Journal of Neurophysiology

Effects of Temperature on Calcium Transients and Ca2+-Dependent Afterhyperpolarizations in Neocortical Pyramidal Neurons

J. C. F. Lee, J. C. Callaway, R. C. Foehring


In neocortical pyramidal neurons, the medium (mAHP) and slow AHP (sAHP) have different relationships with intracellular [Ca2+]. To further explore these differences, we varied bath temperature and compared passive and active membrane properties and Ca2+ transients in response to a single action potential (AP) or trains of APs. We tested whether Ca2+-dependent events are more temperature sensitive than voltage-dependent ones, the slow rise time of the sAHP is limited by diffusion, and temperature sensitivity differs between the mAHP and sAHP. The onset and decay kinetics of the sAHP were very temperature sensitive (more so than diffusion). We found that the decay time course of Ca2+ transients was also very temperature sensitive. In contrast, the mAHP (amplitude, time to peak, and exponential decay) and sAHP peak amplitude were moderately sensitive to temperature. The amplitudes of intracellular Ca2+ transients evoked either by a single spike or a train of spikes showed modest temperature sensitivities. Pyramidal neuron input resistance was increased by cooling. With the exception of threshold, which remained unchanged between 22 and 35°C, action potential parameters (amplitude, half-width, maximum rates of rise and fall) were modestly affected by temperature. Collectively, these data suggest that temperature sensitivity was higher for the Ca2+-dependent sAHP than for voltage-dependent AP parameters or for the mAHP, diffusion of Ca2+ over distance cannot explain the slow rise of the sAHP in these cells, and the kinetics of the sAHP and mAHP are affected differently by temperature.


Brain temperature is an important variable in determining the damage done to cortical neurons after ischemic insults (e.g., Ren et al. 2004). Reversible cooling is also used to inactivate cortical regions during in vivo experiments (e.g., Ferster et al. 1996; Michalski et al. 1993). Furthermore, in vitro experiments are conducted at different temperatures, and one would like to be able to compare experiments between laboratories and to relate the in vitro findings to physiological conditions in vivo. For all of these reasons, it is important to understand the temperature sensitivity of cortical neurons.

Relatively few studies have examined the temperature dependence of passive and active electrical membrane properties in vertebrate neurons. Thompson et al. (1985) provided a detailed look at temperature sensitivity in guinea pig CA1 pyramidal neurons. They determined that cooling from physiological temperatures (37°C) to between 33 and 27°C resulted in increased input resistance, larger and wider action potentials, increased afterhyperpolarization (AHP) amplitude, and increased spike-frequency adaptation (see also Shen and Schwartzkroin 1988). The last two parameters are Ca2+ dependent and were particularly temperature sensitive. Similar effects on input resistance and action potentials were reported for spinal motoneurons (Klee et al. 1974) and hypothalamic neurons (Griffin and Boulant 1995). Passive decay of voltage transients was also found to be greatly prolonged at lower temperatures in layer II/III neocortical pyramidal cells (Trevlyan and Jack 2002).

The effects of temperature are likely to differ between cell types because the coupling of Ca2+ channels to AHP channels varies between cell types (Bayliss et al. 1995; Bowden et al. 2001; Hallworth et al. 2003; Martinez-Pinna et al. 2000; Moyer and Disterhoft 1994; Pineda et al. 1998), and there are differences between cell types in intrinsic Ca2+ buffering [including layer V pyramidal cells (Helmchen et al. 1996) versus layer II/III pyramidal cells (Kaiser et al. 2001)].

More recently, the effects of temperature have been examined on neocortical pyramidal neurons (Volgushev et al. 2000a, b). Cooling from 35°C to room temperature (RT, 20–25°C) led to depolarization, increased input resistance, larger and broader APs, and increased excitability. Further cooling (to <10°C) resulted in a depolarization block of AP production (Volgushev et al. 2000b). The influence of temperature on synaptic transmission was complicated (Volgushev et al. 2000a). While cooling to <20°C led to diminished excitatory postsynaptic potentials (EPSPs), cooling to 20°C from 35°C could lead to increased, decreased or no change in EPSPs. Paired-pulse facilitation was reduced at lower temperatures, indicating altered synaptic release dynamics compared with physiological temperatures (Volgushev et al. 2000a). Similar conclusions about the temperature-dependence of synaptic transmission were reached by Hardingham and Larkman (1998) in neocortex and Aihara et al. (2001) and Fujii et al. (2002) in CA1. These studies indicate that caution is necessary when inferring physiological consequences from experiments at lower temperatures. Rosen and Morris (1994) showed alteration of EPSPs and inhibitory postsynaptic potentials (IPSPs) by temperature in layer II–III rat frontoparietal slices, as well as an altered response of these cells to anoxia. Specifically, cooling reduced the anoxic depolarization and increased input resistance.

One would expect that temperature-related changes in spike height and width would alter Ca2+ entry in cortical neurons (cf. Stewart and Foehring 2001) and thus alter Ca2+ transients in response to APs. In vitro Ca2+ imaging experiments are performed at different temperatures in different labs; however, very little is known concerning the effects of temperature on AP-induced Ca2+ transients in neurons. Borst and Sakmann (1998) reported that in the Calyx of Held, changes in [Ca2+]i in response to a single AP had lower peak, prolonged decay, and larger net charge movement at 24 versus 36°C. Markram et al. (1995) reported that the rise and decay times for Ca2+ transients induced by single back-propagated APs in layer V pyramidal neurons were longer at RT versus 34°C. The temperature dependence of [Ca2+]i in response to trains of action potentials has not previously been addressed.

One consequence of elevated [Ca2+]i in neurons is activation of Ca2+-dependent K+ currents that underlie afterhyperpolarizations (AHPs) and spike-frequency adaptation. In the present study, we used temperature sensitivity to probe the relationship between intracellular Ca2+ concentration ([Ca2+]i) and Ca2+-dependent AHPs. In layer II/III pyramidal neurons from somatosensory cortex, action potentials produce three different AHPs, depending on the number and frequency of spikes (Abel et al. 2004; Lorenzon and Foehring 1992; Pineda et al. 1998). Single action potentials are followed by a fast AHP (fAHP) and a medium AHP (mAHP). The mAHP decays with a time constant of 100–200 ms. With multiple action potentials at higher frequency, a slowly decaying (τ ∼1–2 s) sAHP is evoked. Both the mAHP and sAHP are voltage independent and Ca2+ dependent. Ca2+ entry through specific types of Ca2+ channels activated during the action potential(s) is responsible for evoking the mAHP (P-type Ca2+ channels) and the sAHP (P/Q, N-type Ca2+ channels) (Pineda et al. 1998). The mAHP appears to be due to small conductance Ca2+-dependent K+ channels (sK) (Sah and Faber 2002; Vogalis et al. 2003). The channels underlying the sAHP are unknown. We used changes in temperature to gain insight into the nature of these sAHP channels.

Activation of the sAHP requires multiple (3–5) spikes, and there is a long latency to peak. One potential reason for the slow rise time would be slow diffusion of Ca2+ from sites of entry to the sAHP channels (Lancaster and Zucker 1994; Sah and Faber 2002). Alternative explanations include slow intrinsic channel kinetics (Sah and Faber 2002) or involvement of an intermediate between Ca2+ and the channels (Abel et al. 2004; Sah and Faber 2002; Schwindt et al. 1992b). We examined the temperature sensitivity of the sAHP rise time in neocortical pyramidal neurons to test whether diffusion could explain the sAHP's slow rise. We found this parameter to be highly temperature-sensitive, ruling out diffusion as a primary limiting factor (cf. Sah and McLachlan 1992 for vagal neurons).

We compared input resistance, action potentials, and Ca2+ transients in response to a single AP or trains of APs to test whether Ca-dependent events are more temperature sensitive than voltage-dependent ones and whether the temperature sensitivity is different between the mAHP and sAHP.


The brain was removed from metofane-anesthetized Sprague-Dawley rats (P13-19) and then sliced into 300-μm-thick coronal sections using a vibrating tissue slicer (World Precision Instruments, Sarasota, FL). The tissue was sliced in an ice-cold, high-sucrose solution (pH = 7.3–7.4, 300 mosM/l) containing (in mM) 250 sucrose, 2.5 KCl, 1 Na3PO4, 11 glucose, 4 MgSO4, 0.1 CaCl2, and 15 HEPES. The primary somatosensory and primary motor cortices were dissected from the slices and then transferred to a mesh surface in a chamber containing artificial cerebrospinal fluid (ACSF) at room temperature. The ACSF contained (in mM) 125 NaCl, 3 KCl, 2 CaCl2, 2 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 20 glucose (pH = 7.4, 310 mosM) and was bubbled with a 95% O2-5% CO2 (carbogen) mixture. For recording, slices were placed in a recording chamber on the stage of an Olympus BX50WI upright microscope. Slices were bathed in carbogenated ACSF delivered at 2 ml/min and heated with an in-line heater (Warner Instruments, Hamden, CT) to 31–35°C. Apamin was prepared as a concentrated stock in d,d, H2O and then thawed and added to the ACSF just before recording.

Bath temperature was measured with a thermistor (Warner Instruments: positioned adjacent to the slice in the bath) attached to an analog thermometer (Yellow Springs Instruments, Yellow Springs, OH). Temperature was changed by turning the heater off or on and waiting until the bath temperature reached steady state (stable for >2 min.) at room temperature (RT: 23 ± 1°C) or 33 ± 2°C. The temperature sensitivities of measured parameters were expressed as Q10 values (the proportionate change for a 10°C change in temperature). Q10s were calculated as Mathwhere t2 is 33 ± 2°C and t1 is 23 ± 1°C and X2 and X1 are the corresponding parameters at those temperatures. Because we are comparing room temperature (22°C) versus 33°C, a Q10 ratio close to 1 indicates little or no temperature dependence, whereas a Q10 ratio < 1 suggests a decrease with increasing temperatures. A Q10 ratio > 1 suggests an increase with increasing temperatures.

Pyramidal neurons in layers II and III were visualized with infrared/differential interference contrast (IR/DIC) video-microscopy (Dodt and Zieglgansberger 1990; Stuart et al. 1993) using a ×40 (0.8 NA) Olympus water-immersion objective. Simultaneous whole cell patch clamp and Ca2+ fluorescence imaging records were acquired using an Axoclamp 2A (Axon Instruments; current clamp) or an Axopatch 200B (Axon Instruments; voltage clamp) amplifier in combination with a cooled CCD camera (Sensicam: PCO, Germany). We recorded with borosilicate electrodes (of resistance 4–8 MΩ) produced with a horizontal electrode puller (Sutter Instruments, Novato, CA) and filled with a solution containing (in mM) 130.5 KMeSO4, 10 KCl, 7.5 NaCl, 2 MgCl2, 10 HEPES, 2 adenosine 5′-triphosphate (ATP), and 0.2 guanosine 5′-triphosphate (GTP). Unless otherwise specified, 100 μM fura-2 (Molecular Probes; pentapotassium salt) was added to the intracellular solution. Data were only collected from cells forming a ≥1 GΩ seal.

Due to the time required for fura-2 to equilibrate within various compartments of the cell and subsequent run-down of the sAHP during prolonged recordings, we waited ∼5 min for stabilization of [fura-2] in the soma. At this time, [fura-2] was not steady state in the dendrites so the bulk of our analyses were restricted to somatic Ca2+ (see Abel et al. 2004 for differences between soma and dendritic measurements of [Ca2+]i). The somatic Ca2+ transients were very stable over long recordings (data not shown) and in some experiments, 10 mM myo-inositol was added to the pipette solution to reduce sAHP rundown (cf. Fig. 3B).

Optical data were obtained by exciting the dye (fura-2) at a wavelength of 380 ± 10 nm and measuring fluorescence changes at an emission wavelength of 520 ± 40 nm (filters from Chroma Technology, Brattleboro, Vt.). Electrical and optical data were synchronously acquired on a single Windows platform PC running software written by Dr. J. C. Callaway, based on software developed by Lasser-Ross et al. (1991). Electrical records were digitized with 16 bit resolution at 10 kHz and corrected for the liquid junction potential (10 mV).

The relative change in fura-2 fluorescence (ΔF/F) is closely proportional to the calcium concentration for changes less than ∼50% ΔF/F (Abel et al. 2004; Lev-Ram et al. 1992). We used a calcium calibration buffer kit (Molecular Probes) to prepare solutions of known ratios of K2-EGTA to Ca-EGTA in the internal recording solution, for which we could calculate [Ca2+]free. This allowed us to determine the KD for fura-2 in vitro to be 222 nM. We acquired pairs (at excitation wavelengths of 340 ± 10 and 380 ± 10 nm) of fluorescence intensities from solutions containing [Ca2+]free ranging from 0 to 400 nM. The resulting calibration curve was used to estimate resting calcium in our cells (from ratiometric measurements taken at a holding potential of –70 ± 5 mV).

In our experiments, fluorescence values (at 380 nm) were converted to Ca2+ concentrations using a modification of the method described by Lev-Ram et al. (1992). The following equation was used Math(1)(where [Ca2+]1 is the resting Ca2+ level) to estimate [Ca2+]i from %ΔF/F. This formula was derived by Wilson and Callaway (2000) and employed here because it did not require a measurement of the maximal possible fluorescence change, which requires loading the cell maximally with calcium. Sb380/Sf380 is the ratio of bound to free fura-2 fluorescence (see Grynkiewicz et al. 1985), which we determined in our calibration to be ∼10. Corrections for photo-bleaching were made by subtracting the Ca2+ signal from an equal-length control sweep containing no stimulus. Tissue auto-fluorescence was accounted for subtracting the fluorescence of a nonfura-loaded area of tissue near the cell.

Unless otherwise stated, data are presented as means ± SD. Further analysis was conducted using Igor Pro (Wavemetrics, Lake Oswego, OR) and Kaleidagraph (Synergy Software, Reading, PA). Curve fits used the Levenberg-Marquardt algorithm to determine the best fit by minimizing χ2 values. Additional components were reported for curve fits if the additional component comprised ≥10% of the amplitude.

Technical considerations

If fura-2 was itself temperature-sensitive, this would be a potential complication for interpreting changes in [Ca2+]i with changes in temperature. Our calibrations for KD were performed at RT. The KD for fura-2 binding to Ca2+ is only moderately temperature sensitive. We estimate the Q10 as ∼0.9 (KD increases with cooling) from the published data of Shuttleworth and Thompson (1991) or Larsson et al. 1999). Temperature-related pH changes are also unlikely to alter our conclusions either as Grynkiewicz et al. (1985) reported very little sensitivity of fura-2 to pH changes in the physiological range (see also Lattanzio and Bartschadt 1990). For our carbogen-bubbled ACSF, we measured a pH change of <0.1 over the 22–35°C range. Fura-2 also shows a very slight reduction in fluorescence ratio (380/340 nm) with cooling and some temperature-related alterations in photophysics (Oliver et al. 2000). These photophysical changes were not due to pH or viscocity and appear due to prolonged fluorescence lifetimes at lower temperatures (Oliver et al. 2000). Larsson et al. (1999) reported no change in absorption maxima or isobestic point with temperature (5–37°C).


Recordings were obtained from cells visually identified under IR/DIC and fluorescence imaging as pyramidal cells in layer II or III. All of the cells fired repetitively in a regular spiking pattern (McCormick et al. 1985). Cells with a resting potential negative to –60 mV and APs that overshot 0 mV were chosen for study.

To allow comparison of cells under steady-state conditions at two temperatures (see introduction), data were obtained at one temperature for each cell in most of our experiments.

Table 1 summarizes values for resting membrane potential (RMP), input resistance (RN), and action potential parameters at 33 ± 2°C (hereafter refered to as 33°C) and room temperature (RT: 23 ± 1°C). Depolarization with cooling is a consistent finding in most studies comparing electrophysiological properties at different temperatures (Shen and Schwartzkroin 1988; Thompson et al. 1985; Volgushev et al. 2000a, b; but see Griffin and Boulant 1995). In most cases, we did not directly test the effects of cooling on membrane potential; however, RMP was not significantly different between temperatures, when measured at the initial temperature at the beginning of the experiment (Fig. 1B; Table 1). Thereafter, DC current was used to maintain RMP at approximately −60 mV for all experiments. To determine input resistance, current was injected (500 ms) to hyperpolarize the membrane by 10–20 mV from –60 mV (RN = V/I). Input resistance was highly temperature sensitive and was significantly higher at RT than at 33°C (Fig. 1A, Table 1).

FIG. 1.

Temperature dependence of input resistance and action potential (AP). A: input resistance was increased at lower temperatures. Voltage reponse to 1-s long, −25-pA current injection. Both traces are from the same cell. Inset: pyramidal neurons in layers II and III were targeted. *, recording electrode. Scale bar = 20 μm. B: action potential amplitude and width varied with temperature. Left: single AP from same cell at 26 and 33°C. Box: action potentials from a different cell, expanded scale. Inset: Scatter plot shows data for AP amplitude for 10 cells for which action potentials were obtained at both 26 and 35°C. AP amplitude was greater at 26°C for 9 of 10 cells. C: APs in a different cell are shown at higher gain to illustrate the medium afterhyperpolarization (mAHP). Peak APs were digitally truncated. Note slower rise and decay of mAHP at 26 vs. 30°C.

View this table:

Input resistance (RN) and action potential (AP) parameters

In 10 cells we obtained data for RMP, input resistance, and single spike parameters at both 35°C and RT. In these cells, measurements were made after >2 min at each stable temperature. In these cells, input resistance was reversibly and significantly increased at RT, and we observed no consistent difference in RMP with temperature.

We evoked a single AP with a 5-ms, just-suprathreshold depolarizing current injection at either RT or 33°C. Compared with 33°C, at RT the spike was significantly broader (1/2 width, base width) and the rates of spike rise and repolarization (dV/dt) were significantly slower (Table 1). Spike threshold and amplitude were not significantly different at the two temperatures (Fig. 1B, Table 1), although when measured at both temperatures in the same cell, AP amplitude was larger at RT in 9 of 10 cells tested (significant difference; see Fig. 1B, inset). In these 10 cells, significant differences were also observed for AP half-width and base width and for the rates of spike rise and repolarization but not AP threshold (data not shown). Spike threshold did not vary significantly with temperature in these 10 cells. All of these effects were at least partially reversible. With the exception of spike threshold (unaffected by temperature), all spike parameters show moderate temperature dependence, as indicated by their Q10 ratios (Table 1).

Ca2+ transients from a single AP

The temperature-dependent alterations in the single AP parameters, especially increased spike duration at lower temperatures, would be expected to result in greater Ca2+ entry with each AP (Stewart and Foehring 2001). We elicited single APs (5 ms current injection) and measured fluorescent changes in fura-2 to estimate changes in [Ca2+]i.

In response to a single spike, the amplitude of the somatic Ca2+ transient did not change significantly with increasing temperatures (Fig. 2, Table 2). The Ca2+ transient decay time constants (τdecay) were shorter at 33°C. At RT, the longer decay time for the Ca2+ transient results in intracellular Ca2+ being elevated for a longer time. The time integral of [Ca2+]i tended to be larger at RT, but this difference was not statistically significant (P < 0.22). The peak amplitude of the Ca2+ transient was moderately temperature sensitive, but the decay time was very temperature sensitive (Table 2). In seven cells, we obtained fura-2 data for the soma and proximal apical dendrites (25–50 μm from the soma). The amplitudes were consistently larger and the τdecay consistently shorter for apical dendritic versus somatic transients (Table 2). The Q10s were not significantly different for somatic versus dendritic transients (Table 2).

FIG. 2.

Temperature-dependence of the Ca2+ transient and the mAHP. Due to a single AP. Top: Note little difference in amplitude and slower decay of [Ca2+]i at the lower temperature. Bottom: Same cell as upper traces. The mAHP was relatively insensitive to temperature. The action potential was truncated to emphasize the AHP. Right: Fura-2 fluorescent image of the cell from which these data were obtained. Box indicates where in soma changes in fluorescence were measured. Scale bar =10 μm.

View this table:

Temperature effects on [Ca2+]i and the mAHP in response to a single AP

Temperature dependence of mAHP

We examined the consequences of the broad spikes and prolonged elevations in [Ca2+]i at RT, for Ca2+-dependent events in the cell. A Ca2+-dependent mAHP is evoked after a single action potential (Figs. 1C and 2). Peak mAHP amplitude did not differ significantly between the two temperatures. Conversely, the τdecay of the mAHP decreased significantly with temperature. The mAHP is only moderately sensitive to changes in temperature, however, with Q10 ratios for mAHP amplitude and τdecay ∼ 0.8–0.9 (Table 2).

In a few cells, ImAHP was elicited by a 50-ms voltage step from −60 to +10 mV (data not shown). We do not have spatial control of voltage with this protocol; however, the AHP channels are not voltage dependent, and we have shown that tail currents elicited in this manner reverse at ∼EK, suggesting adequate control for these small, slow currents (Abel et al. 2004). The time to peak (TTP) and τdecay were significantly longer at RT versus 33°C (Table 2). No significant difference was observed for ImAHP amplitude and Q10s were modest.

Ca2+ transients from spike trains

In response to multiple spikes, the individual Ca2+ transients summate to a plateau (Abel et al. 2004; Lasser-Ross et al. 1997). Peak [Ca2+]i corresponds to the end of the spike train. Subsequently, the Ca2+ transient decays back to the resting Ca2+ levels (Abel et al. 2004; Lasser-Ross et al. 1997). Because the plateau Ca2+ level is dependent on the rates of Ca2+ entry and Ca2+ removal, including extrusion mechanisms and Ca2+ buffering, temperature could affect the summated Ca2+ transient after a train of spikes differently than the transient after a single AP. We elicited Ca2+ signals with 10 APs (5-ms suprathreshold current injections) at 50 Hz. The peak amplitude of [Ca2+]i occurred immediately after the final spike of the train and did not differ significantly with temperature (Table 3). The Q10 for peak amplitude was also modest (Table 3).

View this table:

Temperature effects on [Ca2+]i and the AHP in response to trains of APs

In contrast to peak [Ca2+]i, the decay of [Ca2+]i was highly sensitive to temperature. The decay of [Ca2+]i showed considerable variability between cells. At 33°C, the decay of [Ca2+]i in most cells (17 of 28: 61%) was well fit by a single exponential (τ = 452 ± 97 ms). In the remaining 11 cells, the decay was better fit as the sum of two exponentials [τ1 = 389 ± 159 ms (45 ± 13% of amplitude); τ2 = 772 ± 293 ms (55 ± 13% of amplitude)]. At RT, most cells (10 of 13 = 76%) required two exponentials to fit the decay of [Ca2+]i [τ1 = 1128 ± 825 ms (45 ± 10% of amplitude); τ2 = 2429 ± 1423 ms (55 ± 10% of amplitude)]. The remaining three cells were well fit by a single exponential with τ = 929 ± 293 ms. The τdecay for [Ca2+]i was very temperature sensitive (Table 3). Total charge entry, estimated as the time integral of the current, was significantly greater at RT (Table 3).

Temperature dependence of sAHP

In pyramidal neurons, the sAHPs are Ca2+ dependent (Abel et al. 2004; Lorenzon and Foehring 1993; Madison and Nicoll 1984; Pineda et al. 1998; Schwindt et al. 1988a). We recently found that the relationship between IsAHP and bulk [Ca2+]i was sigmoidal (IC50 ∼200 nM, Hill coefficient ∼ 4.5) (Abel et al. 2004). In combination with the requirement for multiple APs to elicit the sAHP, these data suggest that even though the sAHP is not due to sK channels (Bond et al. 2004; Villalobos et al. 2004), the Ca2+ sensor for activation of the sAHP has properties similar to the Ca2+ sensor of sK channels (e.g., calmodulin) and that the sensor responds to a pool of Ca2+ proportionate to the bulk cytosolic [Ca2+]i (Abel et al. 2004).

In current-clamp experiments, we used a standard spike train of 10 spikes at 50 Hz, and the sAHP was evoked along with the mAHP. At a single temperature, measurements of the sAHP and spike train-induced [Ca2+]i were stable for >30 min (Fig. 3B). For the sAHP, peak amplitude was larger (Fig. 3; Table 3), TTP was significantly longer (Fig. 3; Table 3), and τdecay was significantly longer at RT (n = 13 cells) than at 33°C (n = 28 cells: Table 3). The peak sAHP amplitude was moderately temperature sensitive, but the decay of the sAHP was very temperature sensitive (Table 3).

FIG. 3.

Temperature-sensitivity of [Ca2+]i and AHPs after a train of APs. A: Ca2+ transients and AHPs were elicited with a train of 5 ms current injections at 50 Hz, which elicited 10 APs. All data from the same cell. Top: Note small effect of lowering temperature on amplitude of the change in [Ca2+]i but marked slowing of [Ca2+]i decay. Bottom: Note increased amplitude, slower onset and slower decay of the sAHP at the lower temperature. Dotted line = 500 ms after last spike, when sAHP amplitude was measured. Box: Fura-2 fluorescent image of cell from which these data were obtained. Scale bar =10 μm. B: At a single temperature (35°C), the Ca2+ transient and sAHP are stable over time. The internal recording solution contained 10 mM myoinositol to reduce run-down of the sAHP. The sAHP was elicited with a train of ten 5 ms current injections at 50 Hz, which elicited 10 APs. This stimulus was repeated at 5 min intervals. We simultaneously recorded the sAHP and the change in fura-2 fluorescence. The data were normalized by the values for time 0 and plotted as a function of time (diamonds = % dF/F; squares = sAHP amplitude). Note that percent dF/F was essentially unchanged over the 55 min recording and that the sAHP was very stable for >30 min. Similar data were obtained from four cells. Scale bars are the same for data from 0 min versus 55 min.

We also examined IsAHP in voltage clamp (Fig. 4). We used a step to +10 mV (500 ms) from a holding potential of –60 mV to elicit Ca2+ entry. See preceding text for discussion of spatial clamp of voltage (Abel et al. 2004). The tail currents decay with three exponentials, with the slowest corresponding to IsAHP (Abel et al. 2004). The decay of this slow component was significantly slower and had a significantly longer time to peak at RT than at 33°C (Fig. 3; Table 3). The TTP was particularly temperature-sensitive (Table 3). In a few cells (n = 7), IsAHP was isolated from ImAHP by application of 50–100 nM of the selective sK blocker apamin (Fig. 4) (Abel et al. 2004) to allow examination of the τ for the onset of IsAHP. In these cells, the exponential rise to the peak IsAHPonset) was highly temperature sensitive and significantly slower at RT than at 33°C (Fig. 4, Table 3).

FIG. 4.

Temperature-dependence of IsAHP. Currents were elicited by a 500 ms step to +10 mV from −60 mV. Tail currents were elicited on return to −60 mV. A: 50 nM apamin blocked a component of the current with intermediate decay kinetics. This revealed the slow onset kinetics of the slow component (IsAHP). Inset: Fura-2 fluorescent image of the cell from which these data were obtained. Scale bar =10 μm. B: At lower temperatures, IsAHP onset and decay were much slower.


To evaluate the temperature dependence of Ca2+-dependent mechanisms regulating pyramidal cell excitability, we compared passive and active membrane electrical responses and Ca2+ transients in response to a single AP or trains of APs to test whether Ca2+-dependent events are more temperature-sensitive than voltage-dependent ones, whether the slow rise time of the sAHP is limited by diffusion, and whether temperature sensitivity is different between the mAHP and sAHP.

Our principal findings were the temperature sensitivity of the time course of the rising phase of the sAHP is much higher than diffusion (or voltage-dependent spike parameters). This finding rules out diffusion of Ca2+ as the limiting factor in the slow rise of the sAHP. Second, the decay kinetics of the sAHP were also highly temperature sensitive, consistent with an intermediate messenger between Ca2+ and the sAHP channels. Third, the kinetics of decay of [Ca2+]i following trains of APs was also highly temperature sensitive. And last, in contrast to the sAHP kinetics, the mAHP was much less sensitive to temperature, suggesting different mechanisms linking Ca2+ to activation of mAHP (sK) and sAHP channels.

sAHP rise phase kinetics

At physiological temperatures, the sAHP (and the underlying IsAHP) reaches its peak slowly (τ = 122 ms for IsAHP) and decays much slower than the mAHP (τ ∼ 1–2 s for both the sAHP and IsAHP) (Lorenzon and Foehring 1992; Pineda et al. 1998; Schwindt et al. 1988a, b). One potential reason for the slow rise time of the sAHP would be slow diffusion of Ca2+ from sites of entry to the sAHP channels (Sah and Faber 2002; Schwindt et al. 1992a). Alternative explanations include slow intrinsic channel kinetics (Sah and Faber 2002) or involvement of an intermediate between Ca2+ and the channels (Abel et al. 2004; Sah and Faber 2002; Schwindt et al. 1992a; Vogalis et al. 2003). The temperature dependence of aqueous diffusion is estimated at ∼1.3 (Hille 2001). We found many AP-related parameters to have Q10s similar to diffusion (0.7–1.3, depending on direction of change). In contrast, we found the temperature sensitivity of the sAHP rise time in neocortical pyramidal neurons to be much more temperature-sensitive than diffusion [Q10 ∼ 0.2–0.3 (equivalent to Q10 ∼ 3–5 for measurements that increase with temperature: see methods)]. This finding rules out diffusion as the primary limit for the sAHP's slow rise in pyramidal neurons (cf. Sah and McLachlan 1992 for vagal motoneurons). In addition, the very high temperature sensitivity for the rise time of the sAHP suggests that a Ca2+-dependent intermediate may be important, rather than intrinsically slow gating of the sAHP channels.

mAHP versus sAHP

The apamin-sensitive mAHP in pyramidal neurons is mediated by sK-type channels (Bond et al. 2004; Villalobos et al. 2004). Studies in expression systems have revealed that the Ca2+ sensitivity of sK channels is conferred by an integral association with calmodulin (Maylie et al. 2004). This complex is activated by Ca2+ with a Kd of ∼400–500 nM (Hill coefficient 4–5) (Kohler et al. 1996). Our previous study showed that the mAHP decays faster than the decay of [Ca2+]i in either soma or dendrites and bulk cytoplasmic [Ca2+]i was a poor indicator of activation of the current underlying the mAHP (Abel et al. 2004). These data suggest that sK channels in neocortical pyramidal neurons respond to a restricted domain of [Ca2+]i. Our present results indicate modest temperature sensitivity of mAHP kinetics (much less than for the decay of [Ca2+]i), consistent with tight linkage of sK channels and Ca2+ entry. The magnitude of the changes with temperature that we observed for input resistance suggests that temperature-related changes in the membrane time constant could account for much of the changes in the time course of the mAHP.

If the sAHP was also due to sK channels, we would expect similar sensitivity to [Ca2+]i and changes in temperature as the mAHP. This was not the case. The temperature sensitivity of the decay of the sAHP was very high and similar to that of the decay of [Ca2+]i (Q10 ∼ 0.2–0.4), suggesting a closer relationship between the sAHP channels and cytoplasmic [Ca2+]i than mAHP channels and [Ca2+]i. In neocortical pyramidal cells, sAHP amplitude decreased and the decay τ was prolonged at RT. The sAHP was also enhanced by cooling in hippocampal CA1 pyramidal neurons (Shen and Schwartzkroin 1988; Thompson et al. 1985), causing increased spike-frequency adaptation at lower temperatures.

Our previous study suggests that bulk cytoplasmic [Ca2+]i is proportional to the Ca2+ signal that activates the sAHP channels and that the time course of the sAHP is similar (but not identical) to that of somatic [Ca2+]i transients (Abel et al. 2004). With cooling to RT, there was no significant change in the amplitude of the train-induced [Ca2+]i and the decay τ was prolonged. Further, in most cells a second, slower τdecay was evident at RT, suggesting that at least two different classes of mechanisms are involved in restoration of [Ca2+]i. The decay τs were also highly temperature sensitive (Q10 ∼ 0.2–0.4). Again, the values for the τs for sAHP decay and somatic [Ca2+]i decay do not match well, suggesting imperfect tracking of bulk cytosolic Ca2+ by the sAHP channels.

Temperature sensitivity had not previously been investigated for decay of [Ca2+]i following trains of APs. Our findings show similar high temperature sensitivity for the decay of [Ca2+]i following a single AP or a train of APs. In response to a single AP, Borst and Sakmann (1998) reported that at the calyx of Held, changes in [Ca2+]i had lower peak, prolonged decay, and larger net charge movement at 24 versus 36°C. Markram et al. (1995) reported that the rise and decay times for Ca2+ transients induced by single back-propagated APs in layer V pyramidal neurons were longer at RT versus 34°C, with Q10s of 2.6–3.1 for decay and for the rise time ∼3. This rise time is unlikely to be relevant to the rise of the sAHP, which occurs over a much slower time scale. The entry of Ca2+ is largely over at the end of spiking (these channels deactivate with τs ≪ 1 ms) (Lorenzon and Foehring 1995; Stewart and Foehring 2001; but see Marrion and Tavalin 1998).

Possible mechanisms for the enhanced AHP with cooling include greater Ca2+ entry due to broader action potentials or altered buffering (sequestration, extrusion) of intracellular [Ca2+] (Shen and Schwartzkroin 1988; Thompson et al. 1985; Volgushev et al. 2000b). One would expect that changes in spike height and width would alter Ca2+ entry in cortical neurons (cf. Stewart and Foehring 2001) and thus alter Ca2+ transients in response to APs. Increased Ca2+ entry should manifest as an increased peak [Ca2+]i (or perhaps increased time integral of [Ca2+]i). Altered buffering could change the peak and would result in a change in the decay of the [Ca2+]i transient. In layer II/III pyramidal neurons, we found little change in peak [Ca2+]i, but prolonged decay times at lower temperatures. These data suggest that while Ca2+ entry is increased at RT, the mechanisms for restoration of resting [Ca2+]i are even more sensitive to temperature. The similarities between decay of the sAHP and [Ca2+]i may indicate that Ca2+ regulatory mechanisms confer temperature sensitivity and slow kinetics on the sAHP channels.

Passive properties and single AP

We found that input resistance was very temperature sensitive and increased with cooling (cf. Griffin and Bouland 1995; Thompson et al. 1985; Volgushev et al. 2000b). AP threshold was insensitive to temperature. AP amplitude was consistently enhanced at lower temperatures (Fig. 1B). Spike width increased and the rates of rise and fall of the spike were prolonged at RT, compared with 33°C. Q10's for all of these parameters were modest (near the Q10 for diffusion: i.e., 1.3 for increase with increasing temperature, 0.76 for increase with decreasing temperature), with the exception of AP width (at threshold voltage), which had a Q10 of 0.5. All of these findings are consistent with those of Thompson et al. (1985) in guinea pig CA1 pyramidal neurons and Volgushev et al. (2000b) for layer II/II pyramidal cells in rat visual cortex. Volgushev et al. (2000b) attributed the change in resting potential, input resistance, and action potentials primarily to changes in K+ conductance, with little change in Na+ conductance. We found the Q10 for the rate of spike repolarization to be lower than the Q10 obtained by Thompson et al. (1985). This may be related the lesser influence of Ca2+ -dependent K+ conductances in spike repolarization in neocortical pyramidal cells (Lorenzon and Foehring 1993; Pineda et al. 1998; Schwindt et al. 1988b). versus CA1 pyramidal neurons (Shao et al. 1999; Storm 1990).

The broader APs at RT would result in greater Ca2+ entry through voltage-gated channels. Our previous study of Ca2+ currents in response to mock AP waveforms (at RT) suggests that a twofold increase in spike width would result in ∼1.5-fold increase in total charge entry at RT (Stewart and Foehring 2001). Effects on the peak amplitude of the Ca2+ current are more complicated and depend on whether increased spike width is due to slower rise time (peak increases with increased width) or slowed repolarization (decreased amplitude) (Stewart and Foehring 2001). There was little change in the amplitude of the Ca2+ transient in response to a single spike at RT versus 33°C. In contrast the τdecay was very temperature sensitive, being prolonged at RT (Q10: 0.2). In combination, these results lead to an increase in the integral of [Ca2+]i versus time. These findings are similar to those of Borst and Sakmann (1998) in the calyx of Held and Markram et al. (1995) in layer V pyramidal neurons.


Our principal findings were that Ca2+-dependent events were in general much more temperature sensitive than voltage-dependent ones, our Q10 data show that the slow rise time of the sAHP cannot be explained by diffusion of Ca2+ from a remote source of entry, and the sAHP is much more sensitive to temperature than the mAHP. These data provide further evidence that the channels underlying the sAHP have a different relationship to Ca2+ entry than the mAHP channels (Abel et al. 2004; Pineda et al. 1998). These data are also consistent with the sAHP being more closely related to bulk cytoplasmic [Ca2+]i than the mAHP (although the sAHP and IsAHP are slower than the decay of [Ca2+]i at both temperatures). We favor the hypothesis that the sAHP channels are coupled to Ca2+ entry via a cytoplasmic intermediate. We also found that the temperature changes over the 22–33°C range have modest effects on the amplitude and shape of action potentials in superficial pyramidal cells from rat sensorimotor cortex: lower temperatures result in broader spikes and prolonged decay of Ca2+ transients. The Ca2+ transients resulting from single APs, or trains of APs, are prolonged and their time course is highly temperature sensitive.


This work was supported by National Institute of Neurological Disorders and Stroke Grants NS-33579 to R. C. Foehring and NS-42276 to J. C. Callaway.


We are grateful to S. Phillips for excellent technical assistance.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract