Behavioral experiments

Nineteen animals were used to examine the effect of temperature acclimation on escape kinematics. The goldfish used for these studies had comparable body lengths (5°C: 10.93 ± 0.33 cm; 15°C: 11.52 ± 0.22 cm; 25°C: 10.58 ± 0.36; P = 0.132) and weights (5°C: 43.75 ± 2.62 g; 15°C: 46.99 ± 5.60 g; 25°C: 40.37 ± 2.42 g; P = 0.491). We examined kinematics of the C-start behavior in goldfish acclimated to three temperatures: 5, 15, and 25°C. The tank used to examine escape behavior was a 21 × 26-cm clear plastic container with a grid taped to the bottom outside for measuring distance (Fig. 1A). Each square on the grid represented 1 cm². Startle responses were elicited by a rubberized hammer contacting one side of the test container. A black sheet was taped on the side of the tank where the hammer hit so that the fish could not see the hammer approach. This behavioral setup was contained within a plastic tub covered by black sheeting so the fish were distracted as little as possible by movements and sounds in the laboratory. The stimulating hammer was drawn back by a string extending under the outer sheeting. A mirror was placed on a 45° angle above the tank and a high-speed video camera (1200 digital high-speed 10-bit CMOS camera system, PCO/Cooke, Romulus, MI) recorded the escape behavior from the reflection in the mirror at 1 frame/ms on a PC. Recordings started just before the hammer was released and captured the behavior of the fish prior to, during, and following contact of the hammer with the container. Two lights sitting near the camera illuminated the fish via the mirror.

Open in a separate window

FIG. 1.

Analysis of behavioral kinematics. A: still image of a fish in the behavioral test tank. A grid was taped underneath the tank with 1-cm² markings for calibration purposes. A hanging rubber hammer was pulled back by a string to contact the container and the corresponding side of the container was covered with dark paper so that the fish could not see the approach of the hammer. Escape responses were recorded for about 300 ms before, during, and after the hammer contacted the side of the tank. B: examples of the digitized escape behavior for 3 fish at 5, 15, and 25°C. For each frame, the position of the head of the fish was digitized by marking the nose (indicated by red dots) and the back of the skull; the blue line connects these 2 points. Green circles indicate the position of the nose of the fish at the onset of each recording. Recordings were made at 1 frame/ms so the distance between each line indicates the distance moved by the head/ms. C: example of smoothing analysis of the digitized data. The blue trace represents the quantification of the raw data (head speed in this example). The red trace represents the smoothed data used for peak amplitude measurements reported in these studies.

Animals were placed in the recording chamber in their native tank water and taping of the escape behavior began within 5–10 min. Each fish was given 10 escape trials at its acclimated temperature, with about 3–4 min between trials. Fish acclimated to 5°C were then acutely warmed 10° over the course of 1 h in the same aerated tank water (a bubbler and heating unit were added to the water). After 1 h, they were then given an additional 10 trials at 15°C. Similarly, fish acclimated to 25° were acutely chilled (Arctica Titanium Chiller, Petsolutions.com, Beavercreek, OH) in their own aerated tank water over the course of 1 h following an initial 10 trials and then given another set of 10 trials at 15°C. In both cases, water was also circulated through a chiller that continuously monitored water temperature.

The escape response in the goldfish has been well described (Eaton et al. 1977). When one M-cell fires, it simultaneously triggers multiple motor neurons on the contralateral side of the fish's body so that muscles on that side contract and the fish forms a C-bend, or C-start. The C-start consists of two basic parts: Stage 1, which rotates the body around the center of mass; and Stage 2, which involves an axial acceleration that moves the center of mass 2–6 cm (Eaton et al. 1988). Only escapes that fit the definition of a C-start were examined in this study. Nonresponses and escapes immediately into the side of the container were also not included in the analysis of kinematics; i.e., maximum head speed, angular velocity, and angular acceleration were measured in cases where the fish started in the middle of the chamber and was able to execute a full turn. Only the kinematics of Stage 1 were examined because Stage 2 was often constrained, either by the size limitation of the container or by the reverberations of water off of the sides of the container. Escape probability and onset latency were measured for all escapes, since these factors were not influenced by limitations of the testing container. Escapes were analyzed using Image J (National Institutes of Health) by placing a dot in the middle of the fish's nose and the middle of the back of the head for each frame of the 300-ms trial. In this manner, the x and y coordinates were recorded and the location of the head in each frame was digitized (Fig. 1B). For each trial, distance was first calibrated to the grid on the bottom of the container.

