Leg Organization

Crayfish, clawed lobsters, and crabs possess four pairs of thoracic legs (Fig. 20.1) that are used in (p. 472) locomotion and one pair of chelipeds generally not used in walking, except in crayfish (Procambarus clarkii) when the animal walks on dry land. Spiny lobsters walk using their five pairs of walking legs. In most decapod crustaceans, legs are composed of seven segments: coxopodite, basipodite, ischiopodite, meropodite, carpopodite, propodite, and dactylopodite. The morphology and the size of these (p. 473) segments are different for walking legs and chelipeds (Fig. 20.1B–E). Note that an extension of the propodite forms the immobile part of each cheliped and the first two pairs of walking legs of crayfish and lobsters. In some crabs (Callinectes sapidus, Portunus pelagicus), the hindmost leg is modified into a flat paddle used for swimming (Fig. 20.1D).

Click to view larger

Figure 20.1 Organization of decapod crustaceans’ legs. (A–C) Legs and chelipeds of crayfish (Procambarus clarkia) showing their insertion on the body (A), the segments and joints of a walking leg (B), and that of a cheliped (C). (D and E) Walking legs, swimming legs, and chelipeds of a blue crab (Callinectes sapidus), showing their insertion on the body (D) and a detail of a walking leg organization (E).

Crustacean leg joints are bicondylar, allowing movements in a single plane, the orientation of which is rotated by 90 degrees from one joint to the next. Therefore, we can distinguish between joints moving in the horizontal plane (thorax-coxa [TC], basis-ischium [BI], carpus-propus [CP]) (see dashed lines in Fig. 20.1B, E) and joints moving in the vertical plane (coxa-basis [CB], merus-carpus [MC], propus-dactyl [PD]) (see magenta dots in Fig. 20.1B, E).

Three main joints are involved in locomotion (Fig. 20.1): the TC joint allows forward and backward movements of the leg; the CB joint is responsible for upward and downward movements; and the MC joint is responsible for extension and flexion of the leg. Note that in crabs, the PD joint is also involved in locomotion due to the large size of the dactyl (Fig. 20.1E) compared to crayfish (Fig. 20.1B). The limited mobility of IM and PC joints reduces their involvement in walking. From a motor command point of view, the proximal joints seem to play the major roles in postural and locomotion activities. In crayfish, each leg is innervated by 95 motor neurons (MNs): 19 promotor MNs and 13 remotor MNs control forward and backward movements at the TC joint, respectively; 19 anterior and 4 posterior levator MNs and 12 depressor MNs control upward and downward movements of the leg, respectively (Bevengut et al., 1996). This proximal motor innervation represents 73 out of the 95 MNs innervating the whole leg (i.e., 77%). Muscles acting on the distal joints of the leg are innervated by a much smaller number of MNs: only one excitatory MN controls the opener muscle at the CP joint, while two MNs (a phasic and a tonic) command the closer muscle (Lang et al., 1980); at the CP joint the stretcher and the bender muscles are innervated by one MN each, and the MN that innervates the stretcher also commands the opener MN; at the MC joint, the extensor and the flexor MNs are innervated by two MNs each (one phasic MN and one tonic MN; Wiens, 1993). In addition to these excitatory MNs, crustacean legs are also innervated by inhibitory MNs: one common inhibitor (CIMN) that innervates each muscle (Rathmayer & Bévengut, 1986) and two specific inhibitory MNs (SI and OI) that innervate the stretcher and the opener, respectively (Wiens & Rathmayer, 1985). Inhibitory MNs seem to accelerate relaxation of muscle fibers, thereby allowing faster movements to occur.

Leg Insertion and Type of Walking

Using in vitro preparations of the walking system of crayfish (Sillar & Skorupski, 1986; Chrachri & Clarac, 1989), neural networks involved in leg control were intensively studied, either during tonic postural activity or during fictive locomotion (Fig. 20.2A) (Sillar et al., 1987; Chrachri & Clarac, 1989, 1990; Clarac et al., 1991; Cattaert et al., 1994a, 1995; Pearlstein et al., 1998a). One important feature of these in vitro studies is that the central nervous system was dissected out with the motor and sensory nerves with the proprioceptor attached. Thereby, it was possible to classify any neuron with respect to its connections with sensory and motor elements in the network.

Locomotor Network: One Central Pattern Generator per Joint

Locomotor activity relies on coordination of joint movements within each leg to achieve alternation between power stroke (stance) and return stroke (swing), and interleg movements to achieve postural control during body displacement. The organization of the locomotor network has been studied in the in vitro preparation of the walking system of crayfish. In such preparations, the motor nerves display tonic activity, with several MNs discharging at 1–5 Hz, and rhythmic locomotor activity does not spontaneously occur. However, it is possible to elicit a fictive walking activity (i.e., pattern of motor activities resembling locomotor activity but without leg movements) by stimulating descending interneurons (Fig. 20.2A) or by perfusing the in vitro preparation with 10^–5 M oxotremorine (a muscarinic agonist of acetylcholine). Although sometimes very slow (40-sec period—Fig. 20.2B), the fairly stable rhythm induced by oxotremorine is generally considered as locomotor-like (Chrachri & Clarac, 1990; Cattaert et al., 1995) because it displays coordination patterns that are characteristic of forward or backward locomotion.

Oxotremorine exposure, particularly when injected directly into the thoracic ganglia, may cause different groups of antagonistic motor neuron pools (e.g., levator/depressor or promotor/remoter) to display different rhythms or cause only one group of antagonistic MNs to be active (Fig. 20.2B). Therefore, a modular scheme was proposed for the organization of the neural network commanding walking leg movements (Cattaert et al., 1995) in which the leg is commanded by a series of joint CPGs that are coordinated during locomotion.

Although these walking CPGs have not yet been described in detail, several features concerning MNs have been established by exposure to muscarinic agonists or after blocking K+ conductances with TEA (tetraethyl ammonium chloride) (Cattaert et al., 1994a). In these conditions, most of MNs display plateau properties (Fig. 20.2C), and a few MNs seem to participate in the rhythm generator, thanks to their pacemaker properties (Fig. 20.2D). During fictive locomotion, it is possible to modulate the locomotor rhythm by injecting current into most MNs. This feature indicates that MNs are part of the locomotor CPG.

Connections Within and Between Central Pattern Generators Controlling a Leg

Pattern generation not only relies on membrane properties of constitutive neurons but also on synaptic connections. It has been demonstrated that MNs innervating the same muscle are electrically coupled (Chrachri & Clarac, 1989) and that antagonistic MNs inhibit each other via inhibitory (p. 475) glutamatergic synapses (Pearlstein et al., 1998b) involving graded transmission (Chrachri & Clarac, 1989). Reciprocal inhibition is a common feature observed in most CPGs (Marder & Calabrese, 1996; Marder & Bucher, 2001; Marder et al., 2005). Postinhibitory rebound, which is present in some CPG neurons, is also a key mechanism in the production of the rhythm and pattern in network-based CPGs (Satterlie, 1985).

Click to view larger

Figure 20.2 Crayfish walking central pattern generator. (A and B) Fictive locomotion activities recorded in vitro from promotor (Pro n), remotor (Rem n), levator (Lev n), and depressor (Dep n) motor nerves of the fifth leg, after stimulation of a descending interneuron (A) and in the presence of oxotremorine (B). (C) Pacemaker activity induced in an intracellularly recorded Rem MN in the presence of the potassium blocker TEA (adapted from Cattaert & Le Ray, 2001, with permission from Elsevier). (D) Plateau properties induced in an intracellularly recorded Dep MN in the presence of oxotremorine (Oxo). In the control condition, a pulse of depolarizing current produces a depolarization restricted to the stimulating pulse duration (Left). In the presence of oxotremorine (Right), the same pulse induces a depolarizing response (plateau) that outlasts the current pulse. The plateau is stopped by injection of a hyperpolarizing current pulse (adapted from Cattaert & Le Ray, 2001, with permission from Elsevier). (E) Injection of depolarizing current pulses (8nA) in an intracellularly recorded coordinating interneuron induces activation of Pro and Lev MNs and inhibition of Rem and Dep MNs recorded from their respective motor nerves. F, Central coordination of CPG rhythms of right legs 2, 3, and 4, monitored by the activity of the remotor nerves.

The locomotor CPG seems to consist of one coupled oscillator for each joint. Four types of “coordinating interneurons” (INs) have been described that coordinate their activity (Fig. 20.2E) (Chrachri & Clarac, 1989). These coordinating INs command, respectively, forward stance, forward swing, backward stance, and backward swing. The commands are mediated by forward connections of INs to MNs and by backward connections from MNs to INs.

(p. 476) Coordinating interneurons may themselves be under the control of descending interneurons (first termed “command” neurons) (Bowerman & Larimer, 1974a, 1974b). Electric stimulation of such descending modulatory neurons elicits various motor behaviors such as low or high posture and forward or backward locomotion (Bowerman & Larimer, 1976; Bowerman, 1977). These descending modulatory neurons have their cell bodies in the brain and project through the circumoesophageal connectives (from where then can be stimulated experimentally) to the thoracic and abdominal ganglia, where they can elicit well-coordinated walking patterns (Fig. 20.2A).

Interleg Central Coordination

We have seen that coordination of joint oscillators within a single leg is achieved within the central nervous system by a diversity of connections involving MNs and specialized interneurons that ensure coordination observed in locomotor pattern. The situation seems to be very different for interlimb coordination. Indeed, in the in vitro preparation, in the absence of sensory feedback, ipsilateral legs tend to be synchronized (Fig. 20.2F), while contralateral legs are generally not coordinated at all (Sillar et al., 1987). The origin of the alternating activities of ipsi- and contralateral legs observed in vivo is therefore likely due to sensory feedback. Indeed, a very similar synchronous pattern of activity of ipsilateral legs can be observed in vivo after autotomy of the legs (Barnes et al., 1972). Because the autotomy plane is located between the basis and ischium, the first joint (TC) keeps working and allows forward and backward movements. In this situation, all the coxae on the same side move in synchrony. The TCMRO may be partially responsible for this coordinating pattern because sinusoidal stimulation applied to the TCMRO of the fourth ganglion evokes an assistance reflex in that ganglion’s remotor MNs that is synchronized with an interganglionic reflex response in the remotor motor neurons of the third ganglion.

Although CPGs are capable of producing coordinated rhythmic MN activities in in vitro preparation of the central nervous system which resemble that observed during locomotion, sensory feedback also play a determinant role in movement control. The next three sections of this chapter will describe sensory structures, sensory-motor circuits, and their dynamics during locomotion.

Each crustacean leg is equipped with a diversity of proprioceptors and exteroceptors that inform continuously the central nervous system on various kinetic and force parameters of leg joints and segments. Some sensory organs such as muscle receptor organs (Sillar & Elson, 1986; Sillar & Skorupski, 1986; Sillar et al., 1986; Skorupski & Sillar, 1988; Elson et al., 1992; Skorupski, 1992) and chordotonal organs (Sillar & Elson, 1986; Sillar & Skorupski, 1986; Sillar et al., 1986; Skorupski & Sillar, 1988; El Manira et al., 1991a, 1991b; Elson et al., 1992) code information on joint kinematic (position, speed, and acceleration). Others, such as such as stress cuticular detectors (Barnes, 1977; Klärner & Barnes, 1986), code forces exerted on various segments, or external stimuli such as water flow.

Chordotonal Organs

Coding

In crabs, chordotonal sensory afferents have been classically divided into two groups, one consisting of movement-sensitive neurons, and the other consisting of position-sensitive neurons (Wiersma & Boettiger, 1959; Mendelson, 1963; Bush 1965a, 1965b). Because sensory neurons are generally unidirectional, both groups were subdivided into two subgroups according to the direction of the movement. From this schema, four types of coding neurons constitute the sensory equipment (p. 477) (p. 478) of COs: opening or closing movement-sensitive afferents, and open or close position-sensitive cells.

Click to view larger

Figure 20.3 Proprioceptors of CB and TC joints in crayfish leg. (A–D) CBCO. Position (A) and general organization (B) of the CBCO (modified from Cattaert & Le Ray, 2001, with permission from Elsevier). (C) Recording site of the microelectrode (ME) within the terminal arbor of CBCO sensory fibers in the thoracic ganglion (modified from Cattaert & El Manira, 1999). (D) Coding properties of CBCO neurons. (E–I) TCMRO and TCCO (modified from Cattaert & Le Ray, 2001, with permission from Elsevier). Position (E) and general organization (F) of the TCMRO (redrawn from Skorupski et al., 1992, and modified from Cannone & Bush, 1981, respectively, with permission from Springer). (G) Depolarization of the T fiber during electrical stimulation of the receptor muscle (RM) (adapted from Cannone & Bush, 1981, with permission from Springer). (H and I) Coding properties of T and S TCMRO fibers (H), and TCCO (I) (redrawn from Skorupski et al., 1992).

In crayfish, chordotonal sensory neurons do not seem to present the same clear-cut specializations (Le Ray et al., 1997a, 1997b). Although the difference between stretch-sensitive and release-sensitive sensory neurons was also described in crayfish coxo-basipodite chordotonal organ (CBCO), neither seems able to code position exclusively, since they are all responding to movement (with different sensitivities). All CBCO neurons are activated for wide angular sectors, with a peak of activity occurring for a different angle value. In response to a ramps-and-plateaus mechanical stimulation applied to the CBCO strand, CBCO neurons present three distinct patterns of discharge (Le Ray et al., 1997a). A first type of CBCO neurons codes velocity, being activated only during the ramp movement (Fig. 20.3D, left). Two other types of afferents combine position and velocity detection. (1) Phasotonic afferents (Fig. 20.3D, middle) produce a phasic firing during ramp movements and a decreasing tonic discharge during the plateaus. Phasic discharge firing rate depends on movement velocity, and tonic discharge firing rate depends on the angular position at the plateau. (2) Continuously firing afferents display a tonic discharge (whatever the position of the joint—Fig. 20.3D, right), the firing rate of which is modulated during movement ramps. Roughly equal proportions of each neuronal type exist among the 40 sensory neurons of the CBCO.

Muscle Receptor Organs

Myo-Chordotonal Organ

The merus-carpus (MC) joint of crustacean limbs comprises two kinds of proprioceptors that monitor MC movements: chordotonal organs (MC1 and MC2) and myo-chordotonal organ (MCO). MCOs consist of chordotonal organs associated with muscles (Barth, 1934). The complex anatomy of MCOs (Clarac, 1968; Clarac & Masson, 1969) varies among crustacean groups (Clarac, 1968, 1970; Clarac & Masson, 1969). We describe here the organization of MCO in crayfish limbs (Clarac & Masson, 1969). MCO is connected with the dorsal muscular bundle of the proximal head of the accessory flexor muscle (AFM) (Fig. 20.4A, B). It is composed of two groups of sensory neurons: MCO1, which is tightly associated with the muscle fibers of AFM; and MCO2, which is located in the ischium and associated with the AFM via a long elastic strand (Fig. 20.4A). As in other COs, MCO sensory neurons (Fig. 20.4C) are divided into stretch- and release-sensitive neurons (Fig. 20.4C). MC joint opening seems to activate a larger number of sensory neurons than does closing (Clarac, 1970). Among MCO sensory neurons, some are sensitive exclusively to movement, while others code also for maintained positions. The sensitivity of MCO neurons to stretching (MC opening) is increased by the contraction of the accessory flexor muscle (Fig. 20.4C) (Clarac & Vedel, 1971). (p. 479)

Click to view larger

Figure 20.4 Proprioceptors of MC joint in crayfish leg. (A) General position of myochordotonal organ (MCO) and MC chordotonal organ (MC) (redrawn from Clarac & Vedel, 1971). (B) Detailed view of MCO1 Clarac & Masson 1969). (C) Coding properties of MCO when sine wave movements are applied to MC joint (o: opening; c: closing of the joint). The activation of the accessory flexor muscle (stimulation of its motor nerve) increases the sensitivity of MCO to MC movements (redrawn from Clarac & Vedel, 1971).

