Abstract

Introduction

The crustacean stomatogastric ganglion (STG) is a central pattern generator network that sits on top of the crustacean stomach and controls the movements of the pylorus and gastric mill. Over the last few decades, the stomatogastric ganglion has become a model system for the study of central pattern generators (CPG). Some other examples of CPGs include pacemaker networks for hearts of all kinds (vertebrate or invertebrate), and networks in the brain responsible for rhythmic locomotory acts like swimming, walking, or flying. CPGs are able to maintain a functionally relevant rhythmic output without requiring input from external sources, and are therefore largely autonomous entities. CPGs are therefore interesting not only to biologists, but to computer and robotics engineers as well. Of particular importance is the essential non-hierarchical nature of the synaptic connections within these networks, which gives rise to the adaptability that is so easy to perceive yet so hard to understand.

The STG was originally attractive to scientists because they recognized it as a pattern generating network comprised of a mere thirty neurons, which led these early investigators to believe that they could reasonably map out the entire system and use it for computer modeling of CPGs. Since this early hope, much has been learned about the workings of the STG and in this process the extent of the complexities involved in even this apparently simple network has been partially unveiled. Even though this network may not be mastered as early investigators once thought, it is an interesting model for pattern generating networks because so much is known about the physiology and the behavior of the system. This large wealth of knowledge set against the relatively simple crustacean nervous system allows very in depth analyses of the workings at a multiplicity of levels.

One reason that studying the crustacean stomatogastric system is relatively easy is that the patterns generated by this network remain intact even after experimental isolation. In addition, the STG is responsible for two distinct motor patterns -- the gastric rhythm and the pyloric rhythm -- controlled by two different sets of motor neurons that innervate separate regions of the stomach (Maynard, 1972). The complete pyloric circuit consists of 14 neurons, one of which is an interneuron and the rest of which are motor neurons. Most of the contacts between these neurons are inhibitory. The pyloric circuit can be reduced to the essential pyloric circuit (diagram 1) of six nodes which is all that is necessary to generate the firing pattern of the pyloric cycle. Within this circuit, the LP and IC neurons fire asynchronously with PY and VD neurons because of the networks of inhibitory connections. The AB cell of the pyloric network acts as the overall frequency controller for the network because of the strong inhibitory connections it makes with all of the other cells. Thus, any alterations in the frequency of the cycle are due to alterations in the AB cell’s function. The AB cell's firing stifles the firing of all the other neurons in the circuit and starts a new cycle (Mulloney, 1987). The gastric circuit is made up of 11 total neurons -- 1 interneuron and 10 motor neurons. This circuit can also be reduced to an essential circuit (diagram 2) containing 5 nodes. Interestingly, the gastric circuit has a non-cyclic (Beltz, 1984) pattern as well as a cyclic pattern. In the non-cyclic condition, the LG and MG are inhibited and the excitation level reaching DG and AM is insufficient to make them fire (Mulloney, 1987). The stable condition of the non-cyclic gastric pattern can be altered by extrinsic inputs or by extrinsic modulation causing the pattern to become cyclic with the DG and AM alternately firing with the LG and MG.

The stomatogastric ganglion receives innervation from the esophageal ganglion (OG) and the commissural ganglia (COG) in vivo (see diagram 3). The innervation by the commissural ganglia is necessary for normal oscillatory properties, and is responsible for the burstiness and the ability to generate plateau potentials (reviewed in Selverston and Moulins, 1985). Although these are the only three major inputs into the STG, there is a whole array of neurotransmitters and neuromodulators used in these innervations. These neurotransmitters and neuromodulators include: proctolin, FMRFamide-like, Histamine, Substance P-like, octopamine, serotonin, and dopamine.

Most of the work done on the crustacean stomatogastric system has come from work on the Homarus americanus and Panulirus interruptus lobsters. There has also been some work done with the Cancer borealis and Cancer Irroratus crabs. Comparisons with the well studied species show that the STG patterns found in C. maenus share a general functional similarity to all crustaceans, but are more similar to the patterns found in the C. borealis STG. In the following experiment we attempted to compare the established physiology of these crustaceans to a novel species of crab, Carcinus maenus. Although most work in this area delves into the specific inhibitory contacts between cells in the STGs of these crustacean species, we will focus mainly on comparing the whole-nerve patterns since we could not perform single cell recording. Since we were limited to such a large-scale analysis we looked mostly for similarities rather than focusing on any differences we found between the STG patterns of established species and those found inC. maenus.

Three successful dissections and preparations of the Carcinus maenus stomatogastric ganglion (STG) were performed according to the protocol provided by Katherine Graubard. This process involved the excision of the stomach, which was then slit ventrally, the lateral and medial teeth were removed, and the stomach was pinned out flat. Nerves were carefully dissected free of the stomach surface, usually proceeding from the anterior to the posterior end. In addition to the STG itself, an attempt was made to maintain the esophageal ganglion, commissural ganglia, the superior and inferior esophageal nerves (SON, ION), and stomatogastric nerve that form the anterior connections to the STG, as well as the lateral, medial, and dorsal ventricular nerves (LVNs, MVNs, DVN) that constitute the main posterior efferent connections. The isolated preparation was then transferred to a transparent glass and silgard dish, and pinned out in an approximation of the native conformation. The dissection took as long as ten hours, and resulted in an isolated stomatogastric system including most of the major nerve branches of the system. The preparation was maintained at temperatures between 0¡ and 15¡ C at all times.

Voltage potentials were measured by conventional extra-cellular recording with a platinum-iridium pin electrode. Amplified activity was successively recorded from all nerves of the system on 4 channels using a Grass Instruments P-55 AC Pre-Amplifier (Astro-Med Inc., West Warwick, RI) and a VR 100 Digital Data Recorder and ITC16 computer interface (Instrutech Corp., Port Washington, NY). The recordings were visualized on a Power Macintosh computer with Acquire (Bruxton Corp, Seattle WA). Three-second sequences were arranged and analyzed from most recordings with IGOR Pro 3.1 (Wavemetrics, Lake Oswego, OR). The vertical axis was calibrated to show a ten millivolt (mV) step for each tick mark.

Terminology in this paper follows that of Maynard and Dando (1974).

Voltage potentials were recorded from the LVN, MVN, DVN, STN, SON, and ION nerves (see Figure 1) both during spontaneous activity and during treatment of a preparation with the neuromodulators dopamine and serotonin (5-HT) added to the ambient solution. The horizontal axes in all of the graphs below demarcate a three second sample, and each tick mark on the vertical axis demarcates a ten millivolt step in the voltage potential for the recording from each electrode. A 500 millisecond square wave is included at the top of each graph as an internal calibration device.

For the second preparation, recordings were successively made from the STN, and the SON and ION on the right side (see figure 3), then from the LVN, DVN, and STN (see figure 4).

Figure 3. Three second segments of recordings from the STN, SON, and ION of the second preparation.

Figure 4. Three second segments of recordings from the LVN, DVN, and STN of the second preparation.

Following initial measurements of spontaneous activity, several experiments were performed on the third preparation. The initial measurements were recorded from the MVN, DVN, and STN; only the DVN showed activity (see figure 5). The preparation was then bathed in 10-5 M dopamine, and activity was recorded from the DVN. The bursting pattern appeared to return to normal within five minutes. Figure 6 is a time course of DVN bursting at 0 minutes, 2.5 minutes, and 5 minutes after addition of dopamine.

Figure 5. Three second segments of recordings from the MVN, DVN, and STN of the third preparation. The MVN and STN do not appear to be active, but the DVN appears to be relatively normal.

Figure 6. Three second segments of recordings from the DVN at 0 minutes, 2.5 minutes, and 5 minutes after addition of 10-5 M dopamine.

