The sponta- increase the survival of the host, which remains neous firing rate of octopaminergic neurons in fresh while the larva feeds until it is ready to pu- stung animals is approximately 5-fold lower than pate. Hence, with a single but precise sting into in control animals (Fig. 6A). They also respond the head ganglia, the wasp not only converts the less to wind stimuli (Fig. 6B). Alteration in octopa- cockroach into a submissive prey but also alters minergic neurons’ activity could be one of the its metabolism, apparently to the larva’s benefit. mechanisms by which the venom modulates the cockroach escape circuit in the central nervous sys- CONCLUSIONS tem. We propose that the venom injected into the head ganglia removes a descending excitatory in- The specificity and effectiveness of neurotox- put to these neuromodulatory systems, thereby de- ins are the outcome of evolutionary selection on creasing the excitability of specific motor behaviors. one animal’s strategy to incapacitate another Last but not least, stung cockroaches survive (Adams and Olivera, 1994). Here we highlight the longer, lose less water, and consume less oxygen selection of an amazing behavioral strategy by a (Haspel et al., 2005). We hypothesize that the sting venomous insect predator for the delivery of

Fig. 6. Modulation of octopaminergic neurons’ activity in stung animals. A: The top trace is an example of an intracellular recording from an octopaminergic neuron and the bottom trace is the extracellular recording from the abdominal ventral nerve cord in two different experimental groups. Notice the decrease in the spontaneous activity of the octopaminergic neuron in stung compared with control animals. B: A cercal stimulus (square pulse) evoked strong responses in the nerve cord (middle trace) and consequently strong responses in the octopaminergic neuron of control animals. Such a stimulus did not evoke a response in the octopaminergic neuron of stung animals (Note the different time scales in A and B).

Archives of Insect Biochemistry and Physiology Wasp’s Venom Changes Cockroach Physiology 207 neurotoxins into the central nervous system of an- LITERATURE CITED. other insect prey to cause specific and effective behavioral modifications. The unique effect of the Adams ME, Oliviera B. 1994. Neurotoxins: overview of an wasp’s venom on the cockroach locomotory be- emerging research technology. Trends Neurosci 17:151–155. havior contrasts markedly with the paralyzing ac- Bicker G, Menzel R. 1989. Chemical codes for the control of tions of other known solitary wasp venoms, which behaviour in arthropods. Nature 337:33–39. interfere at the neuromuscular junction. Unlike these other species of wasps, Ampulex compressa’s Bräunig P, Pflüger HJ. 2001. The unpaired median neurons venom has no paralytic effect on the cockroach of insects. Adv Insect Physiol 28:185–266. prey’s muscular system (Fouad et al., 1996). Thus, Claasen DE, Kammer AE. 1986. Effects of octopamine, because of its specific behavioral effects, one could dopamine, and serotonin on production of flight motor envisage that Ampulex’s neurotoxins might have output by thoracic ganglia of Manduca sexta. J Neurobiol interesting novel effects on the excitability of neu- 17:1–14. rons or on synaptic transmission. Chemist engi- neers can generate hundreds of neurotoxins in Casagrand JL, Ritzmann RE. 1992. Biogenic amines modu- their labs, but these products are random and of- late synaptic transmission between identified giant inter- ten useless, whereas any natural neurotoxin has neurons and thoracic interneurons in the escape system of the cockroach. J Neurobiol 23:644–655. already passed the ultimate screening test: over a few millions years of co-evolution. Thus, one of Eldefrawi AT, Eldefrawi ME, Konno K, Mansour NA, Nakanishi our main goals is to identify the venom compo- K, Oltz E, Usherwood PNR. 1988. Structure and synthesis nents and their corresponding molecular targets of a potent glutamate receptor antagonist in wasp venom. that are involved in these behavioral modifica- Proc Natl Acad Sci USA 85:4910–4913. tions. By studying the effect of the wasp’s venom Fabre JH. 1879. Souvenirs entomologiques, vol. 1. Paris: on its cockroach host, we hope to further our un- Delagrave (1945 ed). p 108–112. derstanding of the neuronal basis of parasite- induced alterations of host behavior and the Ferton C. 1902. Notes detachees sur l’ instinct des hymen- neurobiology of initiation of motor behaviors in opteres melliferes et ravisseurs II. Ann Soc Entomol Fr 71:499–531. insects. Fouad K, Libersat F, Rathmayer W. 1994. The venom of the ACKNOWLEDGMENTS cockroach-hunting wasp Ampulex compressa changes motor thresholds: a novel tool for studying the neural con- The work described in this manuscript was re- trol of arousal? Zoology 98:23–34. ported at the XXII International Congress of En- Fouad K, Libersat F, Rathmayer W. 1996. Neuromodulation tomology (August 2004; Brisbane, Australia). It of the escape behavior of the cockroach Periplaneta was supported by a grant 96/00472 and is cur- americana by the venom of the parasitic wasp Ampulex rently supported by grant 2001044 from the compressa. J Comp Physiol A 178:91–100. United States-Israel Bi-national Science Founda- Gal R, Haspel G, Libersat F. 2003. Wasp uses specialized sen- tion (BSF) to F. Libersat. We thank Jose Gustavo sors to probe and inject venom inside the brain of its cock- Glusman and Gal Haspel for their valuable assis- roach brain. Neural Plasticity 10:200. tance and two anonymous reviewers for valuable comments on the manuscript. All experiments re- Gnatzy W. 2001. Digger wasp vs. Cricket: (Neuro-) Biology ported in this review comply with the “Principles of a predator-prey-interaction. Zoology 103:125–139. of Animal Care,” publication No. 86-23 (revised Haspel G, Libersat F. 2003. Wasp venom blocks central cho- 1985) of the National Institute of Health and also linergic synapses to induce transient paralysis in cockroach with the current laws of the State of Israel. prey. J Neurobiol 54: 628–637.

December 2005 208 Gal et al.

Haspel G, Rosenberg LA, Libersat F. 2003. Direct injection of tary wasp venoms. In: T. Piek, editor. Venoms of the Hy- venom by a predatory wasp into cockroach brain. J menoptera. London: Academic Press. p 161–307. Neurobiol 56:287–292. Piek T. 1990. Neurotoxins from venoms of the Hymenoptera: Haspel G, Gefen E, Ar A, Glusman JG, Libersat F. 2005. Para- twenty-five years of research in Amsterdam. Comp Bioc- sitoid wasp affects metabolism of cockroach host to favor hem Physiol C 96:223–233. food preservation for its offspring. J Comp Physiol 191(6): 529–534. Rathmayer W. 1962. Paralysis caused by the digger wasp Philanthus. Nature (Lond) 196:1148–1151. Keating C, Orchard I. 2004. The effects of dopamine agonists and antagonists on the secretory responses in the Rathmayer W. 1978. Venoms of Sphecidae, Pompilidae, salivary glands of the locust (Locusta migratoria). J Insect Mutillidae and Bethylidae. In: Bettini S, editor. Handbook Physiol 50:17–23 of experimental pharmacology, Vol. 48. Arthropod venoms. Heidelberg: Springer. p 661–690. Kien J, Altman JS. 1992 Decision-making in the insect nervous system: a model for selection and maintenance of Roeder T. 1992. A new octopamine receptor class in locust motor programmes. In: Kien J, McCrohan CR, Winlow W, nervous tissue, the octopamine 3 (OA3) receptor. Life Sci editors. Neurobiology of motor programme selection. New 50:21–28. York: Pergamon Press. Pergamon studies in neuroscience, Rosenberg LA, Libersat F. 2004. Predatory wasp’s venom in- No. 4. p 147–169. duces long-term lethargy in prey by modulating central Libersat F, Pflüeger HJ. 2004. Monoamines and the orches- neuronal circuits. Soc Neurosci Abst 555.15. tration of behavior. BioScience 54:17–25. Roubaud E. 1917. Le venin et l’ evolution paralysante chez Libersat F, Haspel G, Casagrand J, Fouad K. 1999. Localiza- les hymenopteres predateurs. Bull Biol Fr Belg 51:391–419. tion of the site of effect of a wasp’s venom in the cockroach escape circuitry. J Comp Physiol A 184:333–345. Sombati S, Hoyle G. 1984. Generation of specific behaviors in a locust by local release into neuropil of the natural Nassel DR, Elekes K. 1992. Aminergic neurons in the brain neuromodulator octopamine. J Neurobiol 15:481–506. of blowflies and Drosophila: dopamine- and tyrosine hydroxylase-immunoreactive neurons and their relationship Steiner AL. 1986. Stinging behaviour of solitary wasps. In: with putative histaminergic neurons. Cell Tissue Res Piek T, editor. Venoms of the Hymenoptera. London: Aca- 267:147–67. demic Press. p 63–160.

Notman HJ, Downer RGH. 1987. Binding of [3H]pifluthixol, Stevenson PA, Kutsch W. 1987. A reconsideration of the cen- a dopamine antagonist, in the brain of the American tral pattern generator concept for locust flight. J Comp cockroach, Periplaneta americana. Insect Biochem 17: Physiol A 161:115–130. 587–590. Stevenson PA, Hofmann HA, Schoch K, Schildberger K. 2000. Ocker W, Hedwig B, Elsner N. 1995. Pharmacological induc- The fight and flight responses of crickets depleted of bio- tion and modulation of stridulation in two species of genic amines. J Neurobiol 43:107–120. acridid grasshoppers. J Exp Biol 198:1701–1710 Weisel-Eichler A, Libersat F. 2002. Are monoaminergic sys- Orr N, Orr GL, Hollingworth RM. 1992. The Sf9 cell line as tems involved in the lethargy induced by a parasitoid wasp a model for studying insect octopamine-receptors. Insect in the cockroach prey? J Comp Physiol A 188:315–324. Biochem Mol Biol 22:591–597. Weisel-Eichler A, Haspel G, Libersat F. 1999. Venom of a para- Piek T, Visser JH, Veenendaal RL. 1984. Change in behaviour sitoid wasp induces prolonged grooming in the cockroach. of the cockroach, Periplaneta americana, after being stung J Exp Biol 202:957–964. by the sphecid wasp Ampulex compressa. Entomol Exp App Williams FX. 1942. Ampulex compressa (Fabr.), a cockroach- 195:195–203. hunting wasp introduced from New Caledonia into Ha- Piek T, Spanjer W. 1986. Chemistry and pharmacology of soli- waii. Proc Haw Entomol Soc 11:221–233.

Archives of Insect Biochemistry and Physiology 3. Unpublished work: Sensory mechanisms mediating host CNS localization

During the head sting, the Jewel Wasp injects venom directly and precisely inside the cerebral ganglia of its cockroach prey (Figure 2). One goal I pursued during my studies was to identify the physiological mechanisms which enable the wasp to locate and inject venom precisely into the neuronal targets inside the cockroach head cavity. The investigation presented below has indeed brought me closer to understanding the physiological mechanism of this localization process, however an unexpected decline in wasp population has prevented me, as yet, from concluding this series of experiments. Thus, although published in part in abstract form [Gal et al. 2003, Gal and Libersat 2004], these experiments have not yet been published in peer-reviewed journals. Nonetheless, I strongly believe this investigation is important for the understanding of the unique parasite-host interaction studied in the Libersat lab. In this chapter, I will describe the rationale and methodology of these experiments and discuss the results obtained thus far.

3.1. Background In several species of parasitoid wasps, the ovipositor, which also serves as the venom- delivery device (i.e., the ‘stinger’), is known to possess important sensory roles in host parasitization. Sensory organs ('sensilla') on the stinger are involved in host localization, for example, when the host is buried in the surrounding substrate, and in the detection of chemical factors in the host hemolymph that mediate host recognition and acceptance [Gutierrez 1970, King and Rafai 1970, Ganesalingam 1974, Van Baaren and Nenon 1994]. Different mechanical cues, as well as specific chemical compounds in the host hemolymph (including ions, amino acids and polypeptides), have been identified as stimuli for stinger sensilla to induce oviposition in Hymenopterans [Arthur et al. 1969, Hermann and Douglas 1976, Nettles et al. 1982, Kainoh and Brown 1994, van Lenteren et al. 2007]. High-resolution morphological and ultrastructural examinations, complemented by behavioral and physiological studies, have all been used to identify these sensory organs and, possibly, their mode of action [Ganesalingam 1972, Hawke et al. 1973, Greany et al. 1977, LeRalec and Wajnberg 1990, Brown and Anderson 1998, Bleeker et al. 2004, van Lenteren et al. 2007].

~ 21 ~

In Ampulex compressa, venom injection into the cerebral ganglia is crucial for behavioral manipulation of the cockroach host, and thus for successful parasitization. To locate the cerebral ganglia inside the cockroach head cavity, the wasp could theoretically use (1) a pre-programmed trajectory by which the stinger reaches directly into the cerebral ganglia upon penetrating the head capsule, or (2) sensory organs on the stinger to mediate recognition of ganglia inside the head capsule. Preliminary studies on the wasp stinger, conducted by Dr. Gal Haspel (a former graduate student in our lab) and myself, have revealed axons ascending from the tip of the stinger to the terminal abdominal ganglion (TAG) of the wasp’s ventral nerve cord. Furthermore, we have found that removing the cockroach brain or SEG prior to a sting, as well as freezing the tip of the wasp stinger with liquid nitrogen, increased the wasp’s stinging duration [Gal, Haspel and Libersat, unpublished results; also see Gal et al. 2005, Gal and Libersat 2010]. These observations suggest that the wasp uses sensory feedback from its stinger to discriminate nervous from non-nervous tissue inside the cockroach head capsule, rather than a pre-programmed trajectory. A brief preliminary study led by Haspel, in which a Scanning Electron Microscope (SEM) was used to observe the surface of the stinger, revealed clues for cuticular structures on the tip which resemble sensory organs. The purpose of my project was to continue and extend these preliminary investigations.

3.2. Materials and Methods Overview: Since the wasp cannot use visual cues to locate the brain and SEG inside the cockroach head capsule, I hypothesized that it uses specialized sensilla on its stinger to discriminate nervous from non-nervous tissue inside the cockroach head cavity. To test this hypothesis, I first determined behaviorally whether the head sting involves sensory feedback from the stinger. Next, I examined the external morphology of the stinger in detail with a Scanning Electron Microscope (SEM) to identify possible sensory organs. Once these were found, I used Transmission Electron Microscope (TEM) to characterize the ultrastructure of these sensilla. Lastly, I characterized the physiological responses of the stinger sensilla to different mechanical and chemical cues applied to the tip of the stinger, using electrophysiological recordings from (1) afferents ascending from the stinger and from (2) the ventral nerve cord.

~ 22 ~

Animals: Wasps Ampulex compressa Fabricius (Hymenoptera: Ampulicidae) were reared in plastic boxes (40x50x50 cm) at 30°C ambient temperature and 60% humidity, on a 12h:12h L:D cycle. Cockroaches Periplaneta americana were reared in crowded conditions in plastic boxes (50x50x70 cm) at 26°C ambient temperature and on a 12h:12h L:D cycle. Cockroaches used for experiments were adult males collected individually from the colony after the last molt and then aged 1-4 weeks. All animals were provided with food (honey for wasps, cat chow for cockroaches) and water ad libitum.

Behavioral experiments and cerebral lesions: To investigate whether the wasp can discriminate nervous from non-nervous tissue inside the cockroach head capsule, I measured the duration of the thoracic and head stings. Wasps were introduced to four different groups of cockroaches (n=8 for each group): 1. Cockroaches with the brain surgically removed prior to the sting; 2. Cockroaches with the SEG surgically removed prior to the sting; 3. Sham-operated cockroaches; 4. Cockroaches with the neck connectives cut caudal to the SEG, but from which no neuronal tissue was physically removed. The rationale behind this last control group was to eliminate the possibility that the wasp uses behavioral cues from the cockroach (e.g. leg movements or wing flapping) to identify when the stinger is in its appropriate position inside the head.

To surgically remove the cockroach brain from the head cavity prior to the sting, the cockroach was anesthetized with carbon dioxide and a flap was opened in the dorsal head cuticle to remove parts of the vertex and frons (but not the fenestra), as described in Figure 4. Next, fine micro-scissors were used to severe the optic, ocellar and antennal nerves, as well as the circumesophageal connectives which connect the brain and SEG. The brain was then pulled out and completely removed from the head cavity and the cuticular flap was replaced and allowed to close by hemolymph coagulation. Sham operations for this group of cockroaches included the entire surgical procedure (i.e., opening the cuticular flap and cutting the peripheral nerves) but excluding the circumesophageal connectives cut and brain removal. To surgically remove the SEG from the head cavity, the cockroach was anesthetized and a C-shaped flap was opened in the ventral head cuticle, in the areas where the softer head and

~ 23 ~

neck cuticle fuses with the submentum (Figure 4). The neck muscle, which runs ventral to the nervous system, was very gently pulled aside to expose the SEG and the neck connectives. Then, the neck connectives were severed with fine micro-scissors, followed by severing of the circumesophageal connectives, to allow complete removal of the SEG from the head capsule. The neck muscle was then returned and the flap closed and allowed to heal by hemolymph coagulation. Sham operations for this group included the entire surgical procedure but without cutting the circumesophageal and neck connectives or removing the SEG. In the ‘neck connectives cut’ group, a similar procedure to the ‘SEG removed’ group was performed but only the neck connectives were severed, with no neuronal tissue physically removed from the head cavity.

For cockroaches in which the SEG was removed or the neck connectives cut prior to a sting, it was important to avoid the formation of scar tissue in certain areas around the submentum. This is because the wasp typically penetrates the cockroach’s head capsule through the soft cuticle around the submentum which covers and protects the SEG. A thick scar tissue formed due to hemolymph coagulation around the opened cuticular flap would not enable the wasp to penetrate through the otherwise soft ventral cuticle, possibly damaging the delicate stinger. To avoid this scenario and to enable a normal head sting, I performed a C-shaped incision in the submentum which was then opened laterally, leaving the cuticle on one side of the submentum intact for the stinger to penetrate through (Figure 4). When introducing the cockroach to a wasp, I used a plastic cup to encourage the wasp to approach and attack the cockroach from the intact side of the submentum. Such stings appeared entirely normal, with stinging durations similar to that applied to untreated cockroaches. Stings in which the wasp approached and attacked the cockroach from the scarred side, by contrast, were significantly prolonged and were omitted from the results presented here.

~ 24 ~

Electron microscopy of the stinger: I characterized the external morphology of the wasp’s stinger with a Scanning Electron Microscope (SEM), using a preparation protocol similar to that of Bleeker et al. [2004], with minor modifications. The SEM study described here was conducted in collaboration with Dr. Hans M. Smid and Dr. Louise Vet from Wageningen University in the Netherlands. Briefly, female Ampulex compressa wasps were collected soon after emergence from their cocoons to avoid debris accumulation on the stinger. Wasps were anesthetized and their stingers dissected out and immersed overnight in CCl4 (Aldrich, Milwaukee, WI) at room temperature. Preparations were then transferred to 100% ethanol, rinsed once, critical point-dried and sputtered with 10 nm platinum for observation with a JEOL JSM 6300F field emission Scanning Electron Microscope.

I also used Transmission Electron Microscopy (TEM) to identify the ultrastructure of cuticular sensilla initially revealed by the SEM study. However, the TEM study failed to elicit satisfying results for two main reasons: First, the tip of the A. compressa stinger is extremely small in diameter, which prevents solutions from travelling to the stinger’s tip where sensilla of interest are distributed, resulting in incomplete fixation. Second, producing thin sections (70-80 nm) of the stinger cuticle proved particularly difficult to accomplish, largely due to the mechanical properties and physical instability of the cuticle which breaks easily under the diamond knife used for ultra- thin sectioning. Producing thin sections proved especially problematic for transverse sections, mainly due to the curvature of the stinger (see the micrograph of the stinger in Figure 2B). Indeed, to produce serial sections, the block had to be constantly tilted so that the stinger is cut repeatedly at a similar angle.

The TEM study took place during a two months period in Wageningen University, The Netherlands. The preparation protocol used was initially designed by Dr. Hans M. Smid to study the antennal sensilla of Cotesia (which are considerably larger and stiffer than the A. compressa stinger). The duration of the research period proved too short to provide more than one generation of newly-born wasps, insufficient to adjust the protocol to suit the A. compressa stinger. Nonetheless, for the purpose of future studies, I describe below the protocol I used for ultrastructural study of A. compressa stinger sensilla.

~ 25 ~

Whole stingers were dissected out from newly-born wasps and cut with a sharp razor blade to leave approximately 1 mm from the distal end, on which sensilla of interest are located. This generated a relatively large opening at the proximal end of the cut stinger and a short traveling distance for solutions to penetrate through and reach the stinger tip, yet still allowing relatively convenient handling of the tissue throughout the protocol. The cut stingers were placed on the bottom of a small Petri dish filled with 0.1 M phosphate buffer. To prevent the tissue from floating, I used double-sided adhesive tape glued to the bottom of the dish, taking care not to damage the area of interest or block the proximal end of the cut stingers. The buffer was then replaced with fixative (2.5% glutaraldehyde in 0.1 M phosphate buffer, adjusted to pH 7.0 with NaOH) and the tissue left for fixation for 3 hours at 20ºC. Then, the stingers were washed in 0.1 M phosphate buffer (3x5min) and immersed in 1% osmium tetraoxide in 0.1 M phosphate buffer for 2 hours at 20ºC. Next, stained stingers were thoroughly rinsed (5 x 5min) in 0.1 M buffer, dehydrated with a graded series of ethanol, transferred to vials and rinsed twice with propylene oxide. For embedding, I used Agar 100 resin (Agar Scientific, Essex, UK) mixed to ‘medium’ hardness based on the manufacturer’s instructions. The propylene oxide in the vials was replaced with 1:1 propylene oxide : resin solution, mixing the solution well with a pipette. The partially embedded stingers were left overnight with the vial open for the propylene oxide to partially evaporate. The propylene oxide : resin solution was then replaced with 100% resin, left at 20ºC for 4 hours and transferred to resin-filled sectioning capsules. After 24 hours at 60ºC, the capsules were ready for thin sectioning. Serial sections of 70-80 nm were sliced with a diamond knife, collected on formvar-coated nickel grids, contrasted with uranyl acetate and then observed under a Transmission Electron Microscope.

Electrophysiology of the stinger: To examine the physiological properties of sensory organs on the tip of the wasp’s stinger, I recorded their neuronal responses to relevant stimuli. The project was not completed due to an unexpected decline in the population of female wasps in our colony. The basic recording setup for all experiments was similar and is illustrated in Figure 5. Newborn wasps were cold-anesthetized and their abdomen separated from the thorax. The abdomen was pinned dorsal side up onto a Sylgard-covered Petri dish and the ventral nervous system was exposed after removing surrounding tissue, including efferent nerves and musculature. Throughout

~ 26 ~

the experiments, I found it important to perfuse fresh and aerated bee saline through the nervous system (based on [Galizia et al. 1997]; composition in mM: NaCl 130,

KCl 6, CaCl2 5, MgCl2 4, Sucrose 160, Glucose 25, HEPES buffer 10, pH 6.9). This was achieved by a gravity-based perfusion system as described in Figure 5A. A staple-shaped insect pin was placed beneath the stinger to elevate its distal part above the Sylgard, with only the base of the stinger and the abdominal nerve cord submerged in saline. A ground electrode was placed near the base of the stinger, close to the terminal abdominal ganglion (TAG). Data was acquired and digitized with Micro1401 MKII analog to digital board and analyzed offline with Spike2 data acquisition software (CED, Cambridge, UK).

I first tested whether stinger sensilla possess mechano-sensitive properties which might be used to discriminate nervous from non-nervous tissue inside the cockroach head cavity. I used two different extracellular setups to record neuronal activity in response to mechanical forces applied to the stinger: (1) silver hook electrodes placed beneath the afferent sensory neurons ascending from the stinger to innervate the TAG; (2) a custom-built suction electrode to record responses in the VNC itself, rostral to the TAG (see Figure 5). Ideally, both recording methods should be used simultaneously to study how mechanical forced are represented by neuronal activity in afferent nerves and in the VNC. To apply mechanical forces and simulate the natural stinging process, I used agar prepared to a concentration of 2%. I chose this concentration since it roughly approximates the density of the cockroach brain and is, therefore, often used to embed cockroach brains for histological microtome sectioning. A piece of the agar was inserted into the narrow distal end of a pipette tip, which then was mounted on a micro-manipulator. To apply a mechanical stimulus (denoted as arrows in Figure 9), the pipette tip was lowered so that the tip of the stinger (approximately 50 µm) penetrated the agar. Thus, the mechanical forces applied to the stinger were in parallel to the longitudinal axis of the stinger, mimicking the mechanical forces applied during a natural sting (see schematic drawing in Fig. 2B).

