Unique temperature-activated neurons from pit viper thermosensors

Introduction

Pit vipers (Viperidae: Crotalinae, rattlesnakes, copperheads) have highly evolved sensitive receptors called pit organs, which receive and relay high-resolution thermal information to the brain. This thermal information is thought to be processed and integrated at the optic tectum (28) where it aids in detecting and striking prey (29). Although the pit organ has historically been studied as an infrared (IR) detector, thorough physiological (3) (9) and theoretical studies (17) suggest that the organ responds to thermal rather than photonic stimuli. Our investigations of the optical sensitivity and selectivity of this organ indicate that it is responsive to a broad range of radiation from IR to ultraviolet (26). The lack of spectral selectivity strongly supports the hypothesis that the pit is a highly sensitive thermoreceptor rather than photonic IR detector.

The pit organ shows structural specializations indicative of a high-resolution thermal detector. The temperature sensing terminal nerve endings innervate an ultra-thin (<25 µm) pit membrane (29) that acts as the heat-sensing surface. This thin membrane and the associated terminal nerve mass are suspended in a facial cavity in front of an inner air space, giving the pit organ an extremely low thermal mass. The pit organ also has a rich capillary network that is thought to act as a rapid heat sink (12). This combination of low thermal mass and rapid heat dissipation contributes to the precise temporal and spatial resolution of thermal stimuli.

Multiple and single-unit neuronal activity recorded from pit organ afferents demonstrate that these receptors are sensitive to very small temperature changes (about 0.003°C) (3), and can transmit this thermal information over a broad range of temperatures (15°-37°C). This suggests that the pit organ must have highly-sensitive neurons and ion channel mechanisms capable of receiving and transmitting thermal information and that these complement the anatomical specializations of the pit organ for heat reception. However, the cellular and 4 molecular machinery responsible for converting temperature information into neuronal signals in the pit organ is completely unknown.

A generator potential for thermal stimuli at the pit organ has been identified using extracellular recordings (34), and this supports the hypothesis that the terminal nerve mass may express temperature-sensitive integral membrane proteins. This is also consistent with reports on mammalian thermoreceptors, especially thermal nociceptors, where the activationtemperature threshold, ion conductance, and even behavioral features (7) (6) have been correlated to the presence of a temperature-gated cation channel TRPV1. Thermal nociceptive neurons are fairly plentiful, and thus benefited studies linking TRPV1 to thermal responses. The snake pit organ is a unique system for the study of these lower temperaturethreshold thermoreceptors because it has the highest known density of warm receptors (4) and there are far fewer "warm" receptors in mammals (24) (15) (3). By analogy to TRPV1, ion conductances corresponding to warm receptors may be abundant in neurons that project to the pit organ.

Here we demonstrate for the first time that dissociated neurons from the pit viper TG, which supplies the pit organ (4), shows a heat-sensitive current with a temperature threshold and biophysical properties unlike those identified in mammalian neurons. Voltage clamp recordings revealed an inward monovalent cation current, I T, which increased with heating and tracked temperature change. I T was found in a large proportion of TG neurons that were isolated and had a threshold of about 18°C.

Materials and Methods

Animals. Copperheads (Agkistrodon contortix), Western Diamondback rattlesnakes (Crotalus atrox), and common garter snakes (Thamnophis sirtalis) were caught wild and 5 obtained from a commercial dealer (Glades Herp, Myers FL). A. contortix weight ranged from 100-500 gm, C. atrox from 400-800 gm, and T. sirtalis was less than 100 gm. Snakes were individually housed in either glass aquariums or transparent plastic containers. All containers were supplied with a water bowl, hiding box, and a rock and were placed on a thermal strip to allow for behavioral thermoregulation. The housing room was maintained at an ambient temperature of 24-28°C and a 12/12 light/dark cycle. Animals were fed freshly euthanized baby mice once monthly. All procedures involving the use of animals were conducted in accordance with the "Guide for the Care and Use of Laboratory Animals" published by the National Institutes of Health (NIH publication 85-23, revised 1985).

Culture of Trigeminal Ganglion. Cell bodies that have sensory terminal nerve specializations in the infrared/heat-sensitive pit organ are found in the ganglia of ophthalmic, maxillary, and mandibular branches of the trigeminal nerve (4). Snakes were anesthetized with Isofluorane USP (Abbott Laboratories, North Chicago IL) then decapitatedand the head was placed on ice to minimize cell death after decapitation. The TG was located visually, removed, and placed in ice-cold F12:DMEM (Gibco BRL). Cells were dissociated from the ganglia by treatment with collagenase (1 mg/mL, 45 min, 28°C) and trypsin (2.5 mg/mL, 5 min, 22°C) followed by mechanical dissociation with plastic pipettes. The dissociated cell mixture was layered onto 25 % percoll (Sigma Chemical Co, St Louis MO) and centrifuged for 10 min at 500 xg to remove large cell bodies and debris (11). Cells were cultured in F12:DMEM supplemented with 10% fetal bovine serum, 10 U/mL penicillin, 10 µg/mL streptomycin (Gibco BRL) and 2 mM glutamine (Gibco BRL) on poly-D-lysine coated coverslips or in treated 12-well tissue culture plates (BD Falcon). Cells were maintained at 28°C in 7% CO 2 /93% air for 16-96 hrs.

Results

We performed whole-cell voltage clamp recordings on neurons cultured from the TG.

Our primary culture technique yielded neurons and other cells that were < 60 µm in diameter, the size of neurons we recorded from was 30 ± 9 µm (SD, n=157). Neurons were identified morphologically as phase-bright cells that showed voltage-dependent conductances when 10 mV voltage steps from -90 to + 20 mV (100 or 300 ms) were applied ( Figure 1). Under our culture conditions, neurons did not typically express neurites. Table 1 presents general characteristics recorded from these cells. Figure 1A and C show the two general current responses we recorded from these neurons, with the current voltage relationships shown in Figure 1B and D, respectively. These currents differed only in having a rapid inward component following voltage-dependent outward currents ( Figure 1A, asterisk). The input resistance (R) of neurons of the three species tested was determined from the slope of current voltage relationships from -90 to -60 mV and is presented in Table 1. A two-way analysis of variance shows no significant difference in input R among species tested (F 2, 46 = 0.69 p>0.05) or between temperature-unresponsive and temperature-responsive neurons (see below for criteria, F 1, 46 = 0.51, p>0.05).

Figure 1

Characteristics of TG neurons. Currents resulting from increasing 10 mV voltage steps (A, C) and respective current-voltage relationships (B, D). The traces are representative of the two major types of currents found in TG neurons from both C. atrox and A. contortix.The asterisk (*) shows the inward current discussed in the text. B and D show the currentvoltage relationships at arrows numbered 1 and 2 for each set of current traces. E.Representative long (left) and short (right) action potentials in TG neurons, elicited by a 50 -100 pA pulse (bar).

Table 1

Characteristics of TG neurons

We did not observe cells showing spontaneous action potentials (APs) under current clamp. However, in 14 of 19 neurons (74%) from A. contortix, we were able to elicit action potentials (APs) under current clamp with a 50 -100 pA, 100 ms anodic current injection ( Figure 1E). We recorded two types of APs, which could be distinguished by having a short (mean = 5.4 ± 1.2 ms, SD, n=9) or long (mean = 17.1 ± 1.7 ms, SD, n= 5) spike duration.

Due to the low number of temperature-unresponsive neurons in this sample, we were unable 8 to determine if the distribution of AP type was nonrandom with respect to the temperature response characteristics of the neuron. Temperature responses were elicited by applying heated or chilled HBS as close as possible to the voltage clamped neuron using a polypropylene Pasteur pipette. A thermocouple with rapid temperature response (Physitemp, Clifton NJ) placed within 500 µm of the cell was used to estimate the thermal stimulus delivered to the cell. We determined heating and cooling parameters from 30 representative temperature records in which we sought to determine the current response of neurons to a temperature step. Representative temperature traces are presented in the top panels of Figure Temperature-responsive neurons were defined as having consistent, temporally correlated responses to stimulation and having an inward current of at least 2.0 pA/°C when heated to a minimum of 10°C from room temperature (~20°C). These currents also needed to reverse upon cooling. Using these criteria, we identified 75% (18 of 24) of the TG neurons in C. atrox and 74% (90 of 122) in A. contortix as temperature-responsive (Table 1, Figure 2A,B). Figure 2A-C shows representative inward current responses to application of warmed HBS to TG neurons from all species tested. We called this current I T . The current responses were rapid, temperature-dependent, and followed temperature changes closely. Figure 2D shows a representative response from a temperature unresponsive cell, showing no significant current change due to the application of heated HBS. In temperature-responsive cells, inward current in response to heat was accompanied by a decrease in membrane potential when 9 recorded under current clamp conditions ( Figure 2E). Heating and cooling of TG neurons did not result in the generation of action potentials. I T recorded in C. atrox and A. contortix were very similar ( Table 1). The magnitude of I T was 15.6 ± 11.7 pA/°C (SD, n= 13) in TG neurons from C. atrox and 11.2 ± 14.2 pA/°C (n=36) in TG neurons from A. contortix. The range was 3.9 to 45.4 pA/°C for C. atrox and 2.15 to 63.7 pA/°C for A. contortix. We calculated current density of I T based on the assumption that TG neurons were spherical, with a surface area determined by the formula 4 r 2 . The values for surface density are expressed as nA/°C/cm 2 and are presented in Table 1 for the species tested. Current densities showed high variability and were log 10 transformed for statistical comparisons (40). There was no statistical difference in I T current density in TG neurons of C. atrox vs. A. contortix. We pooled data from both snakes to examine the effect of cell size on current density. There was a weak but statistically significant negative correlation (40) between neuron surface area and the log 10 transformed current density value (r = 0.49, s r = 0.16, n = 31, t = 3.03, p<0.05), indicating that smaller cells tend to express more I T.

Figure 2

Temperature-sensitive currents in TG neurons. A-C. I T in representative TG neuron from all species tested (indicated above trace), showing temperature-dependent response (lower trace) to bath heating (temperature profile in upper trace) D. Current trace from A. contortix TG neuron that did not express I T . E. Under current clamp TG neurons expressing I T show a depolarization (lower trace) in response to heating, as shown in this representative trace from and A. contortix TG neuron.

We also examined heat-responsive currents in TG neurons from the common garter snake (T. sirtalis, Figure 2C, Table 1), a snake that does not have specialized heat sensing organs, but is in the same superfamily (Xenophidia) as the crotaline snakes. TG neurons were isolated and recorded from 3 different snakes. Temperature-activated currents resembling I T were seen in 3 of 20 TG neurons (15%) from T. sirtalis. It was possible to make a statistical comparison between the proportions of I T -expressing neurons in the pooled population of C. atrox and A. contortix TG neurons vs. those of T. sirtalis because the sample sizes were within conservative estimates necessary to maintain statistical power for comparisons of unequal samples (40). The crotalines had a significantly greater proportion of I T containing neurons in their TG than T. sirtalis (one-tailed test for significant of proportions, Z = 6.87, p<0.01). Additionally, TG neurons from C. atrox or A. contortix had a significantly higher I T current density than TG neurons from T. sirtalis (see Table 1 for statistical information).

Cooling cells from RT revealed that I T is activated within the range of temperatures that a pit viper would encounter in the wild (38) (Figure 3A). We examined current voltage relationships of cells that were first cooled to 12-15°C before heating to determine the temperature at which I T was inactive. Cells cooled to 12-15°C showed no further temperature-dependent changes in current, suggesting I T was inactive in this range ( Figure 3B). The current temperature relationship of I T from A. contortix is shown for 5 cells presented in Figure 3C. These cells were chosen because they had stable I T data over a large range of temperatures and current values were subtracted from the baseline that represented the lowest value over the record. The I T vs. T relation showed a nonlinear response, with little current increase at lower temperatures (10-15°C) transitioning to a more rapid rate of increase at just below 18°C. Regression lines for the two components were predicted by least squares method minimizing the trend in residuals (40) and the threshold was determined to be 17.8°C for the steeper current gain per unit temperature.

Figure 3

Characterization of I T . A. Cells expressing I T show an outward current (lower trace) upon cooling from room temperature. This cell had not been previously heated. B. Current-voltage plots of TG neurons at 12°C (open circles), 15°C (filled circles) and 35°C (filled triangles). C. Current-temperature relationships of I T from 5 cells. Each cell is represented by a different symbol. Dotted lines show predicted relationships and were obtained using least squares linear regressions.

The reversal potential (RP) of I T was determined for TG neurons from A. contortix by subtracting currents generated from increasing voltage steps at 15°C from those resulting from heating neurons to 26-35°C ( Figure 4A). We use CsCl electrodes to reduce the contribution of voltage activated K + channels, which may influence the value of the RP (1).

Figure 4

A. Current-voltage relationship of I T from a representative cell recorded with CsCl electrodes. Left insets show currents at 15°C and 35°C resulting from current steps depicted28 beneath them. The current for the plot was measured at 100 ms from the start of the voltage step. The current-voltage plot shows the difference current resulting from the subtraction of currents at 15° C from those at 35°C. B. I T recorded from a TG neuron held at +20 and -65 mV. The temperature trace is representative of the heating protocol applied at both holding potentials.

The RP of I T was -12.7 ± 9.6 mV (SD, n=9). Similarly, cells held at and above 0 mV showed a reversal of the heat-induced current ( Figure 4B). Consistent with the RP, ion substitution experiments indicated that heat may activate a monovalent cation channel with K + permeability. Substitution of NaCl with equimolar N-methyl D-glucamine resulted in a 49% (± 26% SD, n=4) decrease in the magnitude of the integrated I T current per °C ( Figure 5A). More detailed ion substitution experiments are summarized in Table 2. Voltage ramps 11 were used to examine RP of neurons in varying ionic conditions. The RP of I T in 10 mM NaCl shifted to -27.9 ± 4.0 mV (Table 1) from -12.7 mV. Interestingly, increasing CaCl 2 from 2 mM to 10 mM had no effect on the RP of I T (-24.8 ± 3.2 mV, SD, n=3) from cells in 10 mM NaCl. Similarly, a ten-fold reduction in Ca 2+ did not affect the RP in 120 mM NaCl.

Figure 5

A. Sodium dependence of I T . TG neurons were voltage clamped at -65 mV in HBS and I T was recorded in response to application of warmed HBS. The recording media was changed to one in which 120 NMDG was substituted for NaCl, and the cell was heated by application of the substituted media. B. Calcium permeability of the heat-activated current was monitored using fura-2 spectrophotometry and whole-cell voltage clamping simultaneously. This panel shows simultaneous fura-2, temperature and current data for a representative heat responsive TG neuron. TG neurons loaded with fura-2 show no heat induced change in the ratio (stimulation at 340nm/380 nm) of fluorescence emission at 510 nm. The neuron was depolarized (bar) by turning off the holding potential, and this resulted in a strong signal, indicating that cells were loaded and responsive to intracellular Ca 2+ level changes. Temperature and current data were not taken just prior to and after the positive control depolarization, but off-line monitoring of temperature show there was no

Table 2

Ion substitution and reversal potentials of I T

From these data, and estimated intracellular concentrations of Na + and K + of 4 and 140 mM, respectively, we determined that for I T the relative permeability P K /P Na 1.16 from the Goldman, Hodkin and Katz voltage equation (16) .

In order to verify the lack of Ca 2+ permeability in I T suggested by ion substitution experiments, we measured intracellular Ca 2+ using fura-2 in 43 voltage clamped TG neurons from A. contortix. I T -expressing TG neurons did not show an increase in intracellular Ca 2+ ( Figure 5B), further indicating that these heat-activated channels are impermeable to Ca 2+ .

From these data it can be assumed that I T is primarily a monovalent cation conductance.

Previous research has established the pharmacological identity of several temperaturesensitive ion channels (7; 20; 23; 25). We exposed I T -expressing TG neurons from A. contortix to 10 µM capsaicin (in 0.001% [vol/vol] ethanol) to examine if a pharmacologically active TRPV1 homologue is present in these cells (7). Cells did not respond to capsaicin, nor was the magnitude of response to heat stimulus changed (n=4). Similarly, amiloride, which blocks temperature sensitive ENaC channels (2) had no effect on the magnitude of I T in TG neurons (n=5).

Discussion

We have recorded a novel temperature activated current in neurons that supply the pit organ of A. contortix and C. atrox. The current was novel in that it had the lowest threshold 12 of activity of any known heat-activated conductance, and had ion permeability unlike that of any characterized temperature sensitive channel. I T was a monovalent cation conductance that was active at ambient temperatures and increased in response to heating. The effective temperature range of I T was consistent with having a role in sensitive thermal detection of prey and other environmental stimuli. I T was found in a large proportion of the neurons in the TG of C. atrox and A. contortix. Temperature activated neurons were also found in the TG of snakes lacking the pit organ (T. sirtalis), but at a much lower frequency and with lower current density. This suggests that crotaline snakes may have adapted a general warm thermosensor to a specialized function as neuronal signal transducers in a highly sensitive thermoreceptive organ. The ion dependence of I T was found to be different from other known temperature-sensitive channels (7) (25) (39) (13), further illustrating the unique nature of these temperature gated conductances. The presence of a large number of neurons expressing temperature-gated currents in these cells further strengthens the hypothesis that heat detection in the pit organ is not due to photonic IR detection but is a result of highlysensitive heat detection(3) (9) (26).

The use of in vitro preparations to study thermoreceptors

A generator potential for pit thermoreceptors has been recorded from crotaline snakes including Agkistrodon (34). Terashima and colleagues (34) recorded both voltage spikes and slow potentials using extracellular electrodes inserted just below the pit membrane into the terminal nerve mass. These potentials were proportional to stimulus intensity and duration, but drifted in direction due to the instability of the preparation. We chose to use a voltage clamp to obtain more detailed biophysical information on temperature-activated currents, but we were unable to clamp the terminal nerve mass, which probably contains the highest density of cellular structures that cause temperature dependent currents (i.e., heat-sensitive 13 ion channels). Recordings were taken from cell bodies because of their ease of preparation and because temperature responses of neurons have been shown to be accurately replicated in cultured trigeminal and dorsal root ganglion neuron cell bodies (23) (8) (31) (27). This is because heat sensitive ion channels are expressed and active at the cell body in cultured neurons.

Neurons were subjected to a wide range of temperatures during these recording procedures, and some of these neurons showed no response to temperature change. This is consistent with findings in ganglion cultures from mammalian sensory neurons where there was a clear distinction between temperature-sensitive and insensitive cells, and suggests that there was not artifactual temperature induced activation of resting leak or voltage activated currents with thermal stimulus. Cells expressing I T showed characteristic nonlinear relationship between temperature and currents that indicated a threshold just below 18°C.

This type of current-temperature relationship has been seen for temperature-activated currents (36) and for isolated temperature-gated ion channels from mammals (7) (25) (39). The transition from slow to rapid current change as a function of temperature is thought to happen at the temperature in which there is a rapid and reversible change in the ion channel, possibly associated with temperature-dependent gating of the channel (36).

We could not record APs in response to heating of TG neurons, although it is clear from studies on intact preparations that heat sensation in the pit organ is transmitted via APs.

It is possible that the rate of heat stimulus we presented to these cells was too slow to generate APs. Temperature changes in the low thermal mass pit organ should be nearly instantaneous, even if they are small. With slow changes, however, ion channels responsible for stimulating regenerative changes in membrane potential might be inactivated before the AP could take place, or voltage-activated channels responsible for terminating APs might be activated before the I T could significantly change the membrane potential. Two previous 14 studies (3) (9) suggest that pit afferents respond to the rate of temperature change as well as the stimulus intensity. However, changes in AP frequency could be distinguished even with very long temperature rise times (3). It is also possible that the density of I T in dissociated TG neurons was not sufficient to give rise AP-producing depolarization of membrane potential, or that accessory channels involved in AP production in these cells were not adequately expressed. This could be the result of low or altered expression of putative ion channel(s) responsible for I T lowered or altered AP-associated ion channels in dissociated cell culture. Finally, it is possible that the role of I T is not to produce fast enough or robust enough depolarization for AP activation, but instead modulate ongoing changes in membrane potential. This is supported by studies from pit viper trigeminal nerve (3) (9) and trigeminal ganglion neurons (33) showing that these elements are spontaneously active and that temperature change is signaled by alterations in ongoing spike activity. We did not see spontaneous APs in our cells, possibly as a result of dissociated culture, but we speculate that one mechanism by which I T might signal heat is to produce a slow depolarization that increases spike frequency, as has been described for other sensory systems (5; 10).

Neurons expressing a temperature-sensitive current similar to I T were also found in the TG from the common T. sirtalis, a snake that does not have a specialized organ for sensitive heat detection, but at a significantly lower frequency and with lower current density.

It is not surprising that we found thermosensitive cells in the TG from this snake, as this ganglion supplies more general temperature sensors to the face (4). It is also possible that some of the temperature-responsive neurons we recorded from the C. atrox and A. contortix TG are actually cutaneous warm receptors that did not send projections to the pit organ, but to other parts of the face. If this is the case then the frequency of cells expressing I T we observed in the TG is not an accurate estimate of the frequency of warm-receptive neurons sending projections to the pit organ but instead to the entire facial region innervated by the 15 trigeminal nerve. The frequency of TG cells showing various temperature-activated currents that actually send afferents to the pit organ will have to be determined using retrograde labeling from the pit organ in conjunction with physiological recordings in the TG. However, a comparison can be made with the frequency of warm-receptive neurons found in the TG of the T. sirtalis, as both estimates were made using the same sampling procedure. The fact that I T in T. sirtalis was qualitatively similar to that of pit vipers has interesting implications in the evolution of the pit organ and its thermosensitivity. The pit afferents of crotaline snakes have been shown to have a higher number of warm-sensitive fibers than any known animal (3). Our data on proportion of heat sensitive neurons in pit vipers vs. T. sirtalis, are consistent with these findings. Additionally, our data also showed that the current density of I T is much higher in pit vipers than in T. sirtalis. It is possible that pit viper thermosensors evolved as specializations of general cutaneous warm receptors like those that would be found in all snakes. The evolution of a specialized pit organ was accompanied by an expansion of numbers of warm-sensitive neurons supplying the area, resulting in very high thermal sensitivity of the pit organ. Additionally, individual neurons may be more sensitive to temperature, and this would result in these neurons having greater electrophysiological responses to temperature change further enhancing sensitivity of an integrated thermoreceptor like the pit.

I T correlates with behavioral data on snake thermosensing

The concordance in temperature response characteristics between behavioral or nerve recordings and cultured neuronal preparations has often been used to establish a causal link between temperature sensation and temperature-activated currents (8). Our data from snake TG neurons correlated to features discovered from behavioral, whole nerve, or single fiber recordings from pit vipers. Most notably the temperature threshold for I T in pit viper TG was 16 in good agreement with recordings from the pit afferent nerves that showed loss of spontaneous spiking activity at 10-15°C (9). Above this threshold there was a constitutive inward current, which may be related to "background" activity of nerves seen above 18°C.

The temperature tracking characteristics of I T may also be related to the non-adapting spike discharge seen in pit afferents when the pit is heated. The detailed investigations of de Cock Buning et al. (1981) demonstrate that rapid adaptation of spiking frequency is a property of rapid cooling in the low thermal mass pit organ and not an adaptation at the nerve itself.

Adding water to the pit increased its thermal mass and resulted in a non-adapting discharge of pit afferents. This suggests that the neural response is constant throughout the thermal stimulus and is in agreement with the behavior of I T . The temperature tracking characteristics of I T were also consistent with the slow, continuous temperature dependent changes seen in the extracellular generator potentials recorded at the pit membrane (34).

Behavioral data also shows that pit afferents stop firing above 37°C (9) and this does not correlate well with the behavior of I T , which showed responses even at and above 40°C.

This raises the possibility that there may be other temperature-sensitive currents that mediate thermosensitivity at the pit organ. It is possible that there may be a high-threshold temperature-sensitive current that responds at temperatures above 37°C and may be involved with the cessation of the pit response at these temperatures. We already have preliminary data that suggest that the TG may contain neurons that have a transiently-activated coolingsensitive current that may contribute to the sensitivity described in snake thermosensing (30).