Force Receptors

Funnel Canal Organs

Funnel canal organs (FCOs) are present in the legs of all crustaceans; they have been mostly studied in crabs (Schmidt & Gnatzy, 1984; Zill et al., 1985; Libersat et al., 1987a, 1987b). FCOs are innervated by 3–24 sensory cells, each with a 500–1,400 µm-long single dendrite whose sheathed end passes through a canal in the cuticle. Every FCO contains two type I sensory neurons and between one and 22 type II sensory neurons (Schmidt & Gnatzy, 1984). FCOs code both cuticular strains and strains engendered by muscle contractions (Zill et al., 1985; Libersat et al., 1987b). FCOs located at the tip of the dactyl only respond phasically when bending forces are applied to the receptor. They likely encode contact with the substratum and are also probably vibration sensitive. In contrast, FCOs located more proximal on the dactyl express phasic firing for low-amplitude bending and phasotonic responses for higher levels of stimulation. Proximal receptors also encode the direction and velocity of the force applied (Schmidt & Gnatzy, 1984; Zill et al., 1985). In the crab, all FCO afferents discharge during the stance phase of locomotion while they remain silent during the swing phase (Libersat et al., 1987a) and during swimming (Bévengut et al., 1986). In in vitro preparations, the activity of FCO afferents can be recorded in the sensory nerves that innervate the dactyl (DSA, Fig. 20.5D) when a force is applied on its cuticle via a mechanical stimulator. (p. 480)

Click to view larger

Figure 20.5 Force and stress receptors in crayfish leg. (A) Position of CSD2 (cuticular stress detector) in the ischium (redrawn from Clarac, 1976). (B) Organization of CSD2 sensory neurons in main and accessory strands (redrawn from Clarac, 1976). (C) Recordings of CSD2 activity during walking in crayfish (redrawn from Klärner & Barnes, 1986).

Circuits Involving Joint Receptors

Resistance Reflex

Definition of the Resistance Reflex

In an intact, quiescent animal and in an isolated thoracic nervous system, an imposed joint movement will elicit an MN discharge that resists the imposed movement (Clarac et al., 1978). Functionally, this negative feedback reflex helps maintain a given position. In vertebrates, this principle governs the stretch reflex in which sensory feedback is provided by muscle spindles within the muscle. In crustaceans, the sensory feedback reflex is provided by COs (Fig. 20.3) and muscle receptor organs (MROs) (Fig. 20.4). Here, we will illustrate the organization and function of the resistance reflex with the example of CBCO sensory-motor circuits.

Neural Circuits

Monosynaptic Connections

CB Joint:

In the in vitro preparation of the walking/postural networks, CBCO sensory neurons activated during release of the CBCO strand (i.e., coding for upward movements of the leg) make monosynaptic contacts onto intracellularly recorded Dep MNs (El Manira et al., 1991a). In opposition, CBCO sensory neurons activated during the stretch of the CBCO strand (i.e., coding for downward movements of the leg) make monosynaptic contacts onto Lev MNs (El Manira et al., 1991a). Among Dep MNs, some receive input from up to nine CBCO neurons, others receive input from only one CBCO, and three receive no CBCO input (Hill & Cattaert, 2008). MNs that receive only phasic sensory neuron inputs from CBCO neurons (Hill & Cattaert, 2008) display purely phasic responses to CBCO movement and are insensitive to maintained position (Fig. 20.6E). Other MNs display a response during both movement and maintained CBCO strand position (Fig. 20.6F). In such responses to maintained position, the MN depolarization is related to the level of stretch/release (p. 481) of the CBCO strand (Le Ray et al., 1997). Such MNs receive a mixture of CBCO inputs (some coding for movement, and others coding movement and position) (Hill & Cattaert, 2008).

Click to view larger

Figure 20.6 Illustration of reflex resistance in TC and CB joints. (A) Experimental procedure used to apply mechanical stimulation to TCMRO and TCCO in the crayfish in in vitro preparation. (B) Extracellular recording from the promotor and remotor motor nerves during ramp movement applied to the TCMRO + TCCO. R: release, S: stretch. (C) In vitro preparation of crayfish walking system used to study resistance reflex at the CB joint. Movement (mvt) is applied to the CBCO strand while a depressor MN (Dep MN) is intracellularly recorded. (D and E) Intracellular phasic (D) and phasotonic (E) responses of two depressor MNs to ramp release (R) of the CBCO strand.

TC Joint:

TC joint movements are monitored by TCMRO and TCCO. TCMRO stretch induces promotor MN discharge, while TCMRO release activates the remotor MN (Fig. 20.6A) (Elson et al., 1992). Promotor MN bursts can also be induced directly by depolarizing the S fiber or hyperpolarizing the T fiber, while converse stimulation evokes remotor MN bursts. The muscle part of TCMRO is controlled by rm1 MN that is spontaneously active. Rm1 is excited by TCMRO stretch (T fiber depolarization) and inhibited during TCMRO release (T fiber hyperpolarization). When the receptor is almost maximally stretched, rm1 receives a tonic inhibition from the S fiber. During forward movements, the thorax-coxa chordotonal organ TCCO (Skorupski & Bush, 1992) is activated (see Fig. 20.3I) and excites rm1, which prevents the slackening of the TCMRO. Parallel stimulation of both proprioceptors in an otherwise isolated preparation causes reflex responses of promotor and remotor MNs to occur on both stretch and release, which indicates that stretch-evoked reflexes are produced by the TCMRO while release-evoked ones are due to the TCCO (Skorupski, Rawat, & Bush, 1992).