~ 27 ~

To test sensory responses to chemical cues, I used the tip-recording method [Hodgson et al. 1956, van Lenteren et al. 2007]. First, I prepared cockroach tissue homogenates from nervous and non-nervous tissue removed from the cockroach head cavity. Twenty cockroaches were anesthetized with carbon dioxide, their head capsules opened under cold saline and their cerebral nervous tissue (i.e., brain and SEG), as well as trachea and mouthparts muscles (which fill up most of the head cavity), were separated from the head capsule. The brains and SEGs were pooled together in one vial and the non-nervous tissue in another, with both vials containing a mixture of cockroach saline and protease inhibitors, to reach a final concentration (weight tissue/volume) of 0.3 mg tissue/ml solution. Both vials were then thoroughly sonicated to breakdown the tissue and the resulting homogenates were divided into 100 µL aliquots, stored at -20ºC. Before each experiment, one aliquot of each type (i.e., nervous and non-nervous tissue) was thawed in room temperature, homogenated in a vortex, and then pulled into a glass electrode filled with 0.1 M KCl (for controls, KCl alone was used), using negative pressure. This electrode, mounted onto an appropriate electrode holder, served as both the recording and stimulating electrode. To stimulate the stinger sensilla, the electrode was lowered so that the tip of the electrode established contact with the tip of the stinger to close the electrical circuit. Care was taken not to touch the stinger with the glass electrode shaft to avoid exerting mechanical force on the stinger.

~ 28 ~

Polyethylene tubing

Polyethylene Aerated tubing saline Stinger

Drain B Pipette tip

Hook Agar electrodes Stinger Suction electrode

Saline level TAG VNC

Figure 5: Wasp recording set-up. A. The wasp stinger and abdominal nerve cord are exposed on a Petri dish continually perfused with aerated saline. The rate of perfusion is adjusted so to allow only a shallow level of saline on the dish (dashed line), just high enough to cover the nerve cord. This allows for maintaining the stinger dry throughout the experiment by raising it above saline level with a staple-shaped pin placed beneath it (not shown). B. Detail of the recording set-up. The stinger (shown here as a Scanning Electron Micrograph) is held above the saline level in the dish and kept dry throughout the experiment. A pipette tip with a small piece of Agar is used to mechanically stimulate the stinger. Three different types of recording setups are possible: (1) Hook electrodes are used to record activity in afferent sensory neurons; (2) a suction electrode is used to record activity in the ventral nerve cord (VNC), and (3) a tip-recording set-up (not shown), in which the pipette tip is replaced with a recording electrode filled with KCl and nervous or non-nervous tissue homogenates extracted from cockroaches. TAG: Terminal Abdominal Ganglion. See text for details.

~ 29 ~

3.3. Results and Discussion

3.3.1. Behavioral experiments I first confirmed that A. compressa uses sensory feedback from its stinger during the head-sting (Figure 6, and see also [Gal et al. 2005, Gal and Libersat 2010]). The duration of the head sting increased significantly when wasps stung cockroaches with the brain or SEG removed (15-fold and 5-fold, respectively; p<0.001, t-test), as compared with sham-operated or neck-connectives-cut controls. In contrast, the duration of the thoracic sting was similar in all groups of cockroaches, demonstrating that the behavioral changes were specific to the removal of the cerebral ganglia. Thus, the wasp appears to actively search for the cerebral ganglia inside the cockroach head cavity during the head sting.

I also confirmed these searching movements visually. In one set of preliminary experiments (n=2), I removed the cockroach brain prior to the sting and left an open ‘window’ in the dorsal head cuticle (see Figure 4). This allowed me to observe the open head cavity during the sting. I then introduced this cockroach to a wasp and, as the wasp inflicted the head sting, used a stereoscope to observe the stinging process through the flap in the dorsal head cuticle. I first determined visually that the stinger reaches the top of the head capsule (where the brain is normally located) directly from below, i.e., from the direction of the SEG and the esophagus, which the stinger probably penetrates on its trajectory to the brain (as depicted in Figure 2B). I could also see the stinger moving about inside the head capsule, presumably in search of the surgically removed brain. Based on these observations I conclude that the stinger’s trajectory inside the head cavity is not pre-determined, but rather that the wasp uses sensory feedback as suggested above. The stinger moved repeatedly back and forth inside the dorsal head cavity, primarily in ‘stabbing-like’ movements at the dorso- ventral plane but also laterally demonstrating an exceptional steering mechanism. These movements persisted for the entire duration of the sting (approximately 10 minutes), during which time the stinger was occasionally retracted ventrally and then protruded again to continue the ‘stabbing-like’ motions. In one experiment, the wasp injected venom into the empty head cavity at the end of the stinging period, and then completely retracted its stinger. In the other experiment no venom injection was observed at the end of the stinging period.

~ 30 ~

***

Figure 6: Stinging duration after cockroach CNS lesions. Cutting the cockroach’s neck connectives prior to a sting, as well as surgically removing the brain or SEG from the head capsule, does not change the duration of the thoracic sting. In contrast, removing the SEG or the brain, but not cutting the neck connective alone, significantly prolongs the duration of the head sting. Schematics under the graph demonstrate the lesion performed prior to the sting. ***p<0.001; n=8 for each group.

~ 31 ~

3.3.2. Electron Microscopy The stinger of Ampulex compressa is composed of two ventral valves and two fused dorsal valves (Figure 7A). SEM analysis revealed cuticular structures on the stinger which resemble known sensory organs found on the ovipositor of other Hymenopterans. These structures are distributed mainly along the distal two-thirds of the stinger, increasing in density towards the distal end. At the stinger’s tip, the sensilla-like structures are distributed regularly on the ventral valves and form two characteristic sets of triplets on the fused dorsal valve (Figure 7A). Aside from the difference in the distribution pattern, the sensilla on all valves appear to be of a similar structure and consist of a uniporous peg located inside a depressed cuticular groove (Figure 7B). This structure highly resembles contact-chemoreceptive (‘gustatory’) sensilla found on other sensory appendages of other Hymenopterans [Quicke et al. 1999, LeRalec and Wajnberg 1990, Brown and Anderson 1998, van Lenteren et al. 2007].

Based on the results of SEM analysis, I also conducted a preliminary TEM study. However, due to reasons discussed above, this study was not completed and the ultrastructure of the sensilla-like structures was not entirely obtained. In most TEM micrographs, the protruding peg-like structure of sensilla was broken in the preparation, probably due to incomplete fixation or flawed thin-sectioning. However, the preliminary TEM study did yield some useful data, presented in Figure 8. Transverse sections through the stinger revealed that each of the valves encloses a large lumen through which, presumably, axons of sensory neurons run to reach the wasp’s abdominal VNC (Figure 8A, and compare with [van Lenteren et al. 2007]). This lumen branches off laterally in several locations along the stinger to reach the surface where the sensilla-like structures lie (Figure 8A-C). Indeed, presumably neuronal structures were often observed reaching cuticular sensilla (e.g., Figure 8E, F). Thus, it appears that at least some cuticular structures on the surface of the stinger, and especially those at the distal end, are indeed sensory organs. Although the uniporous depressed peg-like structure indicates a contact-chemoreceptive function, a thorough ultrastructural study needs to be completed to positively identify the nature of these sensilla.

~ 32 ~

A B G DV

VV VV

C D

Figure 7: SEM analysis of the wasp’s stinger. A: The tip of the stinger, showing the two ventral valves (VV) and the fused dorsal valve (DV). Arrows point to one of two characteristic triplets of sensilla-like cuticular structures on the dorsal valve. B: Two sensilla-like cuticular structures, each containing a peg-like protrusion (P) within a depressed groove (G). Notice the pore at the tip of the peg, suggesting a gustatory function. C, D: Close-up views of sensilla-like structure from the tip of the dorsal valve.

~ 33 ~

A B C

* Ct L * Ct * *

D C E F

Figure 8: TEM analysis of the wasp’s stinger (preliminary results). A-C: Longitudinal sections of one ventral valve; D-E: Cross sections of cuticular sensilla- like structures on the ventral valve; F: Cross section of cuticular sensillum from the dorsal valve. A. The cuticle (Ct) engulfs a large lumen (L) which branches off (arrow) and runs towards the periphery of the valve, presumably to reach cuticular sensilla. B: Micrograph taken from a more distal position on the valve than that presented in A. C. Detail of B, showing neurons (*) which presumably innervate cuticular sensilla. D. A cuticular sensillum comprised of a peg-like structure (tip broken in preparation) (arrowhead) located within a depressed groove (arrows). Inset: SEM of a cuticular sensillum for comparison. E. The depressed groove of a sensillum (arrow), presumably innervated by neurites (white arrows). F. A cuticular sensillum (tip broken in preparation) from the dorsal valve. Neurites can be seen reaching the peg- like protrusion. Scale bars: A, B: 5 µm; C: 1 µm, D-E: 0.5 µm, F: 1 µm.

~ 34 ~

3.3.3. Electrophysiology Electrophysiological recordings demonstrate that mechanical, as well as chemical cues evoke neuronal activity in stinger afferents and in the VNC. First, spiking activity was recorded in the afferent nerves ascending from the stinger to the VNC, as well as from the VNC itself, in response to mechanical forces applied to the tip of the stinger (Figure 9). This was the result of mechanical forces (rather than chemical cues) applied to the stinger, as periods in which the agar was static (although in contact with the stinger) did not evoke spikes (Figure 9A). Second, the tip of the stinger appears to detect chemical cues as well. This was revealed by ‘tip recordings’ in which a KCl-filled electrode was loaded with nervous or non-nervous tissue homogenates obtained from the cockroach head capsule and established contact with the stinger to close an electrical circuit (care was taken not to exert mechanical forced on the stinger when applying chemical stimuli). The tip recordings revealed strong chemically-evoked activity when the stinger was stimulated with homogenates obtained from tissue in the cockroach head but not by establishing contact with KCl alone (Figure 10). The firing rate immediately after establishing contact with the homogenate was higher for nervous tissue homogenates, as compared with non- nervous tissue homogenates (Figure 10), suggesting that the sensilla are capable of discriminating nervous from non-nervous tissue. Unfortunately, by the time of writing this thesis, not enough experiments had been preformed to obtain a statistically reliable analysis and further study should be conducted to thoroughly characterize sensory responses to both chemical and mechanical cues.

~ 35 ~

Figure 9: Stinger responses to mechanical stimulation. Electrophysiological recordings in two different preparations (see Figure 5): A. Recording from stinger afferents innervating the wasp’s terminal abdominal ganglion. B. Recording from the wasp’s ventral nerve cord. In both cases, a pipette tip filled with soft agar is moved against the tip of the stinger to exert mechanical force (direction of movements indicated by arrow directions). The dashed arrows indicate the onset and offset of contact established between the stinger and agar. Each movement against the stinger evoked spike-like potentials in the stinger afferents, as well as in the ventral nerve cord. Static periods (between blue arrows in A) do not evoke spikes, even though the agar is in contact with the stinger. Scale bars: 1 sec.

~ 36 ~

Figure 10: Stinger responses to chemical stimulation. Electrophysiological tip recordings made from a wasp’s stinger. The stimulating/recording electrode is filled with one of the following: (A) brain homogenate + KCl; (B) muscle homogenate + KCl; or (C) KCl alone. Dashed arrows represent stimulus onset. Scale bar is 10 sec and applies to all traces.

~ 37 ~

4. General discussion

4.1. Parasite-induced manipulation of host behavior in insects

Parasites have evolved remarkable strategies to manipulate the behavior of their hosts in order to perpetuate their own genes. For instance, parasites have been shown to manipulate phototaxis, locomotion, behavioral fevers, foraging behavior, reproduction and a variety of social interactions, to name but a few [Zimmer 2001, Moore 2002, Thomas et al. 2005]. However, although parasite-induced behavioral alterations are a widespread phenomenon, the underlying mechanisms are only beginning to be deciphered. In this section, I will survey host behavioral manipulations resulting from parasites which directly 'hijack' the host’s nervous system. Some indirect aspects of host behavioral manipulations, such as those resulting from alteration of the immune, endocrine or metabolic systems in the host, are thoroughly discussed elsewhere [Beckage and Gelman 2004, Moore 2002] and will not be addressed here.

To optimize their chances of reproduction, some parasitic fungi and worms manipulate the behavior of their insect hosts in a way that ultimately leads to suicidal behavior. For example, the parasitic fungus of the genus Cordyceps produces still unidentified chemicals that alter the navigational sense of its ant hosts to aid in the parasite’s dispersal [Mains 1948, 1949]. These chemicals cause an infected ant to climb to the top of a tree or plant and clamp its mandibles around a leafstalk to stay in place. The fungus feeds on the ant’s internal organs, and then its fruiting bodies sprout out of the ant’s cuticle and release capsules filled with spores (Figure 11A). The airborne capsules explode on their descent from the plant or tree, spreading the spores over the surrounding area to infect other ants and thus start another cycle. A somewhat similar strategy is employed by the Lancet liver fluke (Dicrocoelium dendriticum), another parasite that takes over the navigational skills of ants [Hohorst and Graefe 1961, Moore 1995]. The first intermediate host of this parasite is a terrestrial snail, which feeds on animal droppings containing the fluke’s eggs. The snail then secretes a ball of slime loaded with hundreds of lancet flukes at their juvenile stage (‘cercaria’). Together with the slime ball, the cercaria are then swallowed by an ant, which is the fluke’s second intermediate host. One cercarium then migrates and penetrates the sub-esophageal ganglion or the brain to manipulate ~ 38 ~ the ant’s behavior and increase the chance of the parasite to reach its definitive host [Lucius et al. 1980, Maeyama et al. 1994, Romig et al. 1980]. The infested ant leaves the colony and moves upward to the top of a blade of grass, to which it strongly clamps its mandibles to stay in place, often pointing its abdomen upwards. The ant waits there to be devoured by a grazer passing by, in which the fluke can reproduce. Once this happens, the fluke lays its eggs in the liver of the definitive host, from which the eggs then are flushed by the bile to be exerted through the grazer’s droppings. At this stage they may be again swallowed by a terrestrial snail, thus continuing the life cycle. Interestingly, if the ant was spared during the night, it returns to the ground at daybreak and behaves normally until evening returns. Then, the fluke takes over once again and sends the ant back up the grass for another attempt [Schneider and Hohorst 1971]. The chemicals which the fluke releases to manipulate the behavior of the ant, as in the previous example, have not yet been characterized.

Some nematode worms can also induce suicidal behavior in their hosts. Infection by a nematode of the genus Mermis, for instance, causes ants to seek water, jump in and eventually drown, thereby releasing the mature parasite, which is aquatic [Maeyama et al. 1994]. Similar abnormal suicidal behavior was observed in grasshoppers and crickets infected by Gordian worms (Figure 11B), as infected crickets seek out water and obligingly commit suicide by drowning. Researchers assume that the larvae produce certain chemicals that directly affect the host’s cerebral ganglia, probably disrupting the geotactic sense of the terrestrial host [Biron et al. 2005]. An even more complex behavioral manipulation is that seen in ant colonies which house the caterpillar of the butterfly Maculinea rebeli. This caterpillar is a parasite of ants that lives inside the brood chambers of ant nests, secreting chemicals which manipulate the ants to accept, nurture and protect the caterpillar as if it was their own. The parasitic caterpillar itself, however, also serves as a host for the parasitoid wasp Ichneumon eumerus. To safely reach the caterpillar, which is protected deep within the ant colony, the wasp releases agonistic chemicals to induce fighting between worker ants, locking the colony into combat and thus leaving its caterpillar host defenseless and available for oviposition [Thomas et al. 2002].

~ 39 ~

Probably one of the most exquisite alterations of host behavior is induced in spiders stung by another Ichneumonid wasp, Hymenoepimecis argyraphaga [Eberhard 2000, 2001]. Similar to the caterpillar of M. rebeli, this parasitoid wasp takes advantage of the natural behavior of the host to provide a safe shelter for its larva. In this example, however, the wasp manipulates the host spider to change its normal web-weaving pattern and construct a tailor-made shelter for the future wasp larva. The adult wasp stings its host, the spider Plesiometa argyra (Araneidae), on the spider's web. The sting transiently paralyses the spider and allows the wasp to lay an egg on its abdomen and fly away. The stung host soon recovers from paralysis and resumes apparently normal activity, building normal orb-webs to catch prey, while the parasitic larva grows by feeding on its hemolymph for about two weeks. Then, motivated by a thus far unknown cause, the spider starts weaving a very unique web, repeatedly executing only one sub-routine of the full orb-web construction program while repressing all other routines. The new web is strikingly different from the normal orb-shaped web of P. argyra and is designed to support the wasp’s cocoon so that it is suspended in the air, rather than glued onto a substrate. When this ‘safety net’ is ready, the larva consumes the spider, ultimately killing it, and then pupates on the newly formed web protected from most terrestrial predators. The changes in the spider's weaving behavior must be induced chemically rather than by direct physical interference from the wasp larva, as if the larva is removed immediately prior to the execution of the death sentence, the spider continues to build the specialized cocoon web. The nature of the chemicals involved in this extreme alteration of the spider's behavior remains, however, to be explored.

~ 40 ~

A B

Figure 11: Examples of fatal interactions between parasites and their insect hosts. A. The Camponotus ant, mandibles locked onto a leafstalk, with Hirsutella, the anamorph of Cordyceps unilateralis, emerging from the cuticle (courtesy and copyright of L. Gilbert). B. The hairworm Tellinii spinochordodes emerging from a host cricket, Nemobius sylvestris, after inducing suicidal behavior in the host (courtesy and copyright of F. Thomas). C. An adult wasp Ampulex compressa emerging from the abdomen of its host, the cockroach Periplaneta americana. Figure modified from Libersat et al. [2009].

~ 41 ~

4.2. Hunting strategies of parasitoid wasps

Being parasitic for only a part of their lives and free-living on the others, it is not surprising that parasitoid wasps have evolved strategies to handle their hosts in order to increase the chances of reproduction of their future larvae. However, as briefly mentioned above, most ectoparasitoid wasps do not use strategies as complicated as that of I. eumerus or H. argyraphaga to manipulate their host behavior. Instead, adult parasitoid wasps often incapacitate their host and immediately drag it to a burrow or nest nearby, where the host is relatively concealed from other potential predators. The wasp then lays an egg on the host and seals the burrow, leaving the inert prey inside for the developing larva to feed on. The hunting and host-manipulation strategies of parasitoid wasps are diverse and, at least to some extent, depend on the natural behavior of the host as well as on its size and ability to damage the wasp. The venom of the Beewolf (the Egyptian digger wasp, Philanthus triangulum), for example, contains potent neurotoxins (known as philanthotoxins) which evoke permanent neuromuscular paralysis in the bee prey [Piek et al. 1971]. In Pompilid wasps (such as the well-studied tarantula-hawk, Pepsis) which hunt large spiders, the adult wasp usually disarms the spider of its most formidable weapon, the fangs, with multiple deeply paralyzing stings into the cephalothorax and sometimes directly into the mouth [Steiner 1986, O’Neill 2001]. Sphecoid wasps typically hunt large and potentially harmful prey, usually orthopteroids (e.g. crickets, katydids, grasshoppers, etc), which they sting to evoke a total though usually transient paralysis, sometimes followed by a more specific long-term behavioral manipulation [Steiner 1986]. The transient paralysis allows resistance-free oviposition on the host, and in some instances the returning of the host to normal activity is advantageous for the wasp. For example, similar to the way H. argyraphaga employs the weaving skills of its spider host, larrine wasps (L. anathema, for example) use the tendency of their mole-cricket prey to burrow in the ground [Castner 1988]. When the wasp penetrates the underground refuge of the cricket and attacks it, the cricket may emerge in panic from its burrow, pursued by the wasp. The wasp then paralyses the cricket with multiple stings, lays a single egg on the host and leaves. The cricket soon recovers from paralysis and burrows back into the ground, resuming normal activity. When the larva hatches, it feeds on its host until pupation, protected in the cricket’s burrow from potential predators. In a different species of a Sphecoid parasitoid, the wasp Liris niger preys

~ 42 ~ on non-fossorial crickets (usually Acheta domesticus). The wasp stings the cricket close to the base of the hind legs or in the metathorax, and then three more times into the prothorax, the mesothorax and finally, into the neck. The stings induce total transient paralysis of the legs, with the stung cricket becoming inert and unable to maintain posture for several minutes [Gnatzy 2001, Steiner 1976, 1986]. The wasp then drags the paralyzed cricket to a burrow, glues an egg between its fore and the middle legs, and seals the burrow with soil particles or pebbles. After the burrow has been sealed, the cricket's legs fully recover from paralysis and the cricket can maintain posture and even walk. Nevertheless, the stung cricket never attempts to escape the burrow but rather remains motionless, although not paralyzed, in its tomb. The wasp larva, after hatching from the egg, feeds on the lethargic cricket and then pupates inside it. Thus, the Liris venom induces not only total transient paralysis in the legs to allow oviposition and transportation to the burrow, but also a partial and irreversible paralysis that renders the cricket prey submissive in its tomb. It has been suggested that the latter effect of the Liris venom is a result of the neck-sting, which is, in comparison, not typical for the mole cricket-hunting Larra that does not evoke such long-term effects.

The currently best understood direct manipulation of host behavior occurs between the Sphecoid Ampulicine Jewel Wasp, Ampulex compressa, and its cockroach prey [Hohorst and Graefe 1961, Williams 1942] (Figure 11C), which was the subject of my thesis. Similar to L. niger, this Sphecoid wasp employs a unique strategy to subdue its prey and supply its offspring with a live, yet immobile food reservoir (Figure 1). With a neurotoxic venom cocktail composed of proteins, peptides and sub-peptidic components [Haspel and Libersat 2003, Wiesel-Eichler et al. 1999], the wasp manipulates the behavior of cockroaches in a most interesting and complex manner, which has caught the eye of naturalists centuries ago [de Reaumur 1742, Bingham 1897, Maxwell-Lefroy 1909, Williams 1942]. Upon encountering a cockroach, an adult female Ampulex wasp stings the cockroach in the thorax to transiently immobilize the prey’s front legs. This short-term paralysis is mediated through chloride-channel agonists in the venom which interfere with synaptic transmission in the prothoracic ganglion [Moore et al. 2006] and enables a second sting, this time directed at the cockroach’s head. As a result of the second sting, two consecutive behavioral effects take place. First, presumably due to the effect of dopamine (DA)

~ 43 ~

found in the venom, the cockroach grooms itself excessively for 20-30 minutes, non- stop [Wiesel-Eichler et al. 1999]. Then, as the grooming phase of envenomation decays, the cockroach enters a long-lasting hypokinetic state characterized by little spontaneous or provoked locomotion, an important hallmark of which is the inability to self-initiate locomotion and produce normal escape responses [Fouad et al. 1994]. However, since the sting is not paralytic, the wasp can steer its intoxicated prey, pulling its antenna with the mandibles and walking backwards, into the nest. The cockroach does not try to fight the wasp but rather follows submissively, often through highly uneven terrain or up a tree, and then waits passively in the nest while the wasp seals the opening with debris and leaves. Hypokinesia persists until the cockroach meets its fate and the larva consumes it entirely and pupates inside its abdomen.

The hypokinetic state is a direct result of the venom injected directly and precisely into the two cerebral ganglia within the cockroach head cavity, namely the supra- esophageal ganglion (SupEG, or ‘brain’) and the sub-esophageal ganglion (SEG), both considered ‘higher-order’ neuronal centers which regulate insect locomotion [Fouad et al. 1994, Haspel et al. 2003, Kien and Altman 1992a,b] (Figure 2). Since the insects’ motor centers reside in the thorax, the venom must manipulate descending regulatory inputs from the cerebral ganglia to modulate thoracic motor centers. My main goal was to identify these circuitries and characterize their neuromodulatory effects on different motor patterns in stung and non-stung cockroaches. Below, I will discuss the results of my published work and its significance to the study of venom- induced hypokinesia, in particular, and to insect neuroscience, in general.

~ 44 ~

4.3. Overview and general discussion of the published work

4.3.1. The descending influence of cerebral ganglia on cockroach motor behaviors As a first step in my research, I investigated how the brain and SEG individually affect the selection, initiation, maintenance and expression of different motor behaviors in non-stung cockroaches [Gal and Libersat 2006]. To my knowledge, this work was the first to directly characterize the descending influence of each cerebral ganglion separately on a wide repertoire of cockroach motor behaviors, including walking, escape, righting, flying and swimming. Previous studies on the role of cerebral ganglia in regulating locomotion in invertebrates (and especially in insects) mainly employed a genetic approach, using the fruit fly as a model organism [Strauss et al. 1992, Strauss and Heisenberg 1993, Heisenberg 1994, Ilius et al. 1994, Martin et al. 1999, Strauss 2002], or a surgical approach, using semi-specific lesions in of the cerebral ganglia [Roeder 1963, Hughes 1965, Reingold and Camhi 1977, Graham 1979, Bassler 1983, Kien 1983, Kien and Altman 1984, Zill 1986, Homberg 1989, Kien 1990a,b, Bohm and Schildberger 1992, Keegan and Comer 1993, Strauss and Heisenberg 1993, Schaefer and Ritzmann 2001, Ridgel and Ritzmann 2005, Cornford et al. 2006]. Both methodologies, however, present significant caveats. In the genetic approach, motor deficits were examined after deleterious mutagenesis of specific structures in the brain only, mainly the mushroom bodies or the Central Body Complex (CBC). Although genetic investigations revealed much of the role of these structures in regulating different aspects of locomotion, the overall descending effect of the brain, as well as that of the SEG, were not addressed at all. Cerebral lesions complement these studies but since the insect nerve cord is neuroanatomically organized as a series of interconnected ganglia (in which the brain is most rostral), the role of the SEG in regulating locomotion is difficult to assess directly. Lesion experiments typically included (1) producing ‘brainless’ insects by cutting the circumesophageal connectives between the brain and SEG to remove all descending input from the brain, or (2) producing ‘headless’ insects by cutting the neck connectives caudal to the SEG to remove descending inputs from both cerebral ganglia. Hence, since the SEG is anatomically located between the brain and thorax, the role of the SEG in regulating locomotion could only be inferred by comparing brainless to headless insects.

~ 45 ~

As described in Gal and Libersat [2006], I developed a technique to remove descending inputs from the SEG with minimal damage to neurons descending from the brain, thus producing ‘SEG-less’ cockroaches. This was accomplished by a mid- sagittal slit through the SEG, as demonstrated in Figure 4. Since the vast majority of axons descending from the SEG have contralateral somata while almost all axons descending from the brain to the thorax pass through the periphery of the SEG without crossing over in this ganglion, splitting the SEG in half specifically removes SEG descending interneurons, while leaving brain descending interneurons intact. This was established both anatomically and functionally, as described in the publication.

I then examined motor deficits in brainless, SEG-less and headless cockroaches in a battery of behavioral paradigms. The behavioral analyses were complemented with electromyogram (EMG) recordings from the coxae of behaving cockroaches during the ongoing behavioral tasks, allowing me to investigate how the basic properties of motor behavior (e.g., initiation, maintenance, coordination, etc.) are regulated by the different cerebral ganglia. Based on the data acquired, I have built a model to describe how different stimuli evoke different motor behaviors, and how these are regulated by each cerebral ganglion. With respect to the behavioral manipulation of cockroaches by the Jewel Wasp, the most relevant finding was that the brain and SEG exert opposite descending tonic effects on locomotion. I found that while the general effect of the brain on walking-related behaviors is inhibitory (as was also shown by previous lesion experiments), the SEG sends permissive tonic input to thoracic pattern generators involved in the expression of walking-related behaviors. This finding confirmed a long-standing hypothesis in insect neuroscience.

~ 46 ~

4.3.2. The venom depresses the cockroach’s drive for walking After characterizing the influence of descending inputs on the expression of different motor pattern generators, I focused on studying which and how different motor patterns are expressed in stung cockroaches [Gal and Libersat 2008]. Previous studies have shown that the venom does not directly interfere with sensory or motor systems. Hence, in the context of the behavioral algorithm described in Diagram 1, the sting appears to affect CNS-mediated functions such as the decision to walk and escape, or the initiation and/or maintenance of these behaviors. To study which of these functions are manipulated in stung cockroaches, I used a combination of electromyographic (EMG) chronic recordings from leg muscles in freely moving animals with paradigms traditionally employed to study motivation in mammals. Some of these paradigms, to my knowledge, have never been used with cockroaches prior to this study and have yielded interesting and novel results, especially in the context of ‘behavioral despair’, as will be discussed below. The methodologies are described in detail in the publication.

I first established that the behavioral threshold for the initiation of walking-related behaviors, specifically, is increased in stung cockroaches but not ad infinitum. Previous studies have shown that rhythmic walking behavior cannot be elicited in stung cockroaches with conventional stimuli, such as wind or tactile stimuli. However, I have shown that walking-related behaviors (such as escape, walking or swimming) can be evoked by intense, though behaviorally relevant stimuli, such as electric foot shocks, repetitive tactile stimuli or water immersion. The ability to evoke rhythmic behavior in stung cockroaches also allowed me to examine the expression of locomotory Central Pattern Generators (CPGs) during ongoing rhythmic behaviors. These had never been examined previously, as rhythmic locomotory behavior is completely abolished in stung cockroaches standing on a solid surface. I focused this study on swimming behavior, since swimming in cockroaches involves the same CPGs as walking and, essentially, can be considered as a form of walking on water [Gal and Libersat 2006]). To study the basic properties of walking-related CPGs during ongoing swimming bouts, I used chronic EMG recordings from three coxae simultaneously. My results illustrate that the basic properties of the locomotory CPG (e.g. inter-leg coordination, correlation between cycle period, stance duration and discharge rate of Ds motor neurons, and more) is similar in control and stung

~ 47 ~

cockroaches. Hence, the pattern of locomotor activity in the thoracic segments of stung cockroaches, once initiated by a supra-threshold stimulus, appears to be normal. I also found, however, that for each cycle period recorded during ongoing swimming bouts, the discharge rate of the Ds motor neurons was lower and the stance duration longer, in stung as compared with control cockroaches. This suggests that neuromodulatory permissive inputs to the locomotory CPG, presumably descending from the cerebral ganglia in which the venom is injected, is decreased in stung cockroaches. In agreement with this notion, when stung cockroaches were introduced into a water-filled container (mimicking the Forced Swimming Test used to test motivation in mammals), they swam for significantly shorter durations than did controls, as if they ‘despaired’ faster. Anecdotally, the decrease in swimming duration, together with the normal swimming pattern seen during active swimming bouts, was strikingly similar to the behavioral symptoms observed in mammalian models for depression. Furthermore, the decrease in descending excitatory drive to motor centers in stung cockroaches appears to be highly selective to walking-related behaviors since other behaviors, such as flying and righting (which are also generated at the thorax), remained behaviorally and physiologically intact. This is again analogous to ‘depressed’ mammalian models, which show deficits in some motor behaviors but not in others.

Taken together, my data show that the venom does not induce a general ‘sleep-like’ state in stung cockroaches (as was previously hypothesized by Fouad et al. [1994]) but rather selectively depresses the cockroach motivation, or ‘drive’, to initiate and maintain walking-related behaviors. Since the venom is injected directly into discrete regions of the cockroach cerebral nervous system, one or more of these regions (i.e., the SEG and the mushroom bodies and CBC in the brain) appears to be substantially involved in determining the drive for walking in cockroaches, and probably in other insects as well. The next step in my research was thus to identify the neuronal substrate for the venom-induced hypokinesia. Since the SEG exerts permissive tonic descending input to thoracic locomotory circuits, the most probable target for the venom’s effect in modulating the drive for walking appears to be the SEG. This hypothesis was confirmed in the next publication.

~ 48 ~

4.3.3. Involvement of the SEG in venom-induced hypokinesia A long-standing hypothesis in the study of venom-induced hypokinesia [Fouad et al. 1994] was that the venom is directed at the cockroach SEG. This hypothesis was mainly anecdotal and based on the fact that during the head sting, the wasp inserts its stinger into the ventral part of the cockroach head capsule, in the direction of the SEG. In 2003, Haspel et al. also reported that venom is injected into the brain as well [Haspel et al. 2003]. This raised the question of whether venom injected into the different cerebral ganglia induces different behavioral manipulations in the stung cockroach (i.e., long-term hypokinesia [Fouad et al. 1994], changes in cockroach metabolism [Haspel et al. 2005] or excessive grooming behavior [Wiesel-Eichler et al. 1999]). My studies have shown that stung cockroaches behave similarly to SEG- less cockroaches in many of the behavioral paradigms tested [Gal and Libersat 2006, 2008]. I, therefore, hypothesized that the SEG is at least one neuronal substrate responsible for the venom-induced hypokinesia. To test this hypothesis, I relied on behavioral, electrophysiological and neuropharmacological methods.

As described in [Gal and Libersat 2010], I demonstrated that during the head sting, the wasp uses its stinger to search for the SEG inside the head capsule of its cockroach prey. Hence, the SEG appears to be at least one neuronal target for the wasp when afflicting the head sting, highlighting the potential role of this ganglion in venom-induced hypokinesia. I was also able to mimic the hypokinetic effects of the sting upon pharmacological inhibition of neuronal activity in the SEG using procaine, which produces stung-like behavioral deficits in non-stung cockroaches. Similar deficits were induced by injections of crude milked venom into the SEG of non-stung cockroaches. This was in marked contrast with injections of procaine or venom into the brain of non-stung cockroaches, which induced hyper-, rather than hypoactivity. These findings complement previous lesion studies [Gal and Libersat 2006] and illustrated the up-regulatory role of the SEG in modulating locomotor activity. Finally, I performed extracellular recordings from the SEG of stung and non-stung cockroaches in situ. I found that spontaneous and evoked spiking neuronal activity in the SEG is decreased in stung cockroaches, as compared with controls. Taken together, my results demonstrate that the wasp inhibits neuronal activity in, at least, the SEG to induce hypokinesia. This also further signifies the crucial role of this little- studied ganglion in modulating locomotion in insects.

~ 49 ~

4.4. How the Jewel Wasp ‘hijacks the free will’ of cockroaches: Current mechanistic hypothesis

My work has demonstrated that the SEG selectively sends tonic permissive inputs to the locomotory CPG in the thorax, and that the wasp’s venom suppresses neuronal activity in the SEG to decrease the propensity of expression of locomotory behaviors. Hence, stung cockroaches show an increased threshold for the initiation of locomotion, but not for the initiation of other, non-related behaviors. Similarly, although extremely intense stimuli can surpass the behavioral threshold in stung cockroaches and activate the locomotory CPG in a normal fashion, locomotion cannot be maintained for long periods of time. These deficits of stung cockroaches interestingly parallel deficits in mammals with decreased locomotory motivation, such as those used as models for depression [see Gal and Libersat 2008]. Anecdotally, stung cockroaches could thus be considered as lacking the motivation, or ‘drive’, to move. However, while insect attention and arousal states, and their correlation with mammalian equivalents, have been thoroughly investigated during the last few years [Greenspan et al. 2001, van Swinderen and Andretic 2003, Van Swinderen 2005, Cirelli 2009, Cirelli and Bushey 2008], the neuronal underpinnings of insect motivation have received relatively little attention despite their obvious implications on the regulation of behavior. By employing the naturally-occurring parasitic interaction between the Jewel Wasp and its cockroach prey, my work has highlighted the role of the SEG in the up-regulation of locomotor activity and determining the ‘rest state’ and locomotory motivation of cockroaches, and probably of other insects as well. Below, I will integrate my results with those obtained in previous studies to describe a comprehensive hypothesized model for the hypokinetic state induced in cockroaches by the Jewel Wasp’s venom. I will focus on the mechanisms by which the venom inhibits the well-defined escape circuitry of cockroaches, thereby rendering them unresponsive ‘zombies’. The hypothesized model is depicted in Figure 12.

~ 50 ~

The wasp injects its venom cocktail directly into cerebral regions involved in the modulation of locomotion, namely the SEG and two regions in the brain, the Central Body Complex (CBC) and the mushroom bodies [Haspel et al. 2003, Schaefer and Ritzmann 2001, Ridgel and Ritzmann 2005]. My research has demonstrated that the two cerebral ganglia send a tonic descending signal to the thoracic motor centers, providing further evidence of their role in regulating the motivational state of insects [Gal and Libersat 2006]. In stung cockroaches, wind stimuli applied to the cerci, tactile stimuli applied to the antennae or tactile stimuli applied to the anal plates, each recruiting different premotor pathways that usually produce strong escape responses, are no longer effective [Fouad et al. 1994, 1996, Libersat et al. 1999, Gal and Libersat 2008]. Under natural conditions, wind-sensitive hairs on the abdominal cerci detect minute air movements (as produced, for instance, by a predator’s strike) and excite giant interneurons (GIs) in the terminal abdominal ganglion to mediate escape running (Figure 3). A tactile stimulus applied to the abdomen recruits a different population of GIs that also mediate escape running. In both cases, the GIs activate various interneurons in the thoracic locomotor centers that, in turn, excite various local interneurons or motor neurons associated with escape running. In addition, escape running can be triggered by tactile stimuli applied to the antennae, which recruits specific interneurons descending from the head to the thorax. Thus, touch- or wind-sensitive input from rostral and/or caudal stimuli are carried by three distinct populations of interneurons (Figure 3 and Figure 12 show two of these). These eventually converge onto the same thoracic circuitry that evokes running leg movements in normal but not in stung cockroaches. Previous experiments revealed that the head-sting affects neither the response of the ascending GIs nor that of the descending interneurons [Libersat et al. 1999], and that thoracic interneurons receive comparable synaptic drive from the GIs in control and stung animals [Libersat et al. 1999]. Thus, the ultimate effect of venom injected into the cerebral ganglia must take place at the connection between thoracic interneurons and specific motor neurons that control leg movements. In non-stung cockroaches, escape running requires the recruitment of both fast and slow motor neurons in the thorax. Under natural conditions, when a stung cockroach is standing on the ground, tactile or wind stimuli do not recruit fast motor neurons [Fouad et al. 1996, Libersat et al. 1999]. However, the same motor neurons can be recruited experimentally during other behaviors [Gal and Libersat 2008]. When cockroaches are placed on their back, for instance, the fast

~ 51 ~

motor neurons are fully active and enable the rapid and strong leg movements required for righting [Gal and Libersat 2008]. The same motor neurons are also recruited when submerging a stung cockroach into water to induce swimming behavior, which also requires strong and fast leg movements [Gal and Libersat 2008]. Owing to their bi-functional nature, these motor neurons are also recruited during flying, for instance when a stung cockroach is placed inside a wind tunnel [Libersat 2003]. These observations suggest that the descending control from the cerebral ganglia on the thoracic walking circuitry is exerted on thoracic premotor circuits. In support of this, I have shown that in stung cockroaches, the discharge rate of slow- muscle potentials (which reflect activity in the slow motor neurons) during active swimming bouts is also decreased, compared with that of control cockroaches [Gal and Libersat 2008]. Thus, it seems that the specific effect of the venom on walking- related behaviors requires inhibition of thoracic premotor elements by descending inputs specifically involved in regulating such behaviors. As a consequence, this descending input must bear information regarding the motivation to initiate or maintain walking-related behaviors. However, how could the venom act so selectively, chemically manipulating only a particular subset of descending inputs to modulate specific behaviors?

One possible explanation is that the venom targets a neuromodulatory system that specifically controls selected behaviors. Certain neuromodulatory systems located in the thoracic and cerebral ganglia are involved in the initiation and/or execution of walking. The monoaminergic system is a likely candidate, as alterations in this system in both the thoracic and cerebral ganglia affect specific subsets of behavior [Libersat and Pflueger 2004]. A few important studies have revealed that monoamines might indeed be involved in the venom-induced hypokinesia. A systemic injection of reserpine (a toxin which depletes monoamines in the CNS) to non-stung cockroaches, for instance, induces a hypokinetic state similar to that seen in stung cockroaches [Wiesel-Eichler and Libersat, 2002]. This is presumably due to depletion of the monoamine dopamine (DA) in the CNS, as DA was found to be required for normal escape responses in non-stung cockroaches. An interesting observation in support of this relies on the involvement of DA in the venom-induced excessive grooming behavior, and its relation to the successive hypokinetic state. Similar to cockroaches stung in the head, cockroaches injected with a DA agonist show excessive grooming

~ 52 ~

behavior. However, when cockroaches are stung a second time while already in the hypokinetic state, or when a DA agonist is injected to stung hypokinetic cockroaches, no excessive grooming is observed [Wiesel-Eichler and Libersat, 2002]. Thus, it is likely that the grooming phase, which is presumably mediated through dopaminergic pathways [Wiesel-Eichler et al. 1999], is also mechanistically related to the hypokinetic state. It was later discovered that systemic injection of DA agonists to stung cockroaches cannot ‘rescue’ the cockroaches from their hypokinetic state [Rosenberg et al. 2007]. Hence, if the venom affects dopaminergic pathways in the cockroach CNS to induce hypokinesia, it must affect dopaminergic receptors or targets downstream. This, however, needs to be addressed directly.

The monoamine, octopamine (OA), was also found to play an important role in venom-induced hypokinesia. A recent study showed that in stung cockroaches, the activity of identified OA neurons in the cockroach’s thorax. i.e., the dorsal unpaired median (DUM) neurons, which modulate the excitability of specific thoracic premotor neurons, is compromised [Rosenberg et al. 2006]. The differences observed in the intracellular activity of thoracic DUM neurons between stung and non-stung cockroaches indicate that it is probably calcium currents that are modulated in DUM neurons of stung cockroaches [Rosenberg et al. 2006]. A change in the neuromodulatory environment of DUM neurons, for example due to the removal of descending neuromodulatory input from the cerebral ganglia, could account for such changes. However, the changes in the excitability of thoracic DUM neurons in stung cockroaches are not accompanied by changes in postsynaptic activity, indicating comparable strength of the presynaptic input in stung and non-stung cockroaches. This is in contrast with postsynaptic activity in the thoracic DUM neurons of brainless cockroaches, in which the rate and amplitude of postsynaptic potentials are markedly decreased as a result of brain removal. Thus, it appears that the direct synaptic connection between the brain and thoracic DUM neurons is unaffected by the sting. The results published in my latest article [Gal and Libersat 2010] suggest that the venom modulates neuromodulatory neurons in the SEG which innervate neuropiles in the thorax, brain, or both. Recently, OA-immunoreactive neurons in the SEG of Periplaneta have been identified and their projection patterns described [Sinakevitch et al. 2005]. The SEG contains 16 OA-immunoreactive midline neurons, arranged in three dorsal midline clusters, and two bilateral clusters of five OA-immunoreactive

~ 53 ~

somata, located between the roots of the tritocerebral connectives. At least three of these SEG-DUM neurons provide dense innervation in the protocerebral-bridge and ellipsoid body of the Central Body Complex, regions implicated in controlling locomotion [Strausfeld 1999, Strauss 2002]. Thus, venom-induced modulation of SEG descending and/or ascending neuromodulatory neurons could decrease the excitability of thoracic DUM neurons that, in turn, modulate thoracic premotor neurons, and thereby decrease locomotion. Some evidence suggests that brain OA neurons (which are thus likely to be modulated by ascending SEG-DUM neurons) are involved in the modulation of thoracic DUM neurons, since injection of an OA receptor agonist directly into the brain (but not into the SEG) partially ‘rescues’ stung cockroaches from their hypokinetic state [Rosenberg et al. 2007]. The precise mechanism and specific neuronal circuitry in the SEG, which seems to be the core of the venom-induced hypokinesia, has yet to be discovered.

~ 54 ~

Antennae

Brain H E

Venom Circumesophageal connectives

I SEG

Neck

connectives

G A Thoracic B Ganglia

Leg Muscles Giant Interneurons

F TAG

D Cerci

~ 55 ~

Figure 12: Current model of the neurophysiological events leading to venom- induced hypokinesia in cockroaches stung by Ampulex compressa. Schematic and simplified drawing of a cockroach nervous system, depicting circuitries that affect walking-related behaviors. The walking pattern generator that orchestrates leg movements is located in the thorax. It consists of motor neurons innervating leg muscles (A), ascending neurons from sensory structures on the legs (not shown) and type-A thoracic interneurons (TIAs; B), which synapse onto the motor neurons directly and indirectly via local interneurons (C). The TIAs receive inputs from several interneurons. For example, sensory neurons (D) in the antennae or cerci recruit descending (E) or ascending (F) Giant Interneurons (GIs) in the terminal abdominal ganglion (TAG). The GIs converge directly onto the TIAs to ultimately evoke escape responses. In addition, neurons of the pattern generator receive input from thoracic neuromodulatory cells (G). One example of these is the thoracic dorsal unpaired median (DUM) neurons, which secrete octopamine and modulate the efficacy of premotor-to-motor (B-to-A) synapses. The neuromodulatory cells, in turn, receive tonic regulatory input through interneurons descending from the brain (H) and sub-esophageal ganglion (SEG) (I). This tonic input affects the probability of occurrence of specific motor behaviors by modulating the different thoracic pattern generators directly (not shown) or indirectly (H, I). The wasp, A. compressa, injects its venom cocktail directly into both cerebral ganglia to modulate specific, yet unidentified cerebral circuitries. The current hypothesis states that in the SEG, the venom suppresses the activity of neuromodulatory neurons (I), presumably SEG- DUM neurons, which (i) ascend to the brain to regulate descending neuromodulatory (probably octopaminergic) neurons (H), and/or (ii) descend to the thorax to regulate locomotory CPGs directly (not shown) or indirectly through thoracic neuromodulatory (probably DUM) neurons (G). Hence, the venom injected into the cerebral ganglia decreases the overall excitatory input to the thoracic locomotory pattern generator. As a result, walking-related behaviors are specifically inhibited and stimuli to the antennae or cerci fail to evoke normal escape responses.

~ 56 ~

5. References

Adams ME, Swanson G (1996) Neurotoxins (second edition). Trends Neurosci Suppl 17: S1-37.

Arthur AP, Hegdekar BM, Rollins RL (1969) Component of the host haemolymph that induces oviposition in a parasitic insect. Nature 223: 996-997.

Bassler U (1983) Neural basis of elementary behavior in stick insects (studies of brain function). Springer, Berlin Heidelberg, New York.

Beckage NE, Gelman DB (2004) Wasp parasitoid disruption of host development: implications for new biologically based strategies for insect control. Annu Rev Entomol 49: 299–330.

Bingham CT (1897) The fauna of British India, Hymenoptera, I. pp. 253-254.

Biron DG, Marche L, Ponton F, Loxdale HD, Galeotti N, et al. (2005) Behavioural manipulation in a grasshopper harbouring hairworm: a proteomics approach. Proc R Soc London Sci Ser B 272: 2117–2126.

Bleeker MA, Smid HM, Van Aelst AC, Van Loon JJ, Vet LE (2004) Antennal sensilla of two parasitoid wasps: a comparative scanning electron microscopy study. Microsc Res Tech 63: 266-273.

Bohm H, Schildberger K (1992) Brain neurons involved in the control of walking in the cricket Gryllus bimaculatus. J Exp Biol 166: 113–130.

Brown PE, Anderson M (1998) Morphology and ultrastructure of sense organs on the ovipositor of Trybliographa rapae, a parasitoid of the cabbage root fly. J Insect Physiol 44: 1017-1025.

Camhi JM (1984) Neuroethology: nerve cells and natural behaviour of animals. Sinauer Associates Inc., Sunderland Mass. Chapter 4: A case study in neuroethology: the escape system of the cockroach. pp. 79–105.

~ 57 ~

Camhi JM, Levi A (1988) Organization of a complex movement: Fixed and variable components of the cockroach escape behavior. J Comp Physiol A 163: 317–328.

Casagrand JL, Ritzmann RE (1992) Biogenic amines modulate synaptic transmission between identified giant interneurons and thoracic interneurons in the escape system of the cockroach. J Neurobiol 23: 644-655.

Castner JL (1988) Biology of the mole cricket parasitoid Larra bicolour (Hymenoptera: Sphecidae). In: Advances in Parasitic Hymenoptera Research (Gupta VK, ed), pp. 131–139. Leiden: Brill.

Cornford A, Kristan WB 3rd, Malnove S, Kristan WB Jr, French KA (2006) Functions of the subesophageal ganglion in the medicinal leech revealed by ablation of neuromeres in embryos. J Exp Biol 209: 493-503.

Cirelli C (2009) The genetic and molecular regulation of sleep: from fruit flies to humans. Nat Rev Neurosci 10: 549-560.

Cirelli C, Bushey D (2008) Sleep and wakefulness in Drosophila melanogaster. Ann NY Acad Sci 1129: 323-329. de Reaumur RAF (1742) Memoirs pour server al’histoire des insects, VI. Académie Royale des Sciences, Paris, France. pp. 280-284.

Eberhard WG (2000) Spider manipulation by a wasp larva. Nature 406: 255–256.

Eberhard WG (2001) Under the influence: webs and building behavior of Plesiometa argyra (Aranea, tetragnathidae) when parasitized by Hymenoepimecis argyraphaga (Hymenoptera, Ichneumonidae). J Arachnol 29: 354–366.

Fouad K, Libersat F, Rathmayer W (1994) The venom of the cockroach-hunting wasp Ampulex compressa changes motor thresholds: a novel tool for studying the neural control of arousal? Zoology 98: 23–34.

Fouad K, Libersat F, Rathmayer W (1996) Neuromodulation of the escape behavior of the cockroach Periplaneta americana by the venom of the parasitic wasp Ampulex compressa. J Comp Physiol A 178: 91–100.

~ 58 ~

Gal R, Haspel G, Libersat F (2003) Wasp uses specialized sensors to probe and inject venom inside the brain of its cockroach prey. Neural Plasticity 10: 200 (abstract).

Gal R, Libersat F (2004) Wasp Uses "Brain Sensors" to Sting Cockroach. Abstracts of the 7th Congress of the International Society for Neuroethology, Nyborg, Denmark.

Gal R, Libersat F (2006) New vistas on the initiation and maintenance of insect motor behaviors revealed by specific lesions of the head ganglia. J Comp Physiol A 192: 1003-1020.

Gal R, Libersat F (2008) A parasitoid wasp manipulates the drive for walking of its cockroach prey. Curr Biol 18: 877–882.

Gal R, Libersat F (2010) A wasp manipulates neuronal activity in the sub-esophageal ganglion to decrease the drive for walking in its cockroach prey. PLoS ONE, in press.

Gal R, Rosenberg LA, Libersat F (2005) Parasitoid wasp uses a venom cocktail injected into the brain to manipulate the behavior and metabolism of its cockroach prey. Arch Insect Biochem Physiol 60: 198–208.

Galizia CG, Joerges J, Küttner A, Faber T, Menzel R (1997) A semi-in-vivo preparation for optical recording of the insect brain. J Neurosci Methods 76: 61-69.

Ganesalingam VK (1972) Anatomy and histology of the sense organs of the ovipositor of the Ichneumonid wasp, Devorgilla canescens. J Insect Physiol 18: 1857- 1867

Ganesalingam VK (1974) Mechanism of discrimination between parasitized and unparasitized hosts by Venturia canescens (Hymenoptera: Ichneumonidae). Ent Exp & Appl 17: 36-44.

Gnatzy W (2001) Digger wasp vs cricket: (neuro-) biology of a predator-prey- interaction. Zoology 103: 125–139.

Greany PD, Hawke SD, Carlysle TC, Anthony DW (1977) Sense organs on the ovipositor of Biosteres (Opius) longicaudatus, a parasite of the Caribbean fruit fly Anastrepha suspensa. Ann Entomol Soc Am 70: 319-321.

~ 59 ~

Graham D (1979) Effects of circum-oesophageal lesion on the behavior of the stick insect Carausius morosus. II. Changes in walking co-ordination. Biol Cybern 32: 147– 152.

Greenspan RJ, Tononi G, Cirelli C, Shaw PJ (2001) Sleep and the fruit fly. Trends Neurosci 24: 142-145.

Gutierrez AP (1970) Studies on host selection and host specificity of the aphid hyperparasite Charips victrix (Hymenoptera: Cynipidae). 6. Description of sensory structures and a synopsis of host selection and host specificity. Ann Entomol Soc Am 63: 1705-1709.

Haspel G (2003) Biochemistry and physiological action of Ampulex compressa venom on cockroach prey. PhD Thesis, Kreitman School for Graduate Studies, Ben- Gurion University of the Negev.

Haspel G, Gefen E, Ar A, Glusman JG, Libersat F (2005) Parasitoid wasp affects metabolism of cockroach host to favor food preservation for its offspring. J Comp Physiol A 191: 529–534.

Haspel G, Libersat F (2003) Wasp venom blocks central cholinergic synapses to induce transient paralysis in cockroach prey. J Neurobiol 54: 628-637.

Haspel G, Rosenberg LA, Libersat F (2003) Direct injection of venom by a predatory wasp into cockroach brain. J Neurobiol 56: 287–292.

Hawke SD, Farley RD, Greany PD (1973) The fine structure of sense organs in the ovipositor of the parasitic wasp, Orgilus lepidus Muesebeck. Tissue Cell 5: 171-184.

Heisenberg M (1994) Central brain functions in insects: genetic studies on the mushroom bodies and central complex in Drosophila. In: Neural Basis of Behavioural Adaptations (Schildberger K and Elsner N, eds), pp. 61-79. Stuttgart: Gustav Fischer Verlag.

Hermann HR, Douglas ME (1976) Comparative survey of the sensory structures on the sting and the ovipositor of hymenopterous insects. J Georgia Entomol Soc 11: 223-239.

~ 60 ~

Hodgson ES, Lettvin JY, Roeder KD (1956) Physiology of a primary receptor unit. Science 122, 417-418.

Hohorst W, Graefe G (1961) Ameisen—obligatorische Zwischenwirte des Lanzettegels (Dicrocoelium dendriticum). Naturwissenschaften 48: 229–230.

Homberg U (1989) Structure and functions of the central complex in insects. In: Arthropod Brain: Its Development, Structure and Functions (Gupta AP, ed), pp. 347- 367. New York: Wiley.

Hughes GM (1965) Neuronal pathways in the insect central nervous system. In: The Physiology of the Insect Central Nervous System (Treherne JE and Beaments JWL, eds), pp. 79–112. New York: Academic.

Ilius M, Wolf R, Heisenberg M (1994) The central complex of Drosophila melanogaster is involved in flight control: studies on mutants and mosaics of the gene ellipsoid body open. J Neurogenet 9: 189-206.

Kainoh Y, Brown JJ (1994) Amino acids as oviposition stimulants for the egg-larval parasitoid, Chelonus sp. Near curvimaculatus (Hymenoptera: Braconidae). Biol Control 4: 22-25.

Keasar T, Sheffer N, Glusman G, Libersat F (2006) Host handling: an innate component of foraging behavior in the parasitoid wasp Ampulex compressa. Ethology 112: 699–706.

Keegan AP, Comer CM (1993) The wind-elicited escape response of cockroaches (Periplaneta americana) is influenced by lesions rostral to the escape circuit. Brain Res 620: 310–316.

Kien J (1983) The initiation and maintenance of walking in the locust: an alternative to the command concept. Proc R Soc Lond B Biol 219: 137–174.

Kien J (1990a) Neuronal activity during spontaneous walking. I.Starting and stopping. Comp Biochem Physiol A 95: 607–621.

~ 61 ~

Kien J (1990b) Neuronal activity during spontaneous walking. II. Correlation with stepping. Comp Biochem Physiol A 95: 623–638.

Kien J, Altman JS (1984) Descending interneurones from the brain and suboesophageal ganglia and their role in control of locust behaviour. J Insect Physiol 30: 59–72.

Kien J, Altman JS (1992a) Decision-making in the insect nervous system: a model for selection and maintenance of motor programmes. In: Neurobiology of Motor Programme Selection (Kien J, McCrohan CR, Winlow W, eds), no. 4, pp. 147–169. Oxford: Pergamon.

Kien J, Altman JS (1992b). Preparation and execution of movement: parallels between insects and mammalian motor systems. Comp Biochem Physiol A 103: 15-24.

King PE, Rafai J (1970) Host discrimination in a gregarious parasitoid Nasonia vitripennis (Walker) (Hymenoptera: Pteromalidae). J Exp Biol 53: 245-254.

LeRalec A, Wajnberg E (1990) Sensory receptors of the ovipositor of Trichogramma maidis (Hym.: Trichogrammatidae). Entomophaga 35: 293-299.

Libersat F (2003) Wasp uses venom cocktail to manipulate the behavior of its cockroach prey. J Comp Physiol A 189: 497–508.

Libersat F, Delago A, Gal R (2009) Manipulation of host behavior by parasitic insects and insect parasites. Annual Reviews of Entomology 54: 189-207.

Libersat F, Gal R (2007) Neuro-manipulation of hosts by parasitoid wasps. In: Recent Advances in the Biochemistry, Toxicity, and Mode of Action of Parasitic Wasp Venoms (Rivers D and Yoder J, eds), pp. 93-114. Kerala: Research Signpost.

Libersat F, Haspel G, Casagrand J, Fouad K (1999) Localization of the site of effect of a wasp’s venom in the cockroach escape circuitry. J Comp Physiol A 184: 333– 345.

Libersat F, Pflueger HJ (2004) Monoamines and the orchestration of behavior. BioScience 54: 17–25.

~ 62 ~

Lucius R, Romig T, Frank W (1980) Camponotus compressiscapus Andr´e (Hymenoptera, Formicidae) an experimental second intermediate host of Dicrocoelium hospes Looss, 1907 (Trematodes, Dicrocoeliidae). Z. Parasitenkd. 63: 271–275.

Maeyama T, Terayama M, Matsumoto T (1994) The abnormal behavior of Colobopsis sp. (Hymenoptera, Formicidae) parasitized by Mermis (Nematoda) in Papua New Guinea. Sociobiology 24: 115–119.

Mains EB (1948) Entomogenous fungi. Mycologia 40: 402–416.

Mains EB (1949) Cordyceps bicephala Berk. and C. australis (Speg.) Sacc. Bull Torrey Bot Club 76: 24–30.

Martin JR, Raabe T, Heisenberg M (1999) Central complex substructures are required for the maintenance of locomotor activity in Drosophila melanogaster. J Comp Physiol A 185: 277-288.

Maxwell-Lefroy H (1909) Indian insect life. pp. 207-208.

Moore J (1995) The behavior of parasitized animals. Bioscience 45: 89–96.

Moore J, ed. (2002) Parasites and the behavior of animals. Oxford: Oxford Univ Press.

Moore EL, Haspel G, Libersat F, Adams ME (2006) Parasitoid wasp sting: a cocktail of GABA, taurine, and beta-alanine opens chloride channels for central synaptic block and transient paralysis of a cockroach host. J Neurobiol 66: 811-820.

Nettles WC, Morrison RK, Xie ZN, Ball D, Shenkir CA, Vinson SB (1982) Synergistic action of potassium chloride and magnesium sulfate on parasitoid wasp oviposition. Science 218: 164-166.

O’Neill KM (2001) Solitary wasps: Behavior and Natural History. Ithaca, New York: Cornell University Press. pp. 58-60.

Olivera BM (1999) Conus venom peptides: correlating chemistry and behavior. J Comp Physiol A 185: 353-359.

~ 63 ~

Piek T, Mantel P, Engels E (1971) Neuromuscular block in insects caused by the venom of the digger wasp Philanthus triangulum F. Comp Gen Pharmacol 2: 317– 331.

Piek T, Spanjer W (1986) Chemistry and pharmacology of solitary wasp venoms. In: Venoms of the Hymenoptera (Piek T, ed), pp. 161-308. London: Academic Press.

Piek T, Hue B, Lind A, Mantel P, Marle J van, Vissier JH (1989) The venom of Ampulex compressa - effects on behavior and synaptic transmission of cockroaches. Comp Biochem Physiol C 92: 175-183.

Piek T, Visser JH, Veenendaal RL (1984) Change in behaviour of the cockroach, Periplaneta americana, after being stung by the sphecid wasp Ampulex compressa. Entomol Exp App 195: 195-203.

Quicke, DLJ, Le Ralec A, Vilhelmsen L (1999) Ovipositor structure and function in the parasitic Hymenoptera with an exploration of new hypotheses. Atti dell’Accademia Nazionale Italiana di Entomologia Rendiconti 47: 197-239.

Rappuoli R, Montecucco C (1997) Guidebook to protein toxins and their use in cell biology. Oxford: Oxford University Press. p. 256.

Rathmayer W (1978) Venoms of Sphecidae, Pompilidae, Mutillidae and Bethylidae. In: Handbook of Experimental Pharmacology (Bettini S, ed), Vol 48 Arthropod Venoms, pp. 661-690. Heidelberg: Springer.

Reingold SC, Camhi JM (1977) A quantitative analysis of rhythmic leg movements during three different behaviors in the cockroach, Periplaneta americana. J Insect Physiol 23: 1407–1420.

Ridgel AL, Ritzmann RE (2005) Effects of neck and circumesophageal connective lesions on posture and locomotion in the cockroach. J Comp Physiol A 191: 559–573.

Ritzmann RE, Pollack AJ, Hudson SE, Hyvonen A (1991) Convergence of multimodal sensory signals at thoracic interneurons of the escape system of the cockroach, Periplaneta americana. Brain Res 563: 175-183.

~ 64 ~

Roeder KD (1963) Nerve Cells and Insect Behavior. Cambridge: Harvard University Press. p. 238.

Romig T, Lucius R, Frank W (1980) Cerebral larvae in the second intermediate host of Dicrocoelium dendriticum (Rudolphi, 1819) and Dicrocoelium hospes Looss, 1907 (Trematodes, Dicrocoeliidae). Z Parasitenkd 63: 277–286.

Rosenberg LA, Glusman JG, Libersat F (2007) Octopamine partially restores walking in hypokinetic cockroaches stung by the parasitoid wasp Ampulex compressa. J Exp Biol 210: 4411–4417.

Rosenberg LA, Pflueger HJ, Wegener G, Libersat F (2006) Wasp venom injected into the prey’s brain modulates thoracic identified monoaminergic neurons. J Neurobiol 66: 155–168.

Schaefer P, Ritzmann RE (2001) Descending influences on escape behavior and motor pattern in the cockroach. J Neurobiol 49: 9–28.

Schaefer PL, Kondagunta GV, Ritzmann RE (1994) Motion analysis of escape movements evoked by tactile stimulation in the cockroach Periplaneta americana. J Exp Biol 190: 287–294.

Schneider G, Hohorst W (1971) Wanderung der Metacercariae des Lanzett-Egels in Ameisen. Naturwissenschaften 58: 327–328.

Sinakevitch I, Niwa M, Strausfeld NJ (2005) Octopamine-like immunoreactivity in the honey bee and cockroach: comparable organization in the brain and subesophageal ganglion. J Comp Neurol 488: 233–254.

Steiner AL (1976) Digger wasp predatory behavior (Hymenoptera, Sphecidae) II. Comparative study of closely related wasps (Larrinae: Liris nigra, Palearctic; L. argentata and L. aequalis, Nearctic) that all paralyze crickets (Orthoptera, Gryllidae). Z Tierpsychol 42: 343–380.

Steiner AL (1986) Stinging behaviour of solitary wasps. In: Venoms of the Hymenoptera (ed. Piek T), pp. 63–160. London: Academic.

~ 65 ~

Strausfeld NJ (1999) A brain region in insects that supervises walking. Prog Brain Res 123: 273–284.

Strauss R (2002) The central complex and the genetic dissection of locomotor behaviour. Curr Opin Neurobiol 12: 633–638.

Strauss R, Hanesch U, Kinkelin M, Wolf R, Heisenberg M (1992) no-bridge of Drosophila melanogaster: portrait of a structural mutant of the central complex. J Neurogenet 8: 125-155.

Strauss R, Heisenberg M (1993) A higher control center of locomotor behaviour in the Drosophila brain. J Neurosci 13: 1852–1861.

Thomas JA, Knapp JJ, Akino T, Gerty S, Wakamura S, et al. (2002) Parasitoid secretions provoke ant warfare. Nature 417: 505–506.

Thomas F, Adamo S, Moore J (2005) Parasitic manipulation: Where are we and where should we go? Behav Proc. 68: 185–199. van Baaren J, Nenon JP (1994) Factors involved in host discrimination by Epidinocarsis lopezi and Leptomastix dactylopii (Hym., Encyrtidae). J Appl Ent 118: 76-83. van Lenteren JC, Ruschioni S, Romani R, van Loon JJ, Qiu YT, Smid HM, Isidoro N, Bin F (2007) Structure and electrophysiological responses of gustatory organs on the ovipositor of the parasitoid Leptopilina heterotoma. Arthropod Struct Dev 36: 271- 276. van Swinderen B (2005) The remote roots of consciousness in fruit-fly selective attention? Bioessays 27: 321-330. van Swinderen B, Andretic R (2003) Arousal in Drosophila. Behav Processes 64: 133-144.

Watson JT, Ritzmann RE (1994) The escape response versus the quiescent response of the American cockroach: behavioural choice mediated by physiological state. Anim Behav 48: 476-478.

~ 66 ~

Weisel-Eichler A, Haspel G, Libersat F (1999) Venom of a parasitoid wasp induces prolonged grooming in the cockroach. J Exp Biol 202: 957–964

Weisel-Eichler A, Libersat F (2002) Are monoaminergic systems involved in the lethargy induced by a parasitoid wasp in the cockroach prey? J Comp Physiol A 188: 315–324.

Wessnitzer J, Webb B (2006) Multimodal sensory integration in insects - towards insect brain control architectures. Bioinspir Biomim 1: 63-75.

Williams FX (1942) Ampulex compressa (Fabr.), a cockroach-hunting wasp introduced from New Caledonia into Hawaii. Proc Haw Entomol Soc 11: 221-233.

Zill SN (1986) A model of pattern generation of cockroach walking reconsidered. J Neurobiol 17: 317–328.

Zimmer C (2001) Parasite Rex: Inside the Bizarre World of Nature’s Most Dangerous Creatures. New York: Free Press.

~ 67 ~

לבסוף, חקרתי את המנגנונים הסנסוריים המאפשרים לצרעה לאתר את גנגליוני הראש של התיקן בתוך קפסולת הראש (ה"גולגולת") שלו. על - ידי שילוב של מחקר התנהגותי, מיקרוסקופיה אלקטרונית ורישומים אלקטרו-פיסיולוגיים , הראיתי כי בקצהו של עוקץ הצרעה ממוקמים אברי חוש, ככל הנראה חיישני טעם, המאפשרים לצרעה להבחין בין רקמות עצביות לרקמות אחרות בתוך קפסולת הראש של התיקן במהלך העקיצה.

בעבודה שלהלן אתאר את דרך המחשבה, השיטות והתוצאות שאספתי במהלך חקירתי את מערכת היחסים המיוחדת בין הצרעה לתיקן. בהסתמך על תוצאותיי ובשילוב עם תוצאות מחקרים קודמים אציע מודל מנגנוני המתאר כיצד ארס המוזרק לראש מעכב את המוטיבציה ואת יכולת יצירת התנועה בתיקנים עקוצים, ובכך שולל מהם את 'חופש הבחירה .'

מילות מפתח: צרעה; תיקן; טפילות; נוירואתולוגיה; התנהגות מוטורית; ארס; מוטיבציה; גנגליון תת - לועי; אברי חוש. תקציר העבודה

הצרעה הטפילית Ampulex compressa צדה תיקנים כמקור מזון חי לצאצאיה. הצרעה עוקצת תיקן אל תוך ראשו ומזריקה קוקטייל של ארס ישירות אל תוך גנגליוני הראש שלו. עקיצה זו משרה שינוי התנהגותי בתיקן, שהופך לאחר העקיצה להיפוקינטי ואינו מסוגל ליזום תנועה רצונית במשך מספר ימים. אולם, על אף שאינו נע בעצמו, התיקן העקוץ אינו משותק. תוך הפגנת דפוס הליכה נורמאלי הוא עוקב אחרי הצרעה בעוד היא מושכת באחד ממחושיו ומובילה אותו, ככלב ברצועה , אל תוך מאורה שבחרה קודם לכן. לאחר שהיא ממקמת אותו במאור הת , מטילה הצרעה ביצה בודדת על גוף התיקן ואז יוצאת, אוטמת את המאורה מבחוץ ועוזבת את התיקן העקוץ בפנים. כאשר זחל הצרעה בוקע מהביצה כיומיים לאחר מכן, הוא ניזון מהתיקן החי במשך ארבעה ימים נוספים מבלי שהתיקן ינסה להתנגד או לברוח . כאשר הזחל מוכן להתגלם הוא טווה גולם בתוך בטן התיקן ועובר מטמורפוזה. כחודש לאחר מכן בוקעת מתוך התיקן צרעה בוגרת, מוכנה להמשיך את מחזור החיים.

מטרת המחקר המוצג בעבודה זו הייתה לאפיין את המנגנונים המאפשרים לצרעה להשרות את השינוי ההתנהגותי בתיקן, ובכך לשלול ממנו את "רצונו העצמי" ואת יכולתו לייצר תנועה באופן יזום. כצעד ראשון חקרתי כיצד כל אחד משני גנגליוני הראש, הנחשבים 'מרכזים עצביים גבוהים' בחרקים, מווסת פעילויו ות מוטורי ת שונות בתיקנים שאינם עקוצים. לצורך כך ביצעתי חיתוכים ממוקדים בגנגליוני הראש וכימתתי את התוצאות ברמה פיסיולוגית והתנהגותית. בהסתמך על התוצאות בניתי מודל המתאר כיצד גירויים סנסוריים שונים מביאים לביטוי דפוסי התנהגות שונים בתיקנים, וכיצד אלו מווסתים על - ידי כל אחד מגנגליוני הראש.

בשלב השני חקרתי לעומק אילו וכיצד דפוסי התנהגות בתיקנים משתנים לאחר עקיצת הצרעה. תוצאות המחקר הראו כי הארס אינו משרה מצב 'דמוי-שינה' כללי בתיקנים העקוצים, אלא מדכא באופן סלקטיבי את המוטיבציה שלהם ליזום ולשמר התנהגויות הקשורות בהליכה. רישומים אלקטרו - פיסיולוגיים משרירי הרגליים הראו כי מידע עצבי מעורר היורד מגנגליוני הראש אל מרכזים מוטוריים בחזה מעוכב בתיקנים עקוצים, דבר היכול להסביר את הירידה ביכולתם ליזום ולשמר הליכה.

בשלב השלישי ה תש משתי בכלים התנהגותיים, אלקטרו-פיסיולוגיים ונוירו-פרמקולוגיים בכדי לבדוק מי משני גנגליוני הראש אחראי על השינוי ההתנהגותי שמשרה ארס הצרעה. תוצאותי הראו כי הארס מעכב פעילות עצבית בגנגליון התת - לועי (Sub-esophageal ganglion) של תיקנים עקוצים בהשוואה לתיקנים שאינם עקוצים. יתרה מכך, בעזרת הזרקה של ארס שנחלב מצרעות או של חומר הרדמה מלאכותי ישירות לא תוך הגנגליון התת-לועי, הראיתי כי עיכוב פעילות עצבית בגנגליון זה מעכב יצירת תנועה בתיקנים שאינם עקוצים, בדומה לתופעות שנצפו בתיקנים עקוצים. ירידה בתנועה לא נצפתה, לעומת זאת, כתוצאה מהזרקת ארס או חומר הרדמה לא תוך הגנגליון העל - לועי. נתונים אלו מאשרים את ההשערה לפיה המטרה העצבית העיקרית של הארס ביצירת היפוקינזיה בתיקנים עקוצים היא הגנגליון התת-לועי. בכך גם מדגיש המחקר את חשיבותו של גנגליון זה בוויסות מנגנונים מוטוריים של הליכה בחרקים.

העבודה נעשתה בהדרכת פרופסור פרדריק ליברסאט

במחלקה למדעי החיים בפקולטה למדעי הטבע

מנגנונים של הזרקת ארס ושינוי התנהגות של תיקן על-ידי צרעה טפילית

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

רם גל

הוגש לסינאט אוניברסיטת בן גוריון בנגב

אישור המנחה ______

אישור דיקן בית הספר ללימודי מחקר מתקדמים ע "ש קרייטמן ______

אדר תש"ע מרץ 2010

באר שבע

מנגנונים של הזרקת ארס ושינוי התנהגות של תיקן על - ידי צרעה טפילית

מחקר לשם מילוי חלקי של הדרישות לקבלת תואר "דוקטור לפילוסופיה"

מאת

רם גל

הוגש לסינאט אוניברסיטת בן גוריון בנגב

אדר תש"ע מרץ 2010

באר שבע