Modulation of the hyperpolarization-activated cation current of rat thalamic relay neurones by intracellular pH

Abstract

1. Properties of the hyperpolarization-activated cation current (I_h) were investigated in thalamocortical neurones of an in vitro slice preparation of the rat ventrobasal thalamic complex (VB) before and during changes of pipette pH (pH_p), intracellular pH (pH_i) and bath pH (pH_b) using the whole-cell patch-clamp technique and fluorescence ratio imaging of the pH indicator 2′,7′-bis(carboxyethyl)-5(and -6)-carboxyfluorescein (BCECF).

2. Recording of I_h with predefined pH_p revealed significant shifts in the voltage dependence of I_h activation (V_½) of 4-5 mV to more positive values for a pH_p of 7.5 and 2-3 mV to more negative values for a pH_p of 6.7 as compared to control values (pH_p= 7.1).

3. Application of the weak acid lactate (20 mM), which produced a slow monophasic intracellular acidification, induced a reversible negative shift of V_½ of up to 3 mV. Application of 20 mM TMA, which caused a distinct intracellular alkalinization, shifted V_½ to 4-5 mV more positive values.

4. In slices bathed in Hepes-buffered saline, no significant pH_o dependence of I_h was observed. Changing pH_o by altering the extracellular [HCO₃⁻] in the presence of constant pCO₂ also revealed no significant pH_o dependence of I_h.

5. Rhythmic stimulation of thalamocortical neurones with repetitive depolarizing pulse trains caused an intracellular acidification, which reversibly decreased the amplitude and time course of activation of I_h.

6. The results of the present study indicate that shifts in pH_i result in a significant modulation of the gating properties of I_h channels in TC neurones. Through this mechanism activity-dependent shifts in pH_i may contribute to the up- and downregulation of I_h.

Thalamocortical (TC) neurones can generate intrinsic oscillations due to the regular occurrence of low-threshold Ca²⁺ spikes and a depolarizing pacemaker potential (McCormick & Pape, 1990a;Soltesz et al. 1991). They are also part of a network which can generate both normal and pathological forms of synchronized oscillations, as for instance sleep spindles during slow-wave sleep and spike and wave discharges during generalized absence seizures (Steriade & Deschênes, 1984; Niedermeyer, 1990; Steriade et al. 1993, 1997). The generation of intrinsic oscillations, as well as the network oscillations, are critically dependent upon the h-current (I_h), a cation current activated by membrane hyperpolarization (McCormick & Pape, 1990a;Soltesz et al. 1991; Pape, 1996). Activity-dependent upregulation of I_h in thalamocortical neurones has been shown to result in dampening of intrinsic and network oscillations (McCormick & Pape, 1990b;Leresche et al. 1991; Bal & McCormick, 1996; McCormick & Bal, 1997). An activity-dependent upregulation of I_h mediated by intracellular Ca²⁺ was suggested in models of oscillations in thalamocortical networks (Desthexe et al. 1996). Recently, the role of Ca²⁺-mediated upregulation of I_h for the timing of synchronized thalamocortical oscillations has been directly shown in ferret thalamocortical neurones (Lüthi & McCormick, 1998). However, the mechanism of Ca²⁺-dependent upregulation of I_h in neurones is still controversial (Ingram & Williams, 1996; Budde et al. 1997; Lüthi & McCormick, 1998). Besides Ca²⁺, other ions may be critically involved in the control of voltage-dependent membrane conductances, such as I_h.

For instance, neuronal activity is accompanied by transient shifts in extracellular as well as intracellular pH, both under physiological and pathophysiological conditions (Caspers & Speckmann, 1972; Chesler, 1990; de Curtis et al. 1998; for review see also Kaila & Ransom, 1998). The exact time course of pH_o and pH_i shifts may vary between different neural tissues but most often a transient extracellular alkalinization is followed by a slower persistent extracellular acidification (Chesler & Kaila, 1992). In coronal slices of rat dorsal lateral geniculate nucleus, for instance, recordings with pH microelectrodes have revealed transient alkaline shifts of up to 0.04 pH units upon synchronous activation (Tong & Chesler, 1999). As the pH_o shifts reflect transmembrane flux of acid or base equivalents, they are accompanied by pH_i changes. Either one may lead to modulation of neuronal excitability, since even modest shifts in pH_o and pH_i have been shown to affect a variety of voltage- and ligand-gated ion channels (Moody, 1984; Traynelis & Cull-Candy, 1990; Tombaugh & Somjen, 1996; for review see also Kaila & Ransom, 1998). The possible role of activity-related changes in the concentration of H⁺ on I_h has not been considered in detail as yet, although early reports exist on a depressant effect of H⁺ on gating of I_h in starfish oocytes (Moody & Hagiwara, 1982) and lobster stretch receptor neurones (Edman & Grampp, 1991).

In the present study we investigated the possible modulation of I_h by extracellular and intracellular pH shifts in thalamocortical neurones in a slice preparation of rat ventrobasal thalamic complex (VB). We demonstrate that the voltage dependence and kinetics of I_h are sensitive to the intracellular concentration of H⁺. We hypothesize that modulation of I_h by intracellular pH may be involved in the timing of intrinsic oscillations in TC neurones, thereby contributing to control of synchronized oscillatory activity in thalamocortical networks. A preliminary report of some of the results has been communicated to the European Neuroscience Association (Munsch & Pape, 1998).

METHODS

Slice prepapration

Juvenile Long-Evans rats of postnatal age P12-P16 were deeply anaesthetized with fluothane (∼5 ml in a 2 l beaker) and decapitated. A block of brain tissue containing the VB was rapidly removed and immersed in cold (2-4°C) saline of the following composition (mM): sucrose, 195; glucose 11; Pipes, 20; KCl, 2.5; MgSO₄, 10; CaCl₂, 0.5; pH 7.35 with NaOH. Coronal slices of the thalamus were cut at 300 μm on a vibratome (Model 1000, Ted Pella, Redding, CA, USA), and kept submerged at room temperature (22-25°C) in artificial cerebrospinal fluid (ACSF) containing (mM): NaCl, 125; KCl, 2.5; NaH₂PO₄, 1.25; NaHCO₃, 22-26; MgSO₄, 2; CaCl₂, 2; glucose, 20; pH 7.35 by equilibration with 95 % O₂-5 % CO₂. The [HCO₃⁻] of the superfusate was somewhat lower than can be calculated from the Henderson-Hasselbalch equation for a pH of 7.35 to account for the loss of CO₂ between storage container and experimental bath. However bath pH was routinely checked and adjusted by adding small amounts of NaHCO₃ to the stock solution. In Hepes-buffered ACSF, NaHCO₃ was substituted by an equimolar amount of Hepes and pH adjusted to the desired value by adding NaOH. The [HCO₃⁻] used to change bath pH at constant pCO₂ (5 %) was 15-17 mM for pH 7.2, 7-9 mM for pH 6.9 and 29-31 mM for pH 7.5. Again, actual bath pH was checked with a pH electrode and adjusted by adding small amounts of NaHCO₃ to the stock solution.

All substances were obtained from Sigma (St Louis, MO, USA) except for Hepes (SERVA, Heidelberg, Germany).

Electrophysiological recording

Recordings were performed on VB neurones in the slice preparation. Individual neurones were approached under visual control by infrared videomicroscopy with patch pipettes pulled from borosilicate glass (GC 150TF-10, Clark Electromedical Instruments, Pangbourne, UK) and filled with the following pipette solution (mM): potassium gluconate, 95; potassium citrate, 20; NaCl, 10; Hepes, 10; MgCl₂, 1; CaCl₂, 0.1; 1,2-bis(2-aminophenoxy)-ethane-N,N,N′,N′-tetraacetic acid (BAPTA), 1; MgATP, 3; Na₂GTP, 0.5; pH adjusted to 7.2 with KOH. For recordings with predefined pH values in the pipette Hepes concentration was increased to 30 mM and pH was carefully adjusted using N-methyl-D-glucamine (NMDG) in order to keep the transmembrane driving force for K⁺ constant (DiFrancesco, 1982). Membrane currents and membrane voltages were recorded in the whole-cell configuration using an EPC-9 amplifier operating Pulse software (v. 8.01, HEKA, Lambrecht, Germany). Recordings were started 10-15 min after obtaining the whole-cell configuration to allow stabilization of intracellular pH. Typical electrode resistances were 2-3 MΩ and access resistance ranged from 5-15 MΩ. Series resistance compensation of ≥30 % was routinely used. All measured potentials were corrected off-line for a liquid junction potential of 12 mV. Prior to measurements, slices were transfered to an experimental bath and continuously superfused with ACSF at room temperature (22-25°C). Sometimes, 0.5-1 mM Ba²⁺ or 0.5-1 mM Cs⁺ was added to the ACSF as indicated. In Ba²⁺-containing solutions MgSO₄ was substituted by an equimolar amount of MgCl₂. In solutions containing tetramethylammonium (TMA) or lactate, NaCl was substituted by an equivalent amount of TMA-Cl or sodium lactate.

The steady-state activation curve of I_h was constructed by measuring tail currents elicited by repolarizing the membrane to a fixed potential (-63 mV) following 2 s voltage steps to test potentials between -118 or -108 mV and -48 mV. In contrast to previous observations on thalamic relay cells, the voltage protocol used in the present study did not evoke low voltage-activated calcium currents (I_LVA) when stepping back to -63 mV under the present experimental conditions. Measurements of I_h tail current amplitudes were carried out only in traces uncontaminated by the I_LVA current, whose presence would have otherwise compromised their accuracy. Tail currents were measured 50 ms after termination of the hyperpolarizing voltage pulse, and normalized by:

(1)

The resulting data could be well fitted by a Boltzmann function of the form:

(2)

from which the value of half-maximal activation (V_½) and the slope factor (k) were obtained. The fully activated current vs. voltage (I-V) relationship was obtained by calculating the ratio of I_h amplitudes during activation and deactivation for each voltage step and multiplying it by the amplitude of the maximal tail current for the largest test pulse. From plots of these data against the test potentials, the fully activated conductance underlying I_h (G_h) and the reversal potential (V_rev) could be obtained (Pape, 1994).

The time course of I_h activation was best approximated by single-exponential fits according to the equation:

(3)

where I_h(t) is the the amplitude of the current at time t, a_o and a₁ are constants and τ is the time constant.

pH imaging

VB neurones were loaded with 10-50 μm of the fluorescent pH indicator BCECF (Teflabs, Austin, TX, USA) via the patch pipette after establishing the whole-cell configuration. The dye was allowed to diffuse into the cell and equilibrate for at least 10 min before the recordings were started. Excitation of the dye was performed with a rapid scanning monochromator (TILL Photonics, Munich, Germany) coupled to the microscope (Zeiss, Axioscope FS, Oberkochen, Germany) via a light guide. Excitation light (420 or 440 and 495 nm, 500 ms exposure time) was directed onto the slice with a 505 nm dicroic (Omega Optical, Brattleboro, VT, USA) through a water immersion objective (LUMPLFL 40 × W, Olympus, Hamburg, Germany). Fluorescence was measured with a 12 bit slow-scan CCD camera (Theta System Elektronik GmbH, Gröbenzell, Germany or Princeton Instruments, Trenton, NJ, USA) after passing a 520 nm long-pass filter (Zeiss). Analysis was restricted to the cell somata. Aquisition rate was 0.03-1 Hz in order to minimize photobleaching of BCECF and photodynamic cell damage. Calibration of background-subtracted fluorescence ratios (F₄₂₀ or F₄₄₀/F₄₉₅) with the nigericin-high K⁺ method was not performed since it does not work properly in slice preparations (Trapp et al. 1996a). The null-point calibration method (Eisner et al. 1989) was not used since all pH_i regulatory mechanisms have to be blocked, which was not tolerated by the thalamic slices. However, to obtain some rough estimates for the magnitude of the changes in pH_i induced by weak acids and bases or the activity-induced acidification, we calculated pH_i values from the changes in BCECF fluorescence ratio in response to application of TMA and lactate (Szatkowski & Thomas, 1986; Eisner et al. 1989).

Statistics

The data in the results section are presented as means ±s.d. and as means ±s.e.m. in the figures, as indicated. Statistical significance was tested with an unpaired t test for small samples (Dixon & Massey, 1969) or Student's paired t test from the statistics package provided by Origin, v. 4.0 (Microcal, Northampton, MA, USA), as indicated.

RESULTS

Recording of TC neurones from rat VB in the whole-cell configuration of the patch-clamp technique revealed activation of two inward current components upon membrane hyperpolarization from a holding potential of -38 or -48 mV (Fig. 1), i.e. an instantaneous component and a component developing with a slower time course of activation. By adding 0.5-1 mM Ba²⁺ to the bathing solution (Fig. 1A, middle family of current traces), the instantaneous component was blocked leaving only a small ‘leak’ current and the slow current component. Rapid activation and Ba²⁺ sensitivity are characteristic of I_Kir, as recently reported for TC neurones from cat VB (Williams et al. 1997). The application of 0.5-1 mM Cs⁺ in the presence of Ba²⁺ blocked the second inward current component with slower, time-dependent activation (Fig. 1A, bottom family of current traces), which is characteristic of I_h in thalamic neurones (McCormick & Pape, 1990a;Soltesz et al. 1991). It also shows that there was no contamination of the tail currents by the I_LVA, at least under the present experimental conditions. Analysis of the I-V relationships (n = 4) showed that I_h, measured as the difference of the steady-state inward current 50 ms before the end of the test pulse and the instantaneous current at the begining of the test pulse (I_stst- I_inst) was not affected by the concentrations of Ba²⁺ used in this study (Fig. 1B). In order to isolate I_h, we therefore added 0.5-1 mM Ba²⁺ routinely to the ACSF.

Figure 1. Hyperpolarization-activated currents in TC neurones of VB.

A, families of current traces recorded in a VB neurone upon hyperpolarizing voltage steps from -48 mV in 5 mV increments to -108 mV under control conditions and in the presence of 0.5 mM extracellular Ba²⁺ and 0.5 mM extracellular Cs⁺, as indicated. Holding potential was -38 mV. B, I-V relationship of I_h, measured as the difference between the steady-state inward current 50 ms before the end of the test pulse (I_stst) and the instantaneous inward current (I_inst) obtained at the beginning of the hyperpolarizing voltage steps, before and after the addition of Ba²⁺ and Cs⁺.

pH sensitivity of I_h

In order to test for a possible pH_i dependence of I_h, TC neurones were recorded with different predefined pH values in the patch pipette solution (Fig. 2). Analysis of the activation curves of I_h for pH values of 6.7, 7.1 and 7.5 (Fig. 2A) revealed significant shifts in the half-activation potential. Acidic pH (6.7) led to a leftward shift of V_½ to 2-3 mV more negative potentials, while alkaline pH (7.5) resulted in a shift to 4-5 mV more positive potentials as compared to a control pH of 7.1 (Table 1). In contrast, no significant differences (P > 0.1) in the fully activated conductance underlying I_h and the extrapolated reversal potential of I_h (Table 1) were observed between the different populations of TC neurones recorded with different pH_p values (Fig. 2B). Examination of the current traces (Fig. 2C) revealed that the time constant (τ) of activation of I_h appeared to decrease with an increase in pH_p. Single-exponential fits of current traces recorded with different pH values (Fig. 3A) corroborated the strong voltage dependence of the I_h activation time course, and in addition showed a substantial dependence on pH_p. At more acidic pH (6.7), τ was increased, while at a more alkaline pH (7.5), τ was decreased as compared to a pH value of 7.1 (Fig. 3B). For a test pulse to -93 mV, for instance, the time constant for activation of I_h was 1.74 ± 0.88 s (n = 8) for a pH_i of 6.7, 1.25 ± 0.46 s (n = 8) for pH_i= 7.1 and 827 ± 215 ms (n = 7) for a pH_i of 7.5.

Figure 2. Modulation by pH_p of I_h in TC neurones.

Mean steady-state activation curve (A) and fully activated I-V relationship (B) of I_h, recorded using pipette solutions of different pH values. Lines in A and B represent best fits of a Boltzmann distribution and linear regression analysis to the data points, respectively. Data represent means (n = 7) and s.e.m.C, examples of current families recorded with different pipette solutions as indicated. Inset shows voltage protocol.

Table 1.

pH sensitivity of I_h activation

pHp	V1/2(mV)	k	Gh(pA mV−1)	Vrev(mV)	n
7.5	−82.8 ± 1.7*	7.6 ± 1.8	13.2 ± 3.4	−28.6 ± 2.2	7
7.1	−87.5 ± 2.5*,**	8.3 ± 2.1	12.4 ± 1.9	−31.0 ± 2.1	7
6.7	−90.2 ± 1.9**	8.9 ± 1.3	11.0 ± 3.1	−25.9 ± 3.0	7

Half-maximal activation voltage (V_1/2), slope factor (k), fully activated conductance (G_h) and reversal potential (V_rev) of I_h in TC neurones recorded with Hepes-buffered pipette solutions of different pH values (pH_p). Significant differences are

^*P≤ 0.01, unpaired t test

^**P≤ 0.05, paired t test.

Figure 3. pH dependence of I_h activation kinetics.

A, current traces recorded from VB neurones upon hyperpolarizing voltage steps to -93 and -108 mV, from a holding potential of -48 mV, with pipette solutions of different pH values as indicated. Continuous lines overlying current traces represent single-exponential fits through the data points. B, plot of activation time constant (τ) versus voltage of test steps from -48 to -108 mV for neurones recorded with intracellular solutions of different pH values. Data represent means (n as indicated) and s.e.m.

Modulation of I_h by intracellular alkalinization and acidification

In order to more directly determine the effect of changes in pH_i on I_h in a given cell, weak bases and weak acids were applied with the bathing solution to intracellularly alkalinize and acidify the neurones, respectively. TMA is a weak base producing a monophasic intracellular alkalinization because of the membrane permeability of its non-protonated form (Szatkowski & Thomas, 1989; Eisner et al. 1989; Church et al. 1998). Superfusion of thalamic slices with 20 mM TMA for 10 min led to a distinct intracellular alkalinization of TC neurones, as revealed by the decrease in the BCECF fluorescence ratio (F₄₄₀/F₄₉₅) by about 30 % (mean = 30.7 ± 4.8 %, n = 6, Fig. 4C, bottom trace). Assuming an interstitial pH within the slice of 7.15-7.25, the estimated shift in pH_i was 0.28-0.3 pH units, from an initial pH_i of 6.95-7.05. The effect of TMA-induced intracellular alkalinization on the I_h of TC neurones was measured 8-10 min after TMA addition, as shown in Fig. 4A. For a given test pulse (from -38 to -93 mV in this cell), the steady-state current amplitude of I_h increased as compared to control values. Upon washout of TMA (10 min), I_h amplitude returned to control levels (Fig. 4A). The increase in h-current amplitude was due to a reversible shift in the activation curve to more positive values, as illustrated in Fig. 4B. The half-activation potential significantly (P < 0.01, paired t test) shifted from V_½= -88.8 ± 3.9 mV (k = 9.2± 1.1, n = 5) to V_½= -83.9 ± 4.4 mV (k = 8.8± 1.5, n = 5) in the presence of TMA. The fully activated conductance underlying I_h and the extrapolated reversal potential of I_h were not significantly changed (data not shown). By comparison, weak acids such as lactate have been shown to induce a fall in pH_i (Szatkowski & Thomas, 1989; Church et al. 1998). Application of 20 mM lactate with the superfusate for 10-15 min caused a slow monophasic intracellular acidification, as revealed by the increase in F₄₄₀/F₄₉₅ by about 9 % (mean = 8.8 ± 1.4 %, n = 5, Fig. 4C, top trace; corresponding to an estimated fall in pH_i by 0.08-0.1 pH units), and had only a moderate effect on I_h. The steady-state amplitude of I_h, as measured during a period of 10-12 min after addition of lactate, was reduced upon a step hyperpolarization to -93 mV (Fig. 4A), due to a shift in the mean half-activation potential to V_½= -91.5 ± 3.7 mV (k = 8.7± 0.7, n = 5) as compared to the control value (Fig. 4B, P≤ 0.05, paired t test). Upon washout, the effect of lactate was fully reversible (Fig. 4A and C).

Figure 4. Modulation of I_h by intracellular acidification and alkalinization.

A, family of current responses to hyperpolarizing test pulses (see inset) under control conditions and in the presence of either TMA or lactate, as indicated. B, steady-state activation curves plotted from normalized tail current amplitudes recorded in ACSF (control), 8-10 min after addition of TMA or lactate and after 10 min wash with ACSF. Continuous lines represent best fits of a Boltzmann distribution to the data points. Data represent means (n as indicated) and s.e.m.C, plots of BCECF fluorescence ratio (F₄₄₀/F₄₉₅, expressed as a percentage) versus time measured in TC neurones upon superfusion of the slices with 20 mM TMA or 20 mM lactate. Upward deflection of F₄₄₀/F₄₉₅ indicates an acidosis. Acquisition rate was 1 Hz. The cell was loaded with 10 μm BCECF via the patch pipette.

In the next experimental step, the possibility was investigated that the effect of pH_i changes on I_h is mediated via modulation of adenylate cyclase activity and subsequent changes in the intracellular cAMP concentration (see Pape, 1996). Addition of the adenylate cyclase activator forskolin (10 μm) to the superfusing solution increased I_h elicited by step hyperpolarizations (10 min after forskolin application) from -48 to -98 mV (Fig. 5A) and shifted V_½ to -82.9 ± 2.1 mV (k = 7.8± 1.0, n = 5), as compared to control (V_½= -88.2 ± 2.6 mV, k = 7.5 ± 0.7, n = 5). Additional application of lactate or TMA revealed that forskolin-upregulated I_h was still modulated by lactate-induced intracellular acidification or TMA-induced intracellular alkalinization (Fig. 5A and B). The half-maximal activation voltage significantly (P≤ 0.05, paired t test) shifted to V_½= -85.5 ± 3.7 mV (k = 7.7± 1.1, n = 5) upon exposure to lactate (20 mM) and to V_½= -79.9 ± 3.6 mV (k = 7.1± 1.7, n = 5) after addition of TMA (20 mM). To corroborate these findings, the effects of lactate and TMA were tested on TC neurones recorded with cAMP (100 μm) in the pipette solution, which shifts the voltage dependence of activation presumably by a direct effect on the channels underlying I_h (reviewed by Pape, 1996). In the presence of intracellular cAMP (100 μm), step hyperpolarizations to -98 mV revealed a decrease in I_h amplitude upon addition of lactate and an increase in I_h amplitude following application of TMA (Fig. 5C). Analysis of the activation curves of I_h revealed a modulation of the voltage dependence of activation (Fig. 5D). Lactate (20 mM) significantly (P < 0.01, paired t test) shifted the activation curve to more negative potentials (V_½= -89.9 ± 2.2 mV, k = 7.5± 0.8, n = 5), whereas TMA (20 mM) induced a rightward shift to more positive potentials (V_½= -84.3 ± 2.2 mV, k = 6.6± 0.9, n = 5) as compared to control (with 100 μm cAMP, V_½= -86.6 ± 2.1 mV, k = 7.4± 0.7, n = 5). The findings suggest that the effects of TMA and lactate occur through a pathway independent of the adenylate cyclase-cAMP system and may be due, at least partly, to a direct effect of pH_i on the properties of I_h channels.

Figure 5. Influence of the adenylate cyclase-cAMP system on TMA and lactate effects on I_h.

A, examples of current responses to hyperpolarizing voltage steps from -48 to -98 mV recorded under control conditions in the presence of forskolin (10 μm), and during addition of lactate (20 mM) or TMA (20 mM) to the forskolin-containing solution. B, steady-state activation curves recorded under control conditions, after bath application of 10 μm forskolin and 8-10 min after addition of either 20 mM lactate or 20 mM TMA in the presence of forskolin (data averaged from recordings in 5 cells). C, family of current responses to voltage steps to -98 mV with cAMP (100 μm) in the pipette solution (control) and after addition of lactate (20 mM) and TMA (20 mM). D, activation curves recorded with 100 μm cAMP in the patch pipette solution 8-10 min after application of 20 mM lactate or 20 mM TMA with the superfusing solution (n = 5). Continuous lines in A and B represent fits of a Boltzmann distribution. Error bars were omitted for clarity.

Effects of changes in bath pH

In order to test for a dependence of I_h of TC neurones on extracellular pH, slices were superfused with Hepes-buffered ACSF titrated to different pH values, and h-currents were elicited by step hyperpolarizations as before. Figure 6A illustrates sample current traces obtained from one cell successively superfused with salines of different bath pH (pH_b), as indicated. The I_h activation curves (Fig. 6B) constructed from these experiments (measured 8-13 min after changing bath pH) revealed no significant differences (P > 0.1) between the half-activation voltages of cells recorded in salines of pH_b 6.8 (V_½= -86.6 ± 2.9 mV, k = 9.7± 0.7, n = 5), pH_b 7.2 (V_½= -85.1 ± 2.8 mV, k = 9.5± 0.6, n = 5) and pH_b 7.6 (V_½= -83.8 ± 2.3 mV, k = 8.9± 0.2, n = 5). Furthermore, there were no statistically significant differences in the fully activated conductance or extrapolated reversal potential of I_h (data not shown). Monitoring BCECF fluorescence in a different sample of TC neurones under the same experimental conditions revealed moderate changes in pH_i upon changes of the pH of the superfusate (Fig. 6C). F₄₄₀/F₄₉₅ decreased by 5.9 ± 2.4 % (n = 5) upon changing bath pH for 10-15 min from 7.2 to 7.6 and increased by 5.3 ± 1.4 % (n = 5) upon changing bath pH to 6.8, indicating a moderate intracellular alkalinization of TC neurones in more alkaline superfusate and intracellular acidification in more acidic superfusate.

Figure 6. Effects of Hepes-buffered solutions on I_h and on pH_i.

A, current responses to hyperpolarizing voltage steps from a holding potential of -48 mV to -108 mV recorded in Hepes-buffered bathing solutions of different pH values. B, steady-state activation curves recorded at different pH values of Hepes-buffered salines, as indicated. Continuous lines in B represent best fits to a Boltzmann distribution. Error bars were omitted for clarity. C, plot of BCECF fluorescence (F₄₄₀/F₄₉₅ expressed as a percentage) vs. time measured in TC neurones superfused with Hepes-buffered salines of different pH. Upward deflection of F₄₄₀/F₄₉₅ indicates an acidosis. Aquisition rate was 1 Hz. The cell was loaded with 10 μm BCECF.

Since superfusion of the slices with Hepes-buffered salines may probably wash out most of the HCO₃⁻, and a pH_o sensitivity of I_h may depend on the presence of HCO₃⁻ as the main buffer (Church, 1992), we changed the [HCO₃⁻] of the saline at constant pCO₂ and varied pH_b. Under these conditions, there was no indication of pH_b-dependent changes in either the time course of activation of I_h or the steady-state amplitude of I_h (Fig. 7A). Significant differences in the half-activation voltage of I_h of neurones recorded in salines of pH_b 7.2 (V_½= -91.4 ± 1.9 mV, k = 8.3± 1.1, n = 5) and pH_b 7.5 (V_½= -90.8 ± 2.3 mV, k = 7.8± 0.7, n = 5) (Fig. 7B) and in the fully activated conductance or the reversal potential of I_h did not occur (data not shown). Monitoring BCECF fluorescence under these conditions revealed even smaller changes in F₄₄₀/F₄₉₅ upon changing bath pH (Fig. 7C). Superfusion of the slices for 10-15 min with a solution of pH 7.5 decreased F₄₄₀/F₄₉₅ by 3.7 ± 1.2 % (n = 4), and a bath pH of 6.9 increased F₄₄₀/F₄₉₅ by 2.5 ± 0.4 % (n = 4).

Figure 7. Lack of significant pH_b dependence of I_h and pH_i in HCO₃⁻-buffered saline.

Current responses to hyperpolarizing test pulses (A) and steady-state activation curves (B) of VB neurones recorded in HCO₃⁻-buffered salines of different pH values. Continuous lines in B represent best fits to a Boltzmann distribution. Error bars were omitted for clarity. C, plot of BCECF fluorescence (F₄₄₀/F₄₉₅, expressed as a percentage) vs. time measured in VB neurones superfused with HCO₃⁻-buffered salines of different pH. Upward deflection of F₄₄₀/F₄₉₅ corresponds to an intracellular acidosis. Aquisition rate was 1 Hz. The cell was loaded with 10 μm BCECF.

Activity-dependent effects on I_h

Since increased activity is typically associated with a slow intracellular acidification in mammalian neurones (reviewed in Chesler, 1990; and Kaila & Ransom, 1998), we tested whether activity-dependent changes in pH_i affect I_h in TC neurones. Changes in intracellular pH upon repetitive stimulation under voltage clamp conditions with depolarizing pulse trains were directly monitored with BCECF (Fig. 8A). Repetitive stimulation with 10 depolarizing pulse trains at a frequency of 0.1 Hz led to a pronounced acidification of TC, as revealed by the distinct increase in F₄₄₀/F₄₉₅ of about 40 % (mean = 34.8 ± 6.1 %, n = 11, Fig. 8A). According to the estimations of changes in pH_i induced by TMA and lactate this corresponds to a shift in pH_i of 0.34-0.4 pH units. Immediately before and after 10 trains, I_h was activated by a hyperpolarizing voltage step to -93 mV of 2 s duration from a holding potential of -38 mV (Fig. 8B). In addition, stepping back to -63 mV following the 2 s voltage step now evoked a T current. These pulse train-evoked T currents were not further investigated in this study. The activity-dependent intracellular acidification resulted in a slower time course of activation and a strongly reduced h-current amplitude (Fig. 8B). These findings support the hypothesis that activity-related pH_i shifts may contribute to the control of rhythmic burst firing in TC neurones via modulation of I_h.

Figure 8. Reduction of I_h by activity-dependent intracellular acidification.

A, plot of BCECF fluorescence ratio (F₄₄₀/F₄₉₅) versus time of a VB neurone recorded before, during and after stimulation with 10 pulse trains delivered at 0.1 Hz (each train consisting of 10 pulses of 100 ms duration to +12 mV, separated by hyperpolarizing steps of 25 ms duration to -78 mV; see inset in A). At the points indicated by a and b, test pulses to -93 mV (see inset in B) were applied and the resulting current responses are shown in B. F₄₄₀/F₄₉₅ was monitored in a somatic region of the cell, following loading with 50 μm BCECF via the patch pipette. Upward deflection of F₄₂₀/F₄₉₅ indicates an intracellular acidosis.

DISCUSSION

The present investigation indicates that intracellular H⁺ may act as an modulator of I_h, relating activity-dependent shifts in intracellular pH to changes in the excitability of TC neurones. Activity-related acidification leads to downregulation of I_h, while alkalinization upregulates I_h, due to shifts in its voltage dependence of activation.

Mechanism of I_h modulation by H⁺

The channels underlying I_h are characterized by their sensitivity to low concentrations of extracellular Cs⁺ and their relative insensitivity to extracellular Ba²⁺, which makes these ions suitable tools to isolate the hyperpolarization-activated cation current from other inwardly rectifying conductances such as I_Kir (for review see Pape, 1996). As there is evidence for the existence of I_Kir in TC neurones (Karschin et al. 1996; Williams et al. 1997), and some members of the family of Kir channels have been shown to be particularly sensitive to changes in intracellular pH in the physiological range (for review see Fakler & Ruppersberg, 1996), particular care was taken in the present study to block the instantaneous inward rectifier current and an as yet unidentified leak current with Ba²⁺, in order to study the pH sensitivity of I_h in isolation.

Effects of various ions on I_h have been extensively studied in neuronal preparations (Pape, 1996). Among them, a depressant effect of extracellular H⁺ on I_h was observed in lobster stretch receptor neurones and was suggested to contribute to the control of membrane excitability during varying external conditions (Edman & Grampp, 1991). Cultured neurones of the fetal rat medulla were recently shown to be inhibited by increases in bath pCO₂, mediated by CO₂/H⁺-induced inhibition of I_h (Wellner-Kienitz & Shams, 1998). However, the site of action of H⁺ and the underlying biophysical mechanisms remained largely unclear, in particular because changes in intracellular pH due to variations of pH_o could not be ruled out, as has been shown, for instance, for cultured rat hippocampal neurones (Church et al. 1998). In the present study moderate changes in pH_i were caused by changes in bath pH. However, these changes in pH of Hepes-buffered solutions or changes in the [HCO₃⁻] at constant pCO₂ showed no significant effects on the voltage dependence or conduction properties of I_h. Two major explanations may account for this: either the shifts in pH_i in response to changes in pH_b were insufficient to modulate the channels underlying I_h or the actual pH_o shifts near the cellular membrane were much smaller than those measured in the bath solution (Trapp et al. 1996a).

The shift in the activation curve of I_h, but unaltered fully activated conductance and reversal potential, indicates that changes in pH_i affect the gating properties rather than the conductance of the channels underlying I_h. These conclusions are different from those obtained in lobster stretch receptor neurones and fetal rat medulla neurones, in which H⁺ decreased the conductance of I_h without affecting its voltage dependence (Edman & Grampp, 1991; Wellner-Kienitz & Shams, 1998). Based upon the observation of a small but significant shift in the reversal potential, it was concluded that variations in pH_o lower the I_h channel conductivity for Na⁺ more than for K⁺.

In several neuronal and non-neuronal cell types including TC neurones in rat thalamus, modulation of the voltage dependence of I_h was shown to be mediated by the intracellular adenylate cyclase-cAMP system (for review see Pape, 1996). Therefore, changes in the intracellular cAMP concentration as a result of adenylate cyclase activation would shift the activation curve of I_h to more positive potentials. Different subtypes of adenylate cyclase (type I and VIII) were found to be expressed in the rat thalamus (Matsuoka et al. 1997) and H⁺ is capable of modulating the basal activity of adenylate cyclase (Johnson, 1982), which, in turn, might affect pH_i via the cAMP system. Therefore, the question arises whether H⁺ directly affects h-channels or acts through modulation of adenylate cyclase activity and hence the cAMP level. In the present study lactate- and TMA-induced effects on I_h were not occluded by near-maximal stimulation of adenylate cyclase activity through forskolin (10 μm) or addition of 100 μm cAMP to the internal solution. These results argue in favour of a direct modulation of I_h channels through intracellular H⁺, although additional experiments should be performed addressing this issue. For instance, the genes encoding hyperpolarization-activated cation channels have recently been identified (Ludwig et al. 1998; Gauss et al. 1998; Santoro et al. 1998), and studies on single, re-expressed I_h channels will greatly facilitate the evaluation of the detailed mechanisms underlying modulatory influences.

Possible significance of I_h modulation by H⁺

Modulation of I_h has been shown to be important for the timing of rhythmic membrane potential oscillations in TC neurones (Bal & McCormick, 1996; Lüthi & McCormick, 1998).In particular, activity-related increases in the intracellular Ca²⁺ concentration resulting from rebound Ca²⁺ burst firing were shown to cause a persistent upregulation of I_h by shifting its activation curve to more positive potentials, presumably mediated via adenylate cyclase activity (Budde et al. 1997; Lüthi & McCormick, 1998). In addition, various other types of ions and second messenger pathways with important functional implications for the regulation of I_h have been reported in several neuronal preparations (see Pape, 1996 for review; Ingram & Williams, 1996; Cathala & Paupardin-Tritsch, 1997). This multitude of mechanisms suggests a rather complex scenario of I_h modulation. In the present study, repetitive activation of TC neurones with a stimulation protocol trying to follow as closely as possible the voltage responses during rhythmic oscillatory activity as, for instance, occurring during sleep spindle oscillations (Bal & McCormick, 1996; Budde et al. 1997) resulted in a pronounced transient intracellular acidification. This activity-related acidification is most likely to be due to secondary mechanisms, involving e.g. plasmalemmal Ca²⁺-H⁺ exchange, after stimulus-induced Ca²⁺ influx through voltage-gated Ca²⁺ channels (Schwiening & Boron, 1994; Ou-Yang et al. 1995; Trapp et al. 1996b). Our data provide direct evidence for the contribution of activity-related intracellular pH shifts in the modulation of I_h. Intracellular alkalinization results in upregulation of I_h and intracellular acidification, occurring during repetitive activity, results in downregulation of I_h. Following from this is the conclusion that upregulation and downregulation of I_h, resulting from Ca²⁺ influx and subsequent intracellular acidification due to Ca²⁺-H⁺-exchange, may reciprocally interact, thereby forming an activity-dependent mechanism to control the temporal characteristics of rhythmic oscillations in the thalamocortical network.