The weaver Mutation Causes a Loss of Inward Rectifier Current Regulation in Premigratory Granule Cells of the Mouse Cerebellum

Abstract

Considerable interest has recently focused on theweaver mutation, which causes inward rectifier channel alterations leading to profound impairment of neuronal differentiation and to severe motor dysfunction in mice (Hess, 1996). The principal targets of mutation are cerebellar granule cells, most of which fail to differentiate and degenerate in a premigratory position (Rakic and Sidman, 1973a,b). Two hypotheses have been put forward to explain the pathogenetic role of mutant inward rectifier channels: namely that inward rectifier channel activity is either lacking (Surmeier et al., 1996) or altered (Kofuji et al., 1996; Silverman et al., 1996;Slesinger et al., 1996). We have examined this question by recording inward rectifier currents from cerebellar granule cells in situ at different developmental stages in wild-type and weaver mutant mice. In wild-type mice, the inward rectifier current changed from a G-protein-dependent activation to a constitutive activation as granule cells developed from premigratory to postmigratory stages. In weaver mutant mice, G-protein-dependent inward rectifier currents were absent in premigratory granule cells. A population of putative granule cells in the postmigratory position expressed a constitutive inward rectifier current with properties compatible with mutated GIRK2 channels expressed in heterologous systems. Because granule cells degenerate at the premigratory stage (Smeyne and Goldowitz, 1989), the loss of inward rectifier current and its regulation of membrane potential are likely to play a key role in the pathogenesis of weaver neuronal degeneration.

Ionic channels in neurons have the dual role of controlling membrane potential, thus allowing information coding that is essential for nervous system computation, and of regulating basal cellular processes such as growth, survival, and death (Spitzer, 1991). K⁺ channels play an important role in membrane potential regulation. In the weaver mutation, which causes neuronal degeneration and severe motor dysfunction in mice, an inward rectifier K⁺ channel is altered (Patil et al., 1995), and neuronal degeneration is observed in the cerebellum, substantia nigra, and hippocampus. In particular, most cerebellar granule cells fail to differentiate and die in ectopic position, causing severe disruption of the cerebellar network (Rakic and Sidman, 1973a,b; Sotelo and Changeux, 1974). Although the weaver mutation has been identified, its implications in terms of the pathogenetic mechanism remain largely unknown (Hess, 1996).

It has been shown that the weaver mutant channel expressed in oocytes has lower K⁺ selectivity than the wild-type and has lost the modulatory G-protein control, thus becoming constitutively active (Slesinger et al., 1996). An anomalous Na⁺influx would cause membrane depolarization, possibly facilitating neuronal degeneration through secondary Ca²⁺-dependent processes (Kofuji et al., 1996). The absence of inward rectifier currents in granule cells in culture has suggested, however, that a loss of function was the main effect of the mutation (Surmeier et al., 1996). Here too, uncontrolled membrane depolarization would link the mutation to its pathological consequences. To be realistic, these hypotheses need to be based on a knowledge of the native properties of the inward rectifier current and its effect on membrane potential. This is particularly relevant to granule cells, whose electroresponsiveness follows a specific developmental pattern in situ in terms of both inward rectification and action potential generation (D’Angelo et al., 1997). We therefore have compared inward rectifier currents in wild-type and weaver granule cells, as well as their G-protein modulation and effect on membrane potential, at different developmental stages using whole-cell patch-clamp recordings in acute cerebellar slices.

Here we show that in wild-type mice an inward rectifier current was elicited by G-protein activation in premigratory granule cells. The inward rectifier current changed voltage dependence and kinetics in postmigratory granule cells, thus becoming constitutively active. Inweaver mutant mice, neither constitutive nor G-protein-dependent inward rectifier currents were observed in premigratory granule cells, providing compelling evidence that a loss of inward rectifier function occurs at the time of granule cell degeneration (Smeyne and Goldowitz, 1989; Surmeier et al., 1996). Some neurons in a deep position that expressed a constitutive inward rectifier current with properties compatible with those reported in reconstituted systems (Kofuji et al., 1996; Silverman et al., 1996;Slesinger et al., 1996) apparently corresponded to granule cells that had migrated and developed synaptic connections (Sotelo and Changeux, 1974). A loss of inward rectifier current at the premigratory stage and its regulation of membrane potential are therefore likely to play the key role in the pathogenesis of weaver neuronal degeneration.

MATERIALS AND METHODS

Patch-clamp whole-cell recordings were performed in granule cells in 250-μm-thick mouse cerebellar slices (Edwards et al., 1989). Either wild-type (Swiss strain) or homozygous weaver mutant mice (B6CBACa-A^w-J/A-wv) were used. Homozygousweaver mice were obtained from intercross mating of heterozygous weaver mice (Jackson) and amounted to ∼25% of littermates, consistent with simple mendelian segregation. Homozygous weaver mice were identified on the basis of their evident motor dysfunction, their dramatic cerebellar atrophy, and the abnormal morphology of cerebellar tissue during microscopic observation of the slices (Rakic and Sidman, 1973a,b). The slices were obtained on postnatal days 9–15 (day of birth = P1) from both wild-type and weaver mice. Cerebellar slices were prepared as reported previously (D’Angelo et al., 1995, 1997). Briefly, the mice were anesthetized with halothane (Aldrich, Milwaukee, WI) before they were decapitated. Krebs’ solution for slice cutting and recovery contained (in mm): NaCl 120, KCl 2, MgSO₄ 1.2, NaHCO₃ 26, KH₂PO₄ 1.2, CaCl₂ 2, glucose 11. This solution was equilibrated with 95% O₂ and 5% CO₂, pH 7.4. Slices were maintained at room temperature before being transferred to the recording chamber (1.5 ml) mounted on the stage of an upright microscope (Zeiss Standard-16). The preparations were superfused at a rate of 5–10 ml/min with Krebs’ solution and maintained at 30°C with a feed-back Peltier device (HCC-100A; Dagan Corporation, Minneapolis, MN).

 Data recording and analysis. The excitable responses and membrane currents of the granule cells were recorded using an Axopatch 200-A amplifier. The data were sampled with a TL-1 DMA Interface (sampling time = 250 μsec for current-clamp recordings, 10 μsec for voltage-clamp recordings) and analyzed with P-CLAMP software (Axon Instruments, Foster City, CA). Mossy fiber stimulation was performed with a bipolar tungsten electrode (Clark Instruments, Pangbourne, UK) via a stimulus isolation unit. The stimulating electrode was placed over the mossy fiber bundle, and stimuli were applied at a frequency of 0.1 Hz. Control and test solutions were applied locally through a multi-barrel pipette. Perfusion of control solution was commenced before seal formation and was maintained until we switched to test solutions. QX-314 was obtained from Alomone (Tel-Aviv, Israel), and BAPTA tetrapotassium salt was obtained from Molecular Probes (Eugene, OR).

Patch pipettes were pulled from borosilicate glass capillaries (Hingelberg, Malsfeld, Germany) and had 8–12 MΩ resistance before a seal was formed with a filling solution containing (in mm): potassium gluconate 126, NaCl 4, MgSO₄ 1, CaCl₂0.02, BAPTA 0.1, glucose 15, ATP 3, HEPES 5 (pH was adjusted to 7.2 with KOH), and either GTP 0.1 or GTP-γ-S 0.1. This solution allowed current-clamp recordings of the excitable response as well as voltage-clamp recordings of the inward rectifier current. Considering that this solution was diluted by 10% before use, and assuming a 0.75 activity coefficient for K⁺ salts (Pitzer and Mayorga, 1973), the nernstian reversal potential for K⁺ ions was −85.7 mV. Intracellular Ca²⁺ was buffered at 100 nm, similar to the resting Ca²⁺ concentration measured in granule cells (Marchetti et al., 1995).

After a gigaseal was formed (seal resistance was usually >20 GΩ), electrode capacitance was carefully cancelled before the patch was ruptured to allow for the electronic compensation of pipette charging during subsequent current-clamp recordings (fastcurrent-clamp mode). Depending on the high input-to-series resistance ratio, bridge balancing in current-clamp recordings proved of little effect and was not used routinely (D’Angelo et al., 1995). Membrane potential was measured relative to an agar-bridge reference electrode. Reported membrane potential values have been adjusted off-line for liquid-junction potentials (usually <5 mV). Voltage-clamp recordings were used to record the inward rectifier current (see below) and to estimate the passive granule cell membrane properties. For this purpose, current transients were elicited by 10 mV hyperpolarizing pulses from a holding potential of −70 mV. The transients showed a monoexponential relaxation (time constant = 81 ± 27 μsec;n = 40) and were used to estimate series resistance (21.1 ± 8.7 MΩ; n = 40) and input resistance and capacitance.

The experimental tracings were analyzed using P-Clamp (Axon Instruments) and Origin (Microcal Software, Northampton, MA) software. Data are reported as mean ± SD, and statistical comparisons were performed using Student’s t test.

Recording and analysis of inward rectifier current. The inward rectifier current was activated with hyperpolarizing voltage pulses from a holding membrane potential of −40 mV. Alternatively,I–V relationships were constructed using voltage ramps. Transient current and leakage subtraction were performed either by using tracings recorded before GTP-γ-S action or tracings in which the inward rectifier current was blocked pharmacologically, as explained in Results. The I–V relationships were fitted with a modified Boltzmann equation (Hille and Schwarz, 1978) as follows:

$$I(V)=[G_{max}(V-V_{\text{rev}})]/[1+exp((V-V_{1/2})/k)]$$ Equation 1

where G_max is the maximal conductance,V_rev is the reversal potential,V_1/2 is the potential at which the current is half-activated, and k is the voltage dependence of activation (V⁻¹). OnceV_rev and G_max were known, and because G(V) =I(V)/(V −V_rev), activation curves of the form:

$$G/G_{max}=1/[1+exp((V-V_{1/2})/k)]$$ Equation 2

were plotted and used to compare inward rectifier voltage dependencies.

RESULTS

We investigated inward rectifier currents and their relationship with granule cell electroresponsiveness in wild-type and weaver mutant granule cells in acute mice cerebellar slices. Recordings were performed at different depths in the cerebellar cortex, allowing the electrophysiological properties to be correlated with neuronal developmental stage. Unless stated otherwise, the granule cells were recorded at P10–P15.

Electroresponsiveness in wild-type granule cells

In wild-type mice, at P10–P15 the germinal layer and the premigratory zone, constituting the external granular layer (EGL) of the cerebellar cortex, are clearly recognizable (Fig.1A) (Rakic and Sidman, 1973a). After migration, the granule cells populate the internal granular layer (IGL), become synaptically connected with mossy fibers, and complete their maturation. The electrophysiological properties of granule cells recorded at different depths are shown in Figure1B and Table 1.

Fig. 1.

Electroresponsiveness in developing wild-type granule cells. A, Schematic view of a cerebellar slice at P15 (after Rakic and Sidman, 1973a,b; Sotelo and Changeux, 1974). The indices a through c correspond to different positions in the cortex from which granule cells were recorded [a = germinal layer,b = premigratory zone, c1 = postmigratory zone (immature cells), c2 = postmigratory zone (mature cells)]. B, Positive and negative current pulses have been injected into granule cells recorded using patch pipettes containing GTP and maintained at the indicated resting potential with constant current injection. No inward rectification can be observed in granule cellsa–c1, whereas a marked nonsagging inward rectification (arrowhead) is expressed in the granule cell c2 (also see D’Angelo et al., 1995, 1997). This granule cell is synaptically connected, as demonstrated by the EPSP generated by mossy fiber stimulation (inset: Δ mossy fiber stimulation. Calibration: 10 mV, 100 msec). The current-clamp protocol is illustrated at the bottom. C, Voltage-current plot for the cells in B. Membrane potential has been averaged over the last 500 msec of the tracings.D, Voltage recordings from a granule cell of the premigratory zone using a patch pipette containing GTP-γ-S. Although the cell shows no inward rectification at the beginning of recording (t₀), inward rectification developed after 5 min (t₅). To compare the effect of injected current pulses, tracings have been recorded from an arbitrary membrane potential (−70 mV) maintained with constant current injection. Therefore the 14 mV hyperpolarization occurring with GTP-γ-S action is not seen in the figure. The bottom voltage–current plot reports membrane potentials averaged over the last 500 msec of the tracings, illustrating the induction of inward rectification by GTP-γ-S.

Table 1.

Electrophysiological properties of wild-type and weaver granule cells

		Cm(pF)	gleak(nS)	gleak/Cm (nS/pF)
Wild-type	a  (n  = 8)	3.5  ± 1.6	0.54  ± 0.18	0.15
	b  (n  = 6)	4.5  ± 0.9	0.3  ± 0.1	0.07
	c1  (n  = 7)	3.1  ± 1.1	0.3  ± 0.1	0.1
	c2  (n  = 7)	2.9  ± 0.9	0.6  ± 0.2	0.21
Weaver	awv  (n  = 8)	2.9  ± 0.5	0.5  ± 0.3	0.17
	bwv  (n  = 9)	4.6  ± 0.3	0.6  ± 0.4	0.13
	cwv  (n  = 6)	7.9  ± 2.5*	1.2  ± 1.1*	0.15

Electrophysiological properties of the different granule cell categories used in this paper (for definition, see Fig. 1A). Input membrane capacitance (C_m) has been estimated from passive current transients generated in the voltage-clamp mode just after the whole-cell configuration was established. Leakage conductance, g_leak, has been estimated as the slope of the I–V plots at potentials lower than −40 mV (in c2 and c_wv,g_leak was estimated after the inward rectifier current was blocked pharmacologically). Both C_mand g_leak were significantly greater in putative weaver granule cells in a deep position than in the other groups (*p < 0.05). Theg_leak/C_m ratio, however, was similar.

In an initial set of experiments we used patch pipettes containing 100 μm GTP (Fig. 1B). The granule cells in the germinal layer (a) and premigratory zone (b) of the EGL showed a rather high resting membrane potential (a, −24 ± 1.4 mV, n = 8; b, −22.9 ± 3, n = 11). These cells did not generate any action potential when they were injected with depolarizing currents, although the membrane had previously been hyperpolarized with a constant negative current. The injection of hyperpolarizing pulses did not reveal any inward rectification. The granule cells in the IGL showed either immature (c1) or mature (c2) electrical responses, which were distinguished on the basis of the nature of action potentials and the presence of inward rectification (D’Angelo et al., 1997). Immature responses consisted of slow spikes, absence of inward rectification, and high resting potential (c1, −36.3 ± 9.1 mV; n = 8). Mature responses consisted of fast spikes, inward rectification, and a negative resting membrane potential (c2 −57.5 ± 10.6 mV;n = 6). Similar granule cells were most commonly observed after P21 (seven of eight; data not shown), when migration is concluded and the EGL is no longer evident. TheV–I relationships shown in Figure 1Ccompare voltage response elicited by current injection into the granule cells, evidencing constitutive inward rectification in the granule cell that generates fast spikes.

When the nonhydrolyzable G-protein activator GTP-γ-S was included in the patch pipette, inward rectification could be induced in postmitotic granule cells that did not show it constitutively (Fig.1D). In these cells GTP-γ-S induction of inward rectification was associated with a significant lowering in resting membrane potential (b, −13 ± 3.7 mV,n = 9; c1, −16.1 ± 4.3 mV,n = 8; p < 0.02 in both cases). Inward rectification was manifest as a decrease in input resistance inV–I plots. Inward rectification extended well over the K⁺ reversal potential (−85.6 mV) into the resting membrane potential region, as would be expected from current voltage dependence measured in voltage-clamp experiments (see below). No significant inward rectification or membrane potential changes were observed in granule cells of the germinal layer [a, 4 ± 3.2 mV, n = 8; not significant (NS)], or in fast spiking granule cells of IGL (c2, 1.1 ± 3.2 mV,n = 7; NS). It should be noted that both the constitutive and GTP-γ-S dependent inward rectification didnot sag, indicating that the underlying current had activation kinetics that were faster than passive membrane charging.

G-protein modulation of the inward rectifier current in wild-type granule cells

The nature of the current generating inward rectification, and its modulation by G-proteins (Wickman and Clapham, 1995), has been investigated (Fig. 2A), and the time course of GTP-γ-S action has been compared with that of GTP (Fig. 2B).

Fig. 2.

GTP-γ-S modulation of the inward rectifier current in wild-type cerebellar granule cells. A, The recordings were performed with patch pipettes containing GTP-γ-S. Tracings were taken at the beginning (t₀) and 6 min after the beginning of recordings (t₆). The GTP-γ-S-sensitive current is shown in the subtraction tracings (t₆–t₀). Note that an inward rectifier current develops in granule cells of the premigratory zone (b; P12) and in immature postmigratory granule cells (c1; P12), but not in granule cells in the germinal layer (a; P13). Mature postmigratory granule cells showed a constitutive inward rectifier current, which slightly decreased with time (c2; P22). Note that an outward tail current usually developed on returning to −40 mV, probably reflecting I-A rebound activation. The voltage-clamp protocol is illustrated at the bottom. B, The plots report the time course of inward rectifier modulation with pipettes containing GTP-γ-S (•) compared with pipettes containing GTP (○). The recordings commenced at t = 0. The current is reported relative to control for membrane pulses from −40 mV to −120 mV. The data are reported as mean ± SD (n = 6–9 for each plot). Note that GTP-γ-S elicited an inward rectifier current inb and c1 but not in a. A slight decrease in the constitutive current was observed inc2 with both GTP-γ-S and GTP. Labelsa–c have the same meaning as in Figure1A.

(a) In the germinal layer, no inward rectifier current could be demonstrated, despite the presence of GTP-γ-S in the patch pipette (8 of 8).

(b) In the premigratory zone, the granule cells showed inward rectifier current expression during GTP-γ-S internal perfusion (G–I_IR; 9 of 10). No inward rectifier current was observed at the beginning of the recording or when patch pipettes containing GTP were used (9 of 9), indicating the absence of constitutive activation. The time course of GTP-γ-S action was similar to that reported previously in culture (Surmeier et al., 1996).

(c1) In the IGL, some of the granule cells (7 of 12) showed inward rectifier current modulation during GTP-γ-S internal perfusion (G–I_IR), but no inward rectifier current was measured at the beginning of the recording. Inward rectifier current modulation in these granule cells was therefore similar to that in cells in the premigratory zone. Accordingly, when we used patch pipettes containing GTP, no inward rectifier current was observed in the majority of granule cells (7 of 10), whereas a constitutive inward rectifier current was observed in the remaining granule cells (3 of 10; see point c2).

(c2) In the IGL, some of the granule cells showed a constitutive inward rectifier current (C–I_IR). In addition to being observed at the beginning of the recordings with pipettes containing GTP-γ-S (5 of 12),C–I_IR was also observed when patch pipettes containing GTP were used (3 of 10). At a more advanced developmental stage (P21–P22),C–I_IR was recorded in the majority of granule cells (seven of nine; data not shown).

These results indicate that the inward rectifier current follows a precise developmental time course in situ, leading from G-protein-dependent to constitutive activation.

Biophysical and pharmacological properties of the inward rectifier current in wild-type mice

G–I_IR and C–I_IRdisplayed similar pharmacological sensitivity (Fig.3A). These currents were completely blocked by 1 mm Ba²⁺(97.2 ± 6.1% G–I_IR block,n = 5; 98.6 ± 4.3% C–I_IRblock, n = 4) but were only partially blocked by 5 mm Cs⁺ (68.4 ± 25.4%G–I_IR block, n = 8; 58.6 ± 41.6% C–I_IR block, n = 4). Moreover, these currents were insensitive to 100 μmQX-314 (3.5 ± 2.2% G–I_IR block,n = 4; 16.6 ± 19.3%C–I_IR block, n = 4), a cationic channel blocker known to affect Na⁺channels as well as inward rectifier channels with partial Na⁺ permeability (Perkins and Wong, 1995; Kofuji et al., 1996; Slesinger et al., 1996). G–I_IR andC–I_IR were analyzed using tracings taken after pharmacological block with 1 mm Ba²⁺ for leakage and transient current subtraction (Fig. 3A).

Fig. 3.

Biophysical and pharmacological properties of the wild-type inward rectifier current. A1, A2, Both the current generated by GTP-γ-S (A1,G–I_IR) and that constitutively expressed (A2, C–I_IR) in granule cells were blocked by 1 mmBa²⁺. The Ba²⁺-sensitive currents have been obtained by subtraction (control − Ba). The voltage-clamp protocol is illustrated at the bottom. Theinset illustrates the insensitivity ofG–I_IR to 100 μm QX-314 and partial block by 1 mm Cs⁺ in two different cells (the tracings are leakage-subtracted, demonstrating the extent of block). B1, B2, Tracings obtained using the subtraction protocol shown in A have been used for kinetic analysis. G–I_IR (B1) was fitted with a double exponential function of the formi(t) = A1 × exp(−t/τ₁) +A2 × exp(−t/τ₂) + C. The individual exponential components as well as the biexponential function are superimposed on the experimental tracing (A1 = 40.2 pA; τ₁ = 1.1 msec; A2 = 16.1 pA; τ₂ = 18.5 msec; C = −42).C–I_IR (B2) was fitted with a single exponential function of the formi(t) = A1 × exp(−t/τ₁) + C. The monoexponential function is superimposed on the experimental tracing (A1 = 31.6 pA; τ₁ = 2.7 msec;C = −31.6 pA). The tracings in B1and B2 were generated by a voltage pulse from −40 mV to −120 mV. C1, C2, The voltage dependence of the fast inward rectifier current components ofG–I_IR (C1) orC–I_IR (C2) was obtained using a voltage-ramp protocol. The experimental points have been fitted to Equation 1 to yield the following parameters: (C1)G_max = 1.17 nS;V_rev = −79.66 mV;V_1/2 = −73.33 mV; k = 24.4 mV⁻¹; (C2)G_max= 2.78 nS;V_rev = −84.8 mV;V_1/2 = −82.4 mV; k = 8.94 mV⁻¹. The voltage-ramp protocol is indicated at the bottom.

G–I_IR did not show inactivation (Fig.3A1). G–I_IR activation comprised a fast and a slow component and could be fitted with a biexponential function (Fig. 3B1). The fast component had almost voltage-independent time constants in the submillisecond range and accounted for 84% of G–I_IR amplitude at −120 mV. The slow component had time constants in the 100 msec range and was better resolved at potentials lower than −100 mV. The fast component therefore accounted for most of the G–I_IR in the physiological range of membrane potentials and should correspond to the “instantaneous” rectification reported in other papers (Kofuji et al., 1996; Surmeier et al., 1996). The slow component may itself be composite (Doupnik et al., 1995), although no different subcomponents were detectable in the time scale used here. The voltage dependence of the fast G–I_IR component was investigated using a voltage-ramp protocol (Fig. 3C1). TheI–V relationship was fitted with Equation 1, yielding the Boltzmann parameters V_1/2 andk as well as maximum conductance (G_max) and reversal potential (V_rev) (Table2). Because of its (1) reversal potential close to that of K⁺ ions, (2) fast kinetics, (3) high sensitivity to Ba²⁺ block, and (4) modulation by G-proteins, G–I_IR appeared as the native counterpart of GIRK2-containing channels measured in reconstituted systems (Duprat et al., 1995; Kofuji et al., 1995; Krapivinsky et al., 1995; Silverman et al., 1996; Slesinger et al., 1996). A similar current has been reported in mice granule cells in culture (Surmeier et al., 1996).

Table 2.

Gating properties of inward rectifier currents in wild-type and weaver granule cells

	Gmax(nS)	Vrev (mV)	V1/2(mV)	k (mV−1)
G–IIR (n = 8)	2.6  ± 1.6	−80.5  ± 6.6*	−81.9  ± 8.3	25.1  ± 5.7**
C–IIR (n = 7)	1.4  ± 1.2	−84.2  ± 7.7**	−86.3  ± 7.2	8.1  ± 0.7*
wv–IIR (n = 6)	4.3  ± 2.6	−64.1  ± 6.3	−71.6  ± 1.9	11.6  ± 1.4

Average parameter values characterizing inward rectifier current activation were obtained using Equation 1. The reversal potential (V_rev) was significantly less negative inwv–I_IR than in eitherC–I_IR (**p < 0.01) orG–I_IR (**p < 0.002). The voltage-dependence of activation (k) was significantly different in wv–I_IR as compared withC–I_IR (*p < 0.01) andG–I_IR (**p < 0.001). Once the maximum conductance (G_max) was normalized for the membrane capacitance, the sequence G–I_IR(0.68 ± 0.4 nS/pF) > wv–I_IR (0.56 ± 0.3 nS/pF) > C–I_IR (0.41 ± 0.33 nS/pF) was obtained, with wv–I_IR proving to be statistically indistinguishable from C–I_IR orG–I_IR.

C–I_IR is shown in Figure 3A2.C–I_IR activated monoexponentially with time constants in the millisecond range, whereas no slow components could be detected (Fig. 3B2). The voltage dependence ofC–I_IR was investigated using a voltage-ramp protocol (Fig. 3C2). The results of an analysis similar to that used for G–I_IR are reported in Table 2. No such constitutive fast inward rectifier current has been reported previously in granule cells in culture, although its existence was suggested by voltage recordings in slice preparations (D’Angelo et al., 1995, 1997).

Table 2 and Figure 4 condensed data on the biophysical properties of G–I_IR andC–I_IR. C–I_IR had steeper voltage dependence than G–I_IR, and an activation time constant that was slower than the fast component of G–I_IR. Together with constitutive activation and the absence of a slow component, these were the major properties ofC–I_IR that distinguished it fromG–I_IR.

Fig. 4.

Average kinetics of wild-type inward rectifier currents. A, Average I–Vplots of the fast inward rectifier current components ofG–I_IR and C–I_IRhave been reconstructed using Equation 1 and the average values reported in Table 2. B, Normalized activation curves ofG–I_IR and C–I_IRhave been reconstructed using Equation 2. C, Voltage dependence of the fast activation time constant ofG–I_IR and of the activation time constant of C–I_IR. A–C show that activation kinetics are faster but less voltage dependent inC–I_IR than inG–I_IR.

Electroresponsiveness in weaver granule cells

In weaver mutant mice, cerebellar architecture is profoundly disrupted (Fig. 5A) (Rakic and Sidman, 1973a; Sotelo and Changeux, 1974). Below the germinal layer, large neurons probably corresponding to ectopic Purkinje cells were surrounded by groups of granule cells, and no IGL could be identified. The small neurons beside or below Purkinje cells in a deep position should correspond to granule cells “which have achieved a complete migration reaching their normal position and developing their normal connections” (Sotelo and Changeux, 1974), although no direct differentiation vis-à-vis other neuronal types (stellate or basket cells) could be achieved in the present experiments.

Fig. 5.

Electroresponsiveness in developing weaver granule cells. A, Schematic view of the homozygous weaver cerebellum (after Rakic and Sidman, 1973a,b; Sotelo and Changeux, 1974). The indicesa_wv–c_wvcorrespond to different positions in the cortex, as illustrated in thetop inset (a_wv = germinal layer, b_wv = premigratory zone,c_wv = deep layer). B, Positive and negative current pulses have been injected in granule cells recorded in P10–P15 acute slices of the cerebellar vermis using patch pipettes containing GTP, and maintained at the indicated resting potential with constant current injection. Note that no inward rectification can be observed in the granule cellsa_wv–b_wv, whereas marked sagging inward rectification (arrowhead) is expressed in the granule cell c_wv (also see D’Angelo et al., 1995, 1997). This granule cell is synaptically connected, as demonstrated by the EPSP generated by mossy fiber stimulation (inset: Δ mossy fiber stimulation. Calibration: 10 mV, 100 msec). The current-clamp protocol is illustrated at the bottom. C, Voltage-current plot for the cells in B. Membrane potential has been averaged over the last 500 msec of the tracings. D, Voltage recordings from a granule cell of the premigratory zone using a patch pipette containing GTP-γ-S. Unlike wild-type granule cells, no inward rectification developed after 5 min of GTP-γ-S internal perfusion (t₆ as compared witht₀), nor was membrane hyperpolarization measured (the tracings have been recorded from an arbitrary membrane potential of −70 mV). The voltage–current plot at the bottom reports membrane potentials averaged over the last 500 msec of the tracings, illustrating the absence of inward rectification induction.

The granule cells displayed different electrophysiological behavior depending on their location, as shown in Figure 5B and reported in Table 1 for slices obtained at P10–P15. When patch pipettes containing GTP were used, the granule cells in the germinal layer (a_wv) and in the premigratory zone (b_wv, recorded at 50–100 μm from the pial surface) appeared similar to those of the wild type. They had a rather high resting membrane potential (a_wv, −22 ± 7.2 mV,n = 8; b_wv, −32 ± 2.7, n = 10) and did not generate either action potentials or inward rectification during current injection. Some of the granular neurons in a deeper position (c_wv, recorded at 100–200 μm from the pial surface), however, had a more negative resting membrane potential (−57.6 ± 5.4 mV, n = 6), generated action potentials (in three neurons a solitary fast spike was measured, whereas the other three neurons generated a fast repetitive spike discharge), and showed sagging inward rectification. These neurons therefore exhibited a resting membrane potential and spike discharge comparable to those of wild-type granule cells at an advanced developmental stage in the IGL, although the nature of the inward rectification was different (compare Fig. 1A, panelC2). These neurons could be activated by afferent fiber stimulation with a time delay of 1.5 msec for EPSP activation (Fig.5B, inset). These neurons therefore are putative postmigratory granule cells that developed until they formed functional synapses. The V–I relationships in Figure5B show that subthreshold voltage responses lacked inward rectification in premigratory weaver granule cells, whereas a constitutive inward rectification was evident in putative postmigratory granule cells.

A major difference between weaver and wild-type granule cells was revealed using patch pipettes containing GTP-γ-S. Despite the apparent similarity of their electrical response to the wild-type cells, the weaver granule cells showed neither induction of inward rectification (Fig. 5D) nor hyperpolarization from rest during GTP-γ-S perfusion (a_wv −5.1 ± 4.9 mV, n = 8; b_wv 0 ± 3.3 mV, n = 10; c_wv 5.6 ± 3.2 mV, n = 6; NS in all cases).

G-protein modulation of the inward rectifier current in weaver granule cells

As in wild-type granule cells, the nature of the inward rectifier current in weaver mutant mice, and its G-protein sensitivity, were investigated (Fig.6A), and the time course of GTP-γ-S action was compared with that of GTP (Fig.6B).

Fig. 6.

GTP-γ-S modulation of the inward rectifier current in weaver cerebellar granule cells. A, The recordings were performed with patch pipettes containing GTP-γ-S. Tracings were taken at the beginning (t₀) and 6 min after the beginning of recordings (t₆). The GTP-γ-S-sensitive current is shown in the subtraction tracings (t₆–t₀). Note that no inward rectifier current develops in either granule cells in the premigratory zone (b_wv) or those in the germinal layer (a_wv). Granule cells in a deep position showed a constitutive inward rectifier current, which slightly decreased with time (c_wv; note the different scale than in a_wv,b_wv). The voltage-clamp protocol is illustrated at the bottom. B, The plots report the time course of inward rectifier modulation with GTP-γ-S-containing pipettes (•) compared with GTP-containing pipettes (○). The recordings commenced at t = 0. The current is reported relative to control for membrane pulses from −40 mV to −120 mV. The data are reported as mean ± SD (n = 5–10 for each plot). Note that neither GTP-γ-S nor GTP elicited any inward rectifier current ina_wv or b_wv. A slight decrease in the constitutive current was observed inc_wv with either GTP-γ-S or GTP. Labelsa_wv–c_wvhave the same meanings as in Figure 5A.

(a_wv) As in wild-type mice, granule cells in the germinal layer did not show any inward rectifier currents despite the presence of GTP-γ-S in the patch pipette (10 of 10).

(b_wv) Unlike in wild-type mice, granule cells in the premigratory zone did not show any inward rectifier currents, despite the presence of GTP-γ-S in the patch pipette (14 of 14). These granule cells were interesting, because they may be cells in which mutation has caused a loss of G-protein sensitivity in the inward rectifier current (Surmeier et al., 1996). The weaver currents therefore differed from wild-type inward rectifier currents in their G-protein-dependent modulation.

(c_wv) Putative postmigratory granule cells showed a constitutive inward rectifier current with anomalous properties, wv–I_IR (6 of 7) (Fig.6C). In addition to being observed at the beginning of the recordings in the presence of GTP-γ-S,wv–I_IR was observed also when patch pipettes containing GTP were used (7 of 7).

Biophysical and pharmacological properties of the inward rectifier current in putative weaver granule cells

The weaver inward rectifier current,wv–I_IR, showed distinctive pharmacological properties. Unlike the wild-type inward rectifiers,wv–I_IR was inhibited by 100 μm QX-314 (60.6 ± 19.3% block; n = 5) (Fig. 7A). Sensitivity to external QX-314 at low concentrations is typical of the homomultimericwv–GIRK2 channels expressed in oocytes (Kofuji et al., 1996; Slesinger et al., 1996). The residual current was then blocked by the addition of 0.5 mm Cs⁺ (98 ± 6% block; n = 5) (Fig. 7A). The other major inward rectifier channel sensitive to submillimolar Cs⁺, I_h, has never been reported to be sensitive to submillimolar external QX-314 (although it can be blocked by 10 mm internal QX-314) (Perkins and Wong, 1995). An additional four neurons that expressed an inward rectifier current that was insensitive to QX-314 but was readily blocked by 0.5 mm Cs⁺ have not been considered as wv–I_IR and have been discarded from the present analysis. These neurons had a significantly higher membrane capacitance thanwv–I_IR-expressing neurons (15.4 ± 3.2 pF, n = 4, vs 7.9 ± 2.5 pF,n = 6; p < 0.01), were autorhythmic, and generated an initial burst followed by a steady discharge during current injection and were probably Purkinje cells.

Fig. 7.

Biophysical and pharmacological properties of the weaver inward rectifier current. A,wv–I_IR was inhibited by 100 μm QX-314 and completely blocked by subsequent addition of 0.5 mm Cs⁺. The QX-314– and QX-314 Cs⁺-sensitive currents have been obtained by subtraction (control − QX-314, control − QX-314/Cs⁺). The voltage-clamp protocol is illustrated at the bottom. B, QX-314 Cs⁺-sensitivewv–I_IR has been used for kinetic analysis. wv–I_IR was fitted to a single exponential function of the formi(t) = A1 × exp(−t/τ₁) + C. The monoexponential function is superimposed on the experimental tracing (A1 =184 pA; τ₁ =41.84 msec;C = −203.6 pA). The tracing was generated by a voltage pulse from −40 mV to −120 mV. C, Because of its slow kinetics, the voltage dependence ofwv–I_IR was obtained using steady-state measurements from the QX-314–Cs⁺-sensitive current (seeA). Experimental points have been fitted to Equation 1to yield the following parameters: G_max = 6.88 nS; V_rev = −61 mV;V_1/2 = −79.3 mV; k = 9.5 mV⁻¹.

wv–I_IR was analyzed using tracings obtained after pharmacological block with Cs⁺ for leakage and transient current subtraction after QX-314 sensitivity had been verified (n = 6) (compare Fig. 7A).wv–I_IR had slow kinetics, activating with time constants of tens to hundreds of milliseconds (Fig.7B), and no fast components could be detected. Voltage dependence was analyzed using Equation 1 (Fig.8), and average parameter values are reported in Table 2. The wv–I_IRreversal potential was ∼20 mV more positive (−64.1 ± 6.3 mV;n = 4) than C–I_IR orG–I_IR, consistent with the reduced selectivity of wv–GIRK2 homomultimers for K⁺ ions (Navarro et al., 1996; Kofuji et al., 1996;Slesinger et al., 1996). wv–I_IR is unlikely to contribute significantly to depolarizing putative granule cells, however, because its activation voltage dependence was steep and confined to negative membrane potentials. Moreover, unlike expression systems, there was no net increase in leakage current in putative weaver granule cells compared with wild-type granule cells (Table1).

Fig. 8.

Average kinetics of the weaver inward rectifier currents. A, Average I–Vplots of wv–I_IR have been reconstructed using Equation 1 and the average values reported in Table2. B, A normalized activation curve ofwv–I_IR was reconstructed using Equation 2. The activation curves ofC–I_IR and G–I_IRare replotted for comparison (dotted lines).C, Voltage dependence of the fast activation time constant of wv–I_IR. Note that the activation time constants are one to two orders of magnitude larger than those of C–I_IR andG–I_IR (compare Fig. 4C).A–C show thatwv–I_IR activation kinetics differs from those of C–I_IR andG–I_IR.

DISCUSSION

It has been proposed that cerebellar granule cell degeneration in the weaver mutation is induced by either the absence of G-protein modulation or altered biophysical properties of an inward rectifier current (Kofuji et al., 1996; Silverman et al., 1996;Slesinger et al., 1996; Surmeier et al., 1996). In this paper we report that the expression of inward rectifier currents and their G-protein modulation in granule cells of the mice cerebellum are developmentally regulated in situ. In the weaver mutation, the absence of G-protein sensitivity at the premigratory stage appears to be the primary process in the pathogenesis of the mutated phenotype. The possibility that abnormal channel properties are manifest at more advanced developmental stages in surviving granule cells is discussed.

Development of inward rectifier currents in wild-type mice

We found the earliest functional evidence for inward rectifier current expression in postmitotic premigratory granule cells. In these cells the inward rectifier current was strictly G-protein-dependent (G–I_IR). G–I_IRwas also measured in postmigratory granule cells in the first 3 postnatal weeks. In these cells, however, other ionic currents also differed from those in premigratory granule cells (D’Angelo et al., 1994, 1997), allowing action potentials to be generated. In mature granule cells, which generate fast action potentials, a constitutive inward rectifier current was found (C–I_IR). This may indicate a change in either channel properties or G-protein modulation. On the one hand, the higher voltage sensitivity and the lack of a slow current component inC–I_IR compared withG–I_IR suggest a change in channel properties. GIRK channel properties arise from the assembly of different subunits (Duprat et al., 1995; Kofuji et al., 1995; Krapivinsky et al., 1995), and the expression of slow kinetics depends on