(p. 482) MC Joint:

Unlike the reflex responses involving the CBCO and Lev/Dep MNs, the neural circuits involved in MC joint reflexes have not been described. As is the case of the CB joint, a resistance reflex can be elicited when the MC joint is moved (Vedel & Clarac, 1979) or when MCO or MC_1,2 are directly stretched or released (Vedel et al., 1975). Stretch of the MC_1,2 (or MCO) induces a reflex discharge of flexor MNs, whereas release induces a reflex discharge in extensor MNs. Both motor responses counteract the imposed movement and are therefore typical resistance reflexes. Note that the accessory flexor MNs also display resistance reflex discharges to stretch the MCO or MC_1,2 (Vedel et al., 1975).

Disynaptic Pathways

Excitatory Pathways

Most unitary EPSPs evoked by a single CBCO unit present a disynaptic component in the late phase of the EPSP that is suppressed in the presence of high divalent cation solution. Unlike the monosynaptic component, the late component has a variable latency and amplitude, and it may be much larger than monosynaptic EPSPs. The amplitude depends on the number of recruited INs in the disynaptic pathway, which is controlled by neuromodulators such as serotonin (Le Bon-Jego et al., 2004—see later).

Inhibitory Pathways

Dep MNs receive GABAergic IPSPs from inhibitory interneurons excited by stretch-sensitive CBCO neurons, and Lev MNs receive similar IPSPs from interneurons excited by release-sensitive CBCO neurons (Le Bon-Jego & Cattaert, 2002). Functionally, these cation-sensitive inhibitory pathways participate in the resistance reflex because during an upward leg movement, Dep MNs are activated (counteracting the movement) and at the same time, Lev MNs are inhibited. Similarly, during a downward movement of the leg, Lev MNs are activated (counteracting the movement) and at the same time, Dep MNs are inhibited.

Assistance Reflex

In the preceding paragraphs, we have presented the resistance reflex with involved neural circuits. However, when reflexes are evoked during walking activity, the intensity and the sign of the reflex response change cyclically, indicating that the ongoing motor program controls intrajoint reflexes (Cattaert et al., 1992; Skorupski, 1996; Le Ray & Cattaert, 1997).

Definition of the Assistance Reflex

During spontaneous movements, the sign of a joint reflex may revert from negative (resistance reflex) to positive (assistance reflex). In such occurrences, an imposed movement activates the MNs that assist the movement. Such reflex reversals were observed in vivo and in vitro for the two first leg joints: between TCMRO and the promotor and remotor MNs at the first leg joint (Skorupski & Sillar, 1986; Skorupski, 1992; Skorupski et al., 1992, Skorupski, Vescovi, & Bush, 1994), and between the CBCO and the levator and depressor MNs at the second leg joint (El Manira, Cattaert, & Clarac, 1990; Le Ray & Cattaert, 1997).

Assistance Reflex at TC and CB Joints

TC Joint:

When the locomotor network produces fictive locomotor activity, a phasic stimulation of the TCMRO induces a complete resetting of the ongoing rhythm. Contrary to the case of a tonic preparation, in which movements applied to the TCMRO result in a resistance reflex (see Fig. 20.6B), phasic stimulation applied to a rhythmic preparation evokes an assistance reflex: forward movements activate promotor MNs, and backward movements activate remotor MNs (Fig. 20.7A). Rhythm entrainment by TCMRO has been proposed to result in part from temporally staggered potentials that occur in both S and T fibers when the receptor is stimulated (Elson et al., 1992). As already mentioned earlier for the TCMRO, reflexes evoked by the TCCO release are phase dependent when the preparation is active, while only resistance reflexes are induced in a quiescent preparation. In the active state, both proprioceptor afferents make phase-dependent connections with distinct subgroups of promotor and remotor MNs. A subgroup of promotor MNs is excited in resistance by TCMRO stretch, and a second group is excited in assistance mode by TCCO shortening. In remotor MNs, a comparable division is observed between MNs that are activated in resistance by TCCO release and MNs activated in assistance by TCMRO stretch (Skorupski et al., 1992; Skorupski et al., 1994).

CB Joint:

In an active preparation displaying fictive locomotion, sine wave stretch-release stimulation applied to the CBCO strand can evoke (p. 483) assistance reflexes that entrain the rhythmic activity recorded from both levator and depressor neurograms (Fig. 20.7B). As in vertebrates (Pearson, Ramirez, & Jiang, 1992; Schomburg, Petersen, Barajon, & Hultborn, 1998) and insects (Hess & Buschges, 1999), crayfish sensory-motor connections are strong enough to entrain or even reset the central rhythm (El Manira and Cattaert, unpublished data). Reflex reversal from resistance to assistance involves cyclic modulation of synaptic transmission at sensory-motor synapses (presynaptic inhibition of the CBCO; see later) and activation of assistance reflex INs (ARINs).

Click to view larger

Figure 20.7 Reflex reversal in crayfish locomotor system observed in the presence of oxotremorine in vitro. (A) TC joint: assistance reflex responses recorded from promotor (Pro) and remotor (Rem) motor nerves in an active in vitro preparation, when sine wave movements are applied to the TCMRO (TCMRO mvt). The release of the TCMRO (i.e., forward movement) activates the promotor discharge, while TCMRO stretch (i.e., backward movement) activates the remotor discharge. T and S: intracellular recordings from TCMRO T fiber and S fiber (redrawn from Elson et al., 1992). (B) CB joint: assistance reflex evoked by sine wave movements applied to the CBCO (CBCO mvt) in the levator (Lev) and depressor (Dep) motor nerves in an active in vitro preparation.

Sensory Control of Coordination

Role of Force Receptors in Coordination

Proprioceptive information about the geometry of the leg touching the substrate is essential to ensure efficient interleg coordination when walking on irregular ground and to ensure postural adjustments necessary to preserve the overall vertical body orientation. To these ends, leg force receptors appear to have dramatic effects on the motor pattern. During walking, for example, the stability of the phase relationship between the left and right legs of a given segment is enhanced when the crayfish is loaded (weight increased by 25%–50%) (Clarac & Barnes, 1985).

In the merus-carpus joint of the crayfish cheliped, stimulation of CSD2 afferents evokes a strong increase in the discharge of flexor nerves (Vedel et al., 1975). Moreover, during fictive locomotion, a powerful entrainment of the rhythmic alternation between levator and depressor MNs is provoked by CSD phasic stimulation (Leibrock et al., 1996b).

In the same way, mechanical or electrical stimulation of the FCO afferents induces the resetting of MN activities in the corresponding leg and in adjacent legs in vivo (Schmidt & Gnatzy, 1984; (p. 484) Libersat et al., 1987a). Moreover, during locomotion in the intact crab (Libersat et al., 1987a, 1987b), electrical stimulation of the DSA sensory nerve results in levation of the proper leg and depression of adjacent ones. This reflex is phase dependent, since it is more effective at the end of the stance than at the beginning. If one leg (P3) is blocked in upward position (Fig. 20.8), the phase relationship between adjacent legs is modified (Fig. 20.8A, middle). The rhythmic electrical stimulation of P3 dactyl restores the original phase relationship (Fig. 20.8A, right).

The role of dactyl force proprioceptors in interleg coordination was confirmed in vitro in the crayfish. Electrical stimulation applied to the sensory nerve of the DSA triggers a burst of activity in the levator motor nerve and inhibits the discharge of the depressor motor nerve from the same leg (Fig. 20.8B, left); it produces the inverse effect on the adjacent leg (Fig. 20.8B, right).

Click to view larger

Figure 20.8 Leg coordination by dactyl sensory afferents (DSAs). (A) DSA electrical stimulation restores interleg coordination during free walking in the crab (Carcinus maenas). The activities of the depressor muscles of legs 2–4 are represented by filled rectangle in three conditions: when leg3 is free during walking (left), when leg3 is blocked in elevated position without (center) and with (right) DSA electrical stimulation (redrawn from Libersat et al., 1987b). (B) Neurograms from levator (Lev) and depressor (Dep) motor nerves of the fourth leg during stimulation of DSA of either the fourth (left) or the fifth (right) leg of the crab. Conversed reflex responses are evoked.

(p. 485)

Modulation of Sensory-Motor Circuits

The descending commands, central pattern generators, coordinating neurons, assistance and resistance reflexes, and modulatory effects all come together dynamically in the intact, freely behaving animal. These dynamic interactions mediate the control of posture and of forward, backward, and sideways locomotion, and of transitions between them. Experiments in which spikes from the CBCO sensory nerve and electromyograms from levator muscle were recorded in a freely behaving animal found that orthodromic and antidromic CBCO afferent activity differed from activity recorded from in vitro preparations where the feedback loop was open (Le Ray et al., 2005a). This result indicates that the postural and locomotor control circuitry functions differently when feedback loops are open than when they are closed.

Hybrid System Experiments

To understand the role of proprioceptive feedback in the dynamic control of locomotor behavior, it would be ideal to record the activity of both proprioceptive afferents and leg motor neurons in a freely behaving animal and compare the recordings made with the feedback loops open and closed. The recent development of the neuromechanical simulation software AnimatLab (see http://www.AnimatLab.com) (Cofer et al., 2010) made it possible to simulate the crayfish leg in a computational model and substitute that model leg for the real leg in open- and closed-loop experiments (Fig. 20.9). These “hybrid” experiments made it possible to determine the role of feedback in the dynamic control of locomotion by opening and closing the feedback loops at will.

Click to view larger

Figure 20.9 The hybrid system experimental setup. (Left) The crayfish leg and body model, showing a blowup of the coxo-basal joint (black X), with Dep tonic and phasic muscle heads (pink and red), Lev tonic and phasic muscle heads (light blue and blue), and the CBCO (yellow) all spanning the joint. Leg raises (blue) release the CBO (green arrow up) and leg depressions (red) stretch the CBCO (brown arrow down). (Right) Crayfish in vitro nerve cord. Five thoracic and two abdominal ganglia of the ventral nerve cord were pinned into a saline-filled dish where pin electrodes recorded activity of Lev MNs, Dep MNs, and CBCO afferents. 50 μM oxotremorine was introduced anterior to the nerve cord and equal amounts of saline were withdrawn posterior to the nerve cord. Middle: An interface that links model CBCO stretch and release movements to identical movements of the live CBCO by a speaker-driven probe, and links recorded Lev and Dep MN units to corresponding model motor neurons through a switchable electronic interface.

(Adapted from Chung et al., 2015, with permission)

The model leg was part of an accurate three-dimensional model of a 9 cm crayfish in which the dimensions and masses of leg segments, the angular ranges of leg hinge joints, and the attachments and mechanical properties of muscles were all based on experimental measurements (model elements are italicized). The model existed “under water” in a virtual physical world. In these hybrid system (p. 487) experiments (Chung et al., 2015), motor units recorded from the different motor nerves of an in vitro preparation (Fig. 20.9) drove the corresponding model muscles in the model leg: Dep motor units excited Dep muscles, and Lev motor units excited Lev muscles. The resulting leg movements around the CB joint stretched and released a model CBCO strand that spanned the joint. The length changes of the model CBCO were then used to drive an actuator attached to one end of the live CBCO strand, the other end of which was pinned into the dish containing the nerve cord. Stretch and release movements of the live CBCO excited appropriate afferents in the CBCO nerve, and their activity projected back to the thoracic ganglion that contained the motor neurons, thereby closing the feedback loop. A switch between the recorded motor nerve and the model could be opened or closed at will, to open or close the feedback loop.

The isolated ventral nerve cord is ordinarily in a “quiescent” state, with several CBCO afferents and all leg motor nerves displaying tonic activity. In these experiments, movement of the model leg up or down excited release-sensitive or stretch-sensitive CBCO afferents and resistance reflex responses from the Dep and Lev MNs (Chung et al., 2015). These and other resistance reflex responses are key to maintaining the static posture of a stationary animal (El Manira et al., 1991a).

The quiescent state is maintained by a persistent outward current present in elements of the locomotor CPG and in the assistance neurons (Cattaert et al., 1994b; Le Ray & Cattaert, 1997). By keeping these neurons strongly polarized, this current prevents spontaneous bursting and assistance reflex responses to leg perturbations. The current is blocked by muscarinic agonists such as oxotremorine, which act to depolarize the neurons and bring the CPG and assistance neurons closer to threshold.

Gradual exposure of the in vitro preparation to 50 μM oxotremorine altered the quiescent state responses to include assistance reflex responses: the leg lift stimuli evoked a brief resistance response from Dep MNs and a stronger and longer lasting Lev MN assistance reflex response (Chung et al., 2015). With continued exposure, those Lev MN assistance reflex responses became Lev/Dep burst pairs: a burst of Lev MN activity immediately followed by a similar burst of Dep MN activity.

With chronic oxotremorine exposure, the in vitro preparation generated tonic motor activity that became irregular sequences of Lev/Dep motor burst pairs (Fig. 20.10). This spontaneous low-frequency (~1/30 sec) motor pattern occurred when the preparation was connected to the leg model in open loop, that is, when the motor activity had no effect on the model leg. When the feedback loop was closed by connecting the motor activity to the model muscles, a series of Lev/Dep bursts began immediately upon closing the loop and continued until the loop was opened. The leg moved up when the Lev motor neurons were active and down when the Dep motor neurons were active, and CBCO afferents were excited by the leg movements. The burst frequency in closed loop was nearly three times higher (~1/10 sec) than in open loop, but the durations of both Lev and Dep portions of each burst pair were no different. Bursts of CBCO activity occurred in response to each of these movements, with release-sensitive afferents excited when the leg was raised and stretch-sensitive afferents excited when the leg was lowered.

Recordings from intact animals suggested that closed-loop CBCO afferent responses to stretch and release differed from open-loop afferent responses recorded from in vitro preparations (Le Ray et al., 2005a). The hybrid experiments showed how opening and closing the loop affected the MN responses to those afferents (Bacque-Cazenave et al., 2015). “Pre-burst” Lev MNs were active before the main Lev MN burst and rapid rise of the leg; “onset burst” Lev MNs were active during the initial phase of the burst and the rapid rise of the leg; and “in-burst” neurons were active throughout the burst. CBCO feedback enhanced the pre-burst MNs’ activity before the burst onset and leg rise and reduced it afterward. The responses of onset burst neurons were enhanced for the short interval around the burst onset and leg rise, and in-burst neuron responses were enhanced generally. These results show that individual MNs have defined roles during walking motor bursts and that those roles are shaped by CBCO feedback.

Simulating Locomotor Network Function

Cattaert and Le Ray (2001) have proposed how neuromodulatory-induced changes in thoracic circuitry can mediate transitions from static posture to walking. This proposal was explored in a computational simulation of the hybrid experiments using a network model of the thoracic circuitry that was connected to the crayfish leg model used in those experiments (Fig. 20.11). The network/leg model was tested by simulating the hybrid experiments and comparing the simulated experimental results with (p. 488) the actual experimental results (Bacque-Cazenave et al., 2015).

Click to view larger

Figure 20.10 Open and closed loop responses of the hybrid experimental preparation and the computational model. (A) Responses of the hybrid experimental preparation (Fig. 20.9). Left: CB joint angle (deg), and spikes recorded from the CBCO sensory nerve, and from the Dep and Lev motor nerves when the feedback loop was open (left) and when it was closed (right). Both trials began with the simulation off and the nerve cord was isolated. The simulation start with the model in both open and closed states is marked by the fall of the leg (drop in CBCO angle). (B) Responses of the computational model of the hybrid experiments (Fig. 20.11) in open loop (left column), closed loop (middle column), and closed loop with no Stretch or Release Assistance Reflex interneurons (right column). Traces from top to bottom: 1. CB angle, as in A; 2: CBCO length shows stretch (up) and release (down); 3 and 4: Membrane potential changes, including spikes, in the Release Rate Afferents (3, green) and Stretch Rate Afferents (4, brown); 5 and 6: Tonic and Phasic Lev MN responses (blues); 7 and 8: Tonic and Phasic Dep MN responses (pink and red). Time bar: 10 sec.

(Adapted from Bacque-Cazenave et al., 2015, with permission)

The responses of the network/leg model were similar to the experimental responses recorded when the preparation was quiescent. In closed loop, Lev and Dep MNs were tonically active and the leg was held stationary. Imposed movements of the leg up or down excited resistance reflexes. The Dep and Lev INs, which form a half-center oscillator to create the central pattern generator (Pearlstein et al., (p. 489) 1998b), were silent in the quiescent state, as were the Stretch and Release Assistance Interneurons that mediate disynaptic assistance reflexes. Downward leg movements stretched the CBCO and excited stretch-sensitive afferents that monosynaptically excited tonic and phasic Lev MNs (Fig. 20.11). Their effect was to resist the downward leg movement. Upward leg movements shortened the CBCO and excited release-sensitive afferents that monosynaptically excited tonic and phasic Dep MNs to resist the upward movement.

Click to view larger

Figure 20.11 Neuromechanical model of hybrid preparation. (Left) Crayfish leg and body model as in Figure 20.9. (Right) Neural circuit model substituted for in vitro ventral nerve cord. Colored circles represent neurons or neuron classes. Solid lines: excitatory synapses (forked endings) from MNs to muscles or from MNs to PAD Inhibitory interneurons, or inhibitory synapses (solid balls) between CPG Lev and Dep interneurons. Dotted lines: connections mediating resistance reflexes; Dashed lines: connections mediating assistance reflexes. Black lines with arrows: Targets of OXO modulation. See text for explanation.

(Adapted from Bacque-Cazenave et al., 2015, with permission)

Exposure to oxotremorine was simulated by an OXO interneuron that brought the Dep and Lev INs and the Release and Stretch Assistance neurons near to threshold by blocking a persistent outward current in each target neuron. In this state, disynaptic assistance reflexes were also excited by leg movements. CBCO release-sensitive afferents that responded to a leg lift excited a Release Assistance neuron; it then excited tonic and phasic Lev MNs to help the leg lift. The result was a brief resistance reflex excitation of the Dep MNs immediately followed by a more vigorous assistance reflex excitation of the Lev MNs.

A stronger stimulation of the OXO interneuron further strengthened the assistance reflex responses and also excited inhibitory neurons (e.g., Dep and Lev PADIs) that blocked the resistance reflex responses. OXO stimulation also brought the CPG closer to threshold, which enabled some of the leg lift assistance reflex responses to excite a Lev/Dep MN burst.

In open loop, a still stronger stimulation of OXO led to spontaneous low-frequency Lev/Dep MN bursting. These bursts were very similar to the irregular, low-frequency Lev/Dep bursts observed in the open -loop experiments (Fig. 20.11). Gradually (p. 490) increasing rates of tonic Lev and Dep MN activity between bursts led to either a Lev/Dep tonic and phasic MN burst or a Dep/Lev burst at irregular intervals of about 30 sec. This burst rate is determined by the intrinsic properties of the half-center oscillator. When the feedback loop was closed, spontaneous Lev/Dep bursting at the threefold higher frequency of 1/10 sec began immediately and continued throughout the simulation. This pattern of responses, including reflex reversal, assistance reflex-triggered bursts, irregular open-loop bursting, and higher frequency bursting in closed loop, resulted from increases in the OXO stimulus and closely resembled responses recorded in the hybrid experiments during exposure to a gradually rising concentration of oxotremorine (Fig. 20.11) (Chung et al., 2015).

The model’s ability to replicate the hybrid experiment results suggests that corresponding mechanisms are at work in the hybrid experimental preparation. The increased Lev/Dep burst frequency of the closed-loop model occurred because the assistance reflex triggered the Lev MN burst early in the intrinsic burst cycle. Following each burst, gradual increases in Lev MN activity evoked assistance reflex responses that created a positive feedback loop for Lev MN activity. This positive feedback initiated the Lev MN burst; the Dep MN burst was triggered by a similar positive feedback and by disinhibition when the Lev neurons stopped firing. The early Lev/Dep burst reset the cycle and thereby increased the burst frequency.

Other proprioceptive organs also provide feedback both during static posture and locomotion, including the TCCO and TCMRO at the thoracic-coxa joint of the leg (Sillar et al., 1986) and cuticular stress detectors (CSDs) in response to stress in leg cuticle during walking (Leibrock et al., 1996a). Proprioceptive feedback from the TCCO and TCMRO serves leg promotion and remotion resistance reflexes in quiescent preparations and assistance reflexes in active preparations, similar to the CBCO (DiCaprio & Clarac, 1981; Skorupski & Sillar, 1986). Positive feedback from TCMRO afferents can entrain the locomotor CPG rhythm and increase its frequency. Cuticular stress detectors (CSDs) respond to cuticular forces in the standing leg and excite the levator MNs in a negative feedback reflex (Leibrock et al., 1996b). In rhythmically active preparations, CSDs reverse their reflexes and entrain levator and depressor MN activity. We do not yet know how feedback from these afferents interacts with feedback from CBCO afferents to help control posture and locomotion. To address this, additional experiments that explore cross-joint reflexes and phase-dependent reflex modulation are needed, together with attempts to incorporate all of these elements in a computational reconstruction of this dynamic system.

Models and simulations together provide a window into a complex dynamic system that makes it possible to explore its mechanisms and identify which parameters and parameter values are critical for its performance. Another recently published model asked whether the network organization and interleg reflexes that characterize the six-legged stick insect might also account for the walking gaits and gait transitions observed in crayfish (Grabowska et al., 2015). Like the single leg model described herein, in which sensory feedback was critical for speeding the rhythm, in the eight-legged model, the phase of sensory feedback between legs is critical for establishing stable coordination patterns among the legs.