Dopamine treatment was halted by repeated washes, and the preparation was allowed to return to its true baseline bursting pattern, as recorded from the LVN and DVN. Recordings from the LVN appear larger than recordings from the DVN despite theoretical concurrence of the axon potential sizes; this is probably due to a weak signal from the electrode placed next to the DVN. The STN was then severed, and the LVN and DVN patterns were clearly altered (see figure 7).

Figure 7. Three second segments of recordings from the LVN and DVN. At the top, bursting patterns returned to normal after washing off the dopamine treatment. At the bottom, both the LVN and DVN patterns were clearly altered by severing the STN.

After severing the STN, we tested the effect of a higher concentration of dopamine (10-4 M). This time it took the preparation much longer to approach baseline bursting patterns. Figure 8 displays a time course of the bursting patterns recorded from the LVN and DVN at 0 minutes, 2.5 minutes, 5 minutes, 7.5 minutes, and 10 minutes after treatment with dopamine.

Figure 8. Three second segments of recordings from the LVN and DVN at 0, 2.5, 5, 7.5, and 10 minutes after treatment with 10-4 M dopamine.

dopamine treatment was removed by repeated washes, and the preparation was allowed to return to baseline bursting patterns. The preparation was then bathed in a treatment of 10-4 M serotonin (5-HT). Figure 9 displays a time course of the bursting patterns recorded from the LVN and DVN at 0 minutes, 2.5 minutes, 5 minutes, and 7.5 minutes, after treatment with 5-HT.

Figure 9. Three second segments of recordings from the LVN and DVN at 0, 2.5, 5, and 7.5, minutes after treatment with 10-4 M 5-HT.

Discussion

Several general patterns of STG function are immediately perceptible from the data. The SON bursting patterns in the first preparation (figure 2) are roughly symmetrical, which shows the necessary bilateral symmetry of the system. The STN in the first and second preparations provides the steady input to the STG that has been proven to be important in P. interruptus and C. borealis. As expected, the large spikes in the SON and the ION patterns are correlated in the second preparation (figure 3), and the LVN and DVN patterns are highly correlated in the third preparation (figures 4, 7-9). Interestingly, the comparative frequency of the spontaneous DVN cycle was quite different between the preparations. The first preparation had a 1 second cycle, the second preparation had a 3 second cycle, and the third had a 2 second cycle. The first preparation probably provides the best estimate of in vivo baseline cyclicity, since there was a long delay between the dissection and the recording of potentials in the second preparation, and the STN was non-functional in the third preparation. Because the STN was nonfunctional in the third preparation, the effect of severing the STN was unexpected. The DVN cycle slowed considerably and gained numerous frequent large spike potentials, indicating that the STN still had some effect on the STG neurons even though there were no visible action potentials traveling through the STN.

The effects of 10-5 M and 10-4 M dopamine and 10-4 M 5-HT on the third preparation are a more complicated issue. In the first dopamine treatment (10-5 M), the frequency of the cyclic pattern in the DVN immediately increased to about double the baseline rate (from a 2 second to a 1 second period), accompanied by the activation of semi-regular large spike potentials. The increase in frequency is due to the strong effect of dopamine on the AB neuron, which controls the frequency of the pyloric rhythm. The large spike potentials are distinct from the normal bursting pattern, and signify the release of a particular class of neurons (like LP) from normal cyclical inhibition patterns. After five minutes, the frequency of the cycle slows to about one cycle every 1.5 seconds, and the large spike potentials were absent. This may be due to a process of desensitization or habituation to the high dopamine concentrations in solution, a process which may be mediated by the cAMP second messenger system (Hempel, 1996). A similar but slower process was found for the 10-4 M dopamine treatment, in which the LVN and DVN cycles remained in sync. The LVN and DVN cycles did not speed up immediately in this case, but were at their maximal frequency (about 1 cycle per second) and maximal number of large spike potentials in the LVN after 2.5 minutes of dopamine treatment. They slowed appreciably after 10 minutes of treatment, and activated far fewer large spike potentials.

Subsequent treatment with 10-4 M 5-HT showed an immediate and sustained increase in the frequency of the cycle (about 1 cycle per 3/4 second), coupled with inconstant amounts of large spike activity. This appears to contradict Beltz et al. (1984), who found that 5-HT slows the LVN cycle down considerably under similar deafferented conditions in Cancer irroratus. They also stated that LVN recording includes action potentials from the LP, PY, and PD neurons, and that the normal alternation of the LP and PD neurons is overpowered by treatment with dopamine. The AB neuron, being the frequency controller neuron, is putatively responsible for quickening the frequency in response to dopamine and to 5-HT. As mentioned earlier, the PD neuron tends to fire simultaneously with AB. Therefore, we hypothesize that the LP neurons are responsible for the irregular large spike potentials recorded in both the dopamine and the 5-HT treatments.

Although the disagreement between our 5-HT data and Beltz's data may be due to species differences, it is far more likely that it is due to our preparation's previous exposure to dopamine, or to experimental error. Further studies of the C. maenus STG should include intracellular recordings of the individual neurons' firing patterns, in order to make a more complete comparison with the STG of other species. Furthermore, in vivo extracellular recording (with a hook electrode) of firing patterns in the LVN or DVN would allow characterization of the modulation of the pyloric and gastric rhythms by sensory cues (seeing or smelling food) and behavior (feeding). The basic pyloric and gastric cycles are well enough understood to permit studies correlating behavior to the neural modulation of the STG.

Special thanks to G. Frank Gwilliam and Ian Kasman in the Biology Dept. at Reed College for their support and help with the technicalities of electrical recording, and to Katherine Graubard and Jana Labenia at the Dept. of Zoology, University of Washington for their excellent tutorial and protocol for the extraction of a functional STG.

Bibliography

Beltz B, Eisen J, Flamm R, Harris-Warrick RM, Hooper SL, and Marder E. Serotonergic innervation and modulation of the stomatogastric ganglion of three decapod crustaceans (Panulirus Interruptus, Homarus Americanus, and Cancer Irroratus). J. Exp, Biol. 1984; 109, 35-54.

Harris-Warrick RM, Marder E, Selverston AI, Moulins M (eds.). Dynamic Biological Networks: The Stomatogastric Nervous System. MIT Press, 1992

Hartline DK, Maynard DM. Motor patterns in the stomatogastric ganglion of the lobster Panulirus argus. J. Exp. Biol., 1975. 62: 405-20.

Heinzel HG. Gastric mill activity in the lobster. I. Spontaneous modes of chewing. J. Neurophysiol., 1988; 59: 528-50.

Hempel CM, Vincent P, Adams SR, Tsien RY, Selverston AI. Spatio-temporal dynamics of cyclic AMP signals in an intact neural circuit. Nature, 1996; 384: 166-9.

Kennedy D, Selverston AI, Remler MP. Analysis of restricted neural networks. Science, 1969; 164: 1448-96.

Marder E, Christie AE, Kilman VL. Functional organization of cotransmission systems: lessons from small nervous systems. Invert. Neurosci ., 1995; 1: 105-12.

Marder E. Modifiability of pattern generation. Curr. Opin. Neurobiol., 1991; 1: 571-6.

Maynard MR, Dando MR. The structure of the stomatogastric neuromuscular system in Callinectes sapidus, Homarus americanus and Panulirus argus (Decapoda Crustacea). Phil. Trans. R. Soc. London B Ser. 1974; 268: 161-220.

Mulloney B. (1987). Neural Circuits. In Selverston, A. I., and Moulins, M. (Ed.), The Crustacean Stomatogastric System. Germany: Springer-Verlag.

Harris-Warrick RM, Marder E. Modulation of neural networks for behavior. Annu. Rev. Neurosci., 1991; 14: 39-57.

Selverston AI. Modeling of neural circuits: what have we learned? Annu. Rev. Neurosci., 1993; 16: 531-46.

Selverston AI. Pattern generation. Curr. Opin. Neurobiol., 1992; 2: 776-80.

Selverston AI, Moulins M. Oscillatory neural networks. Annu. Rev. Physiol., 1985; 47: 29-48.