Calculation of swimming kinematics: angular velocity, angular acceleration, and head speed

Velocities and accelerations were calculated in Matlab. From the head and body positions, body angle (θ) was determined with arctangent (allowing θ to exceed the interval [−π, π] so as to remain continuous), then θ and the x and y components of head position were smoothed using a low-pass filter (fourth-order Butterworth forward and reverse digital filter with a cutoff frequency of 1/20 the sample frequency). Angular velocity (ω) and head velocity (v) were then calculated from the smoothed quantities using finite differences with respect to time (t) and a step size of two, i.e.

The step size for finite differences was chosen to balance reduction of noise with elimination of features (both increase as the step size is increased). For the purpose of calculating angular acceleration (α), a smoothed version of angular velocity (ω*) was computed, using another low-pass filter (fourth-order Butterworth, but with a cutoff of 1/32 the sample frequency). Then angular acceleration was calculated using finite differences with a step size of four

Electrophysiological experiments

Fish were initially anesthetized using 20 mg/l 3-aminobenzoic acid ethyl ester (MS-222; Sigma, St. Louis, MO). They were then injected intramuscularly with curare (1 μg/g body weight), to paralyze them for the duration of the experiment, and mounted in the recording chamber. They were respirated by flowing water from their acclimation tank through their mouth and over their gills. The water was maintained at the temperature of their acclimation tank with a chiller and contained 20 mg/l MS-222. The spinal cord was exposed for antidromic stimulation of the M-cell that, because the M-cell extracellular spike is distinctively recognizable, allows identification of the field in the intact brain. The fish medulla was exposed to permit intracellular recording from the M-cell. An additional opening was made lateral to the medulla over the inner ear of the fish so that the VIIIth nerve could be exposed for stimulation. Fish brains were then continually superfused with saline (in mM: 124.0 NaCl, 5.1 KCl, 2.8 NaH₂PO₄·H₂O, 0.9 MgSO₄, 1.6 CaCl₂·2H₂O, 5.6 glucose, and 20.0 HEPES, pH 7.2) at the appropriate acclimation temperature (5, 15, or 25°C). Fish saline was maintained at the same temperature as the perfusion water by hanging a beaker in the perfusion water so that the rim was above water level, whereas the rest of the beaker was submerged. Saline temperature was then measured throughout each experiment.

Recording procedures and identification of the M-cell have been described previously (Faber and Korn 1978). M-cell intracellular responses to sound as well as to posterior eighth nerve (orthodromic) and spinal cord (antidromic) stimulation were recorded in the soma (50 μm lateral to the axon cap) as well as distally along the cell's lateral dendrite using 4- to 7-MΩ microelectrodes filled with 5 M potassium acetate (KAc) and 0.1 M potassium chloride (KCl). All recordings were performed in current clamp (Axoclamp 9A; Axon Instruments Foster City, CA). Data were digitized using a Digidata (Axon Instruments) and collected and analyzed using pCLAMP 10.2 software. A Master-8 (AMPI) was used to synchronize stimulation of the spinal cord or VIIIth nerve (Digidata) with recordings.

Warm acclimation increases escape probability and reduces directionality

Since studies of acute temperature change have demonstrated that cellular properties such as input resistance can change with temperature, one might also expect the integration of information by the M-cell to be altered. One way to address this is by examining the effect of temperature acclimation on M-cell threshold properties. Since the M-cell triggers the escape response, a determination of escape probability is an indicator of successful M-cell firing. Probability of escape in animals acclimated to 25°C was 100.0 ± 0.0% (n = 7) for all animals, a value significantly greater (P = 0.015) than both the 5°C- (74.8 ± 8.8%, n = 6) and 15°C-acclimated populations (75.6 ± 5.7%, n = 6), respectively, which were not significantly different from each other (Fig. 2B). Therefore M-cell firing was greater in the population of animals acclimated to 25°C than those acclimated to 5 and 15°C.

Another measure that addresses neural processing by the M-cell circuit is the ability of an animal to escape in the appropriate direction; i.e., away from an aversive stimulus instead of toward it (e.g., Fig. 3C). This measure addresses the balance between excitatory and inhibitory inputs to the right versus left M-cell in the presence of an abrupt stimulus. Escapes used for these studies required that a fish initially be oriented perpendicular to a wall hit by a hammer so that a clear determination of “toward” and “away” could be made (Fig. 3C). A comparison of cumulative trials for all animals at each temperature demonstrated that 5 and 15°C populations possessed similar directionality at their acclimation temperatures, escaping away from the hammer 82.6% (N = 23) and 83.3% (N = 30) of the time, respectively. In contrast, animals acclimated to 25°C made only 58.3% (N = 36) correct escapes (P = 0.045, χ² test; Fig. 3D). Note that escape directionality for 25°C-acclimated animals is essentially the same as chance. When directionality percentages were first calculated for each animal, and then averaged for each temperature, similar results were seen [5°C = 85 ± 7% (n = 6); 15°C = 87 ± 8% (n = 6); 25°C = 61 ± 7% (n = 7); P = 0.06]. Therefore the ability to discriminate appropriate escape directionality (i.e., firing of the appropriate M-cell) is disrupted in the 25°C population.

M-cell action potentials are broader in cold, reduced in warm

To better understand the cellular properties underlying firing of the M-cell in the three populations, we recorded from the M-cell intracellularly in animals that had been acclimated to the same conditions as those used in behavioral experiments. In all three acclimation groups (5, 15, and 25°C), resting membrane potentials were in the range of −79 to −83 mV in both the soma and dendrite and did not differ significantly between populations (Kruskal–Wallis, one-way ANOVA on ranks, P = 0.10). We first examined characteristics of the action potential (AP), since this event reflects kinetics of sodium and potassium channels as well as cellular input resistance. Spike height was essentially the same for the three populations (Fig. 4A) and the antidromically evoked AP varied by no more than 5% in amplitude in each population.

Open in a separate window

FIG. 4.

Effect of temperature acclimation on M-cell action potential (AP) kinetics. For all plots, black points represent measurements for each fish; red represents mean ± SD. Top left: representative APs from 5°C (blue), 15°C (black), and 25°C (red) animals superimposed by aligning the peaks. Dashed lines represent the 10–90% rise and decay of the AP. Quantifications of AP kinetics were made for the 3 acclimated groups (5, 15, and 25°C). A: peak amplitude (mV): 5°C = 37.29 ± 2.29 (n = 9); 15°C = 36.88 ± 1.03 (n = 13); 25°C = 35.41 ± 2.55 (n = 8); no significant differences (NSD). B: width at half-amplitude (half-width) (ms): 5°C = 0.65 ± 0.04 (n = 9); 15°C = 0.49 ± 0.02 (n = 13); 25°C = 0.36 ± 0.01 (n = 8); all groups are significantly different. C: latency from antidromic stimulation to AP onset (ms): 5°C = 0.59 ± 0.04 (n = 9); 15°C = 0.51 ± 0.04 (n = 13); 25°C = 0.39 ± 0.02 (n = 8); 5 and 25°C groups are significantly different. D: rate of rise (mV/ms): 5°C = 129.74 ± 8.01 (n = 9); 15°C = 179.68 ± 6.49 (n = 13); 25°C = 228.59 ± 19.28 (n = 8); all groups are significantly different. E: rate of repolarization (mV/ms): 5°C = −42.95 ± 4.81 (n = 9); 15°C = −61.22 ± 3.58 (n = 13); 25°C = −74.18 ± 9.34 (n = 8); 5 and 25°C groups are significantly different.

We next took a closer look at the functioning of sodium and potassium channels during the AP by examining the kinetics of depolarization and repolarization. AP half-width varied significantly between populations, as it was 31.3% broader in 5°C and 26.9% narrower in 25°C- versus 15°C-acclimated animals (Fig. 4B). Compared with 15°C animals, the changes in 5°C-acclimated animals corresponded to a reduction in the slope of the 10–90% rising phase of the AP of 40.3% and a decrease of 29.8% in the 10–90% decay slope (Fig. 4, D and E). Animals at 25°C showed similar temperature-related changes in comparison to 15°C animals with a significant increase in rise slope of 27.2%; however, decay kinetics were not significantly different between 15 and 25°C populations.

We also examined the time from antidromic stimulation to AP onset, which reflects conduction velocity of the M-axon, since animals in different groups were not significantly different in length and all animals were stimulated antidromically in the same region of the spinal cord (see methods). AP onset latency was 25.5% faster in 25 than in 15°C animals, and 5°C populations were 15.7% slower (Fig. 4C), although only 5 and 25°C populations were significantly different. Increases in M-axon conduction velocity in warm-acclimated populations should at least partially account for reduced C-start onset latencies.

M-cell input resistance is greater in cold, reduced in warm

Although the alterations in AP kinetics described earlier indicated that there were alterations in channel kinetics, it was likely that M-cell neuronal membrane properties differed between populations as well. We therefore first looked at input resistance, which should increase when more channels are closed and decrease when channels open. To determine the input resistance of the M-cell, we performed experiments using two electrodes where current was injected into the M-cell using one electrode while potential changes were recorded with the other (Fig. 5A). Compared with 15°C, values of M-cell input resistance were significantly greater in 5°C fish (43.0%) and lower in 25°C (12.7%; Fig. 5B), indicating changes in the number of open channels and therefore conductance. Injection of current ramps into the M-cell further demonstrated this (Fig. 5D) because, although the threshold membrane potential was not significantly different between these groups (Fig. 5E), the amount of time it took to reach threshold was significantly less in cold-acclimated groups (Fig. 5F). Thus the amount of current required to bring the cell to threshold is decreased in cold-acclimated animals with higher input resistances and a larger number of closed channels and increased in warm-acclimated animals with lower input resistances.

Open in a separate window

FIG. 5.

Effect of temperature acclimation on M-cell membrane properties. For all plots, black points represent measurements for each fish; red represents mean ± SD. A and B: input resistance of the M-cell in acclimated fish. A: voltage changes during current injection in 3 fish: one acclimated to 5°C (blue), one to 15°C (black), and one to 25°C (red). The M-cell was penetrated with 2 electrodes while current steps were injected through one electrode and voltage was recorded with the other. B: measurements of input resistance from acclimated animals (MΩ): 5°C = 0.113 ± 0.005 (n = 2, N = 5); 15°C = 0.079 ± 0.002 (n = 2, N = 8); 25°C = 0.069 ± 0.003 (n = 2, N = 9). All groups are significantly different. C: 63% attenuation was seen at (μm): 5°C = 317.40 ± 31.76 (n = 5); 15°C = 239.37 ± 31.76 (n = 5); 25°C = 253.16 ± 16.48 (n = 4); P = 0.303. D: using 2 electrodes, current ramps were injected into the M-cell through one electrode (bottom, black trace), while voltage was recorded with the other (top: single traces from 5°C (blue), 15°C (black), and 25°C (red) animals; middle: averaged traces). E: measurements of threshold voltage (mV): 5°C = 13.17 ± 0.19 (N = 29); 15°C = 12.61 ± 0.12 (N = 27); 25°C = 13.01 ± 0.11 (N = 35); 5 and 25°C groups are significantly different. F: measurements of time to threshold (ms): 5°C = 11.43 ± 0.10 (n = 2, N = 29); 15°C = 16.09 ± 0.26 (n = 2, N = 27); 25°C = 18.24 ± 0.18 (n = 2, N = 35). All groups are significantly different.

We also estimated changes in space constant in the three populations by examining the attenuation of antidromic spike height by recording sequentially from two sites with the same electrode (attenuation factor = V_dend/V_soma, where the average distance between recording sites was ~220 μm). From these values, we were able to estimate the space constant; i.e., distance required for decay of 63% of the signal (Fig. 5C). In contrast to acute temperature studies which showed significant differences in the space constant (Preuss and Faber 2003), there were no significant differences between groups, suggesting that acclimation of membrane properties compensated at least partially for temperature effects on cable properties.

Excitatory synaptic transmission is reduced in cold

The large myelinated club endings (LMCEs) that contact the lateral dendrite of the M-cell from the auditory (VIIIth) nerve form mixed electrical and chemical synaptic contacts onto the M-cell; thus acclimation could alter passage of current through both gap junctions as well as glutamate receptors at these sites. To determine the effects of temperature acclimation on these contacts, we stimulated the VIIIth nerve orthodromically using an extracellular electrode while recording mixed EPSPs from the M-cell lateral dendrite near the LMCEs. We first measured coupling potential half-width at a peak amplitude of 20 mV, with the assumption being that similar coupling potential amplitudes represent stimulation of similar numbers of VIIIth nerve fibers across populations. This measure is also indicative of the half-width of the presynaptic (VIIIth nerve) action potentials, albeit with some filtering. Coupling potential half-width was 28.9% longer in 5°C- and 35.0% shorter in 25°C-acclimated animals than in 15°C animals and it varied significantly between the three populations (Fig. 6A). These results have implications for synaptic transmission from the VIIIth nerve onto the M-cell.

Open in a separate window

FIG. 6.

Effects of temperature acclimation on excitatory neurotransmission: VIIIth nerve-M-cell mixed synapses. For all plots, black points represent measurements for each fish; red represents mean ± SD. A: coupling potential (CP) half-width. Top: representative excitatory postsynaptic potential (EPSP) in the M-cell in response to direct stimulation of fibers of the VIIIth nerve. Note the downward signal is the stimulus artifact, followed first by the rapid, large-amplitude CP then by the smaller-amplitude EPSP due to the chemical component of the mixed synapse. Bottom: CP half-width at 20 mV: 5°C = 0.36 ± 0.01 (n = 3); 15°C = 0.28 ± 0.03 (n = 3); 25°C = 0.18 ± 0.02 (n = 3); all groups are significantly different. B: latency from the onset of the stimulus artifact to the peak of the coupling potential. Top: representative EPSP in the M-cell in response to direct stimulation of fibers of the VIIIth nerve. Red dots on the M-cell EPSP indicate the artifact onset and CP peak, respectively. Bottom: latency for the 3 acclimated populations: 5°C = 0.44 ± 0.01 (n = 3); 15°C = 0.35 ± 0.04 (n = 3); 25°C = 0.23 ± 0.003 (n = 3); 5 and 25°C groups are significantly different. C: τ from a double-exponential fit of the decay of the chemical EPSP (α = 0.01). Top: representative traces of EPSPs in the M-cell at CP peak = 20 mV in animals acclimated to 5°C (blue), 15°C (black), and 25°C (red), and the corresponding fit of the decay of the chemical EPSP. τ₁ represents the early portion of the decay, τ₂ the later portion. Bottom: τ₁ and τ₂ (ms): 5°C: τ₁ = 0.82 ± 0.15, τ₂ = 3.95 ± 1.11 (n = 3); 15°C: τ₁ = 0.85 ± 0.11, τ₂ = 6.53 ± 1.23 (n = 3); 25°C: τ₁ = 0.48 ± 0.11, τ₂ = 2.24 ± 0.47 (n = 3). Both τ₁ and τ₂ are not significantly different between groups. D: measurements of synaptic strength. Left: CP peak/EPSP peak for many stimulation strengths in 3 fish, one acclimated to 5°C (blue), one to 15°C (black), and one to 25°C (red). Right: slopes from the relationship shown on the left. PSP/CP slope: 5°C = 0.27 ± 0.02 (n = 3); 15°C = 0.41 ± 0.06 (n = 3); 25°C = 0.36 ± 0.02 (n = 3); 5 and 15°C groups are significantly different.

We also measured the time from the onset of the stimulus artifact to the peak of the coupling potential, which reflects the conduction velocity of VIIIth nerve fibers (Fig. 6B). Similar to the M-axon, the latency from stimulation onset to CP peak was shorter in warm-acclimated populations, indicating increased conduction velocity in these animals; similarly, longer latencies indicate slower conduction velocities in cold-acclimated populations. In general, increases in conduction velocity in warm-acclimated populations would reduce the temporal separation of inputs to the two M-cells, resulting in the reduced right–left discrimination and impaired directionality seen in this population.

We next examined the chemical component of the mixed EPSP onto the M-cell. The decay time constant of the PSP reflects glutamatergic channel properties and is an important factor in determining whether responses will summate with repetitive stimulation, a condition favorable to induction of long-term potentiation (LTP) at these synapses. In addition, chemical synaptic transmission has been shown to contribute to the long-term modulation of the club ending gap junctions (Smith and Pereda 2003). The decay of the chemical PSP was best fit with a double-exponential function (α = 0.01), which resulted in two values for τ : τ₁, which averaged 0.5–1 ms, and τ₂, which was about 5 ms (Fig. 6C). Although values for both τ₁ and τ₂ were somewhat reduced at 25°C, no significant differences between groups were apparent for either measure.

We also looked at the strength of excitatory chemical synaptic transmission onto the M-cell in each of the three populations by examining the ratio of chemical PSP amplitude to CP amplitude at various stimulation strengths. Since the amplitude of the chemical PSP depends on coactivation of LMCEs (reviewed in Pereda et al. 2004), and the amplitude of the coupling potential presumably reflects the number of active presynaptic fibers, this ratio should reflect the strength of chemical synaptic transmission for VIIIth nerve fiber populations of similar size, or their input–output relationship. Therefore for each fish we plotted the amplitude of the chemical PSP versus the amplitude of the coupling potential over a range of stimulus intensities, determined the linear fit of this relationship, and then compared the slopes of the fits (Fig. 6D, left). The strength of chemical synaptic transmission in the 15°C population was 47.5% greater than that in the 5°C fish and 12.2% greater than that in the 25°C fish, and only differences between the 5 and 15°C populations were significant (Fig. 6D, right).

Direct effects of temperature on the M-cell and synaptic physiology

Acute and acclimated effects of temperature on synaptic transmission have been described for invertebrates as well as various vertebrate neuromuscular junctions (e.g., Hodgkin and Katz 1949; Macdonald and Montgomery 1986; Rosenthal and Bezanilla 2000; Zhurov and Brezina 2005). In general, these studies have demonstrated that synaptic transmission—a complex, multifactorial process—is more sensitive than membrane properties to extreme temperatures and that temperature effects on transmission are sometimes specific to the preparation in question. Few studies have examined the effects of temperature on synaptic transmission in the vertebrate CNS and those that have focused on acute temperature effects (Eccles et al. 1975; Friedlander et al. 1976; Kushmerick et al. 2006; Preuss and Faber 2003). Although in this study we observed spike broadening in cold, decreases in temperature have been associated with decreased amplitude and increased duration of PSPs at excitatory synapses (Cox and Macdonald 2008). Thus physiological processes that involve multiple factors, such as synaptic transmission, are likely to have complex relationships to temperature.

Temperature acclimation causes structural changes that can influence physiological function on a cellular level. Matheson and Roots (1988) demonstrated that optic nerves in goldfish acclimated to colder temperatures had larger-diameter fibers, a change that would counteract cooling-induced increases in input resistance. The fact that differences in input resistance were not matched by similar differences in the estimated space constant is likely attributable to our method of measuring the latter. The space constant would be more accurately measured using hyperpolarizing current injection at multiple locations along the M-cell and it is possible that there is a significant difference that was not detected using our method of measurement.

In this study, we recorded from the M-cell intracellularly and demonstrated that the direct impact of temperature in acclimated animals was less than that in acute studies. For example, Preuss and Faber (2003) showed a greater than threefold increase in the duration of repolarization following a 10°C drop in temperature. Hodgkin and Katz (1949) also demonstrated that acute temperature decreases have a greater effect on repolarization than depolarization kinetics and Carpenter (1981) showed that potassium channels are more sensitive than sodium channels to temperature. However, in this study, rising–falling phases were influenced approximately the same percentages in each temperature group, indicating some amelioration of direct temperature effects. The fact that acclimation to cold speeds up kinetics of potassium channels has important implications for signal processing; for example, spike broadening with cooling should enhance transmitter release and acclimation can reverse this process.

The inhibitory inputs onto teleost M-cells are well documented (Faber and Korn 1978; Furukawa and Furshpan 1963; Hatta and Korn 1998; Kimmel et al. 1985; Triller and Korn 1981; Zottoli and Faber 1980). Each M-cell is known to receive three types of inhibitory glycinergic input: recurrent, feedback, and feedforward. We therefore characterized the activity of glycinergic transmission by examining feedback inhibition, which should be predictive of effects on all three types of inhibition. Contrary to studies of acute temperature exposure that found reduced inhibition in response to acute cold exposure (Friedlander et al. 1974; Preuss and Faber 2003), we have shown that inhibition acclimates to cold in exactly the opposite manner—by increasing in magnitude. In both this study and studies of acute temperature changes, these alterations in inhibition appear to underlie differences in excitability because animals with reduced inhibition are hyperexcitable regardless of changes in input resistance.

The fact that inhibitory transmission is strongest at 5°C could contribute to the decreased amplitude and duration of excitatory synaptic transmission at this temperature if there is an associated increase in tonic inhibition with greater shunting of excitatory inputs to the M-cell. Since we did not observe a change in M-cell time constant, it is likely that increased glycinergic channel open times or repetitive firings of the inhibitory interneurons are responsible for the long duration of inhibition. However, the reduction of excitatory transmission seen in this population could also be due to the method of measuring the strength of chemical synapses. Since input resistance is greater in the cold-acclimated population, the amplitude of the coupling potential might not represent similar size fiber populations; i.e., at 5°C a 20-mV CP might represent a smaller number of fibers. Since chemical PSP amplitude depends on coactivation of LMCEs, coupling potentials might represent smaller fiber populations at 5°C.

We thank S. Lindell, E. Enos, W. Mebane, J. Simmons, M. Peterson, H. Luther, and R. Langill at the Marine Biological Laboratories in Woods Hole, MA for technical advice and support. We thank the Grass Foundation, Grass Fellows, and Fellowship Directors, C. Carr, G. Haspel, and D. Soares for valuable advice and conversation. We thank E. Marder, A. Taylor, H. Bierman, L. Tang, and especially A. Pereda for critical reading of this manuscript and/or helpful discussion and comments. We also thank E. Marder and L. and C. Szabo for generous support and assistance.

Present address of T. Preuss: Department of Psychology, Hunter College, New York, NY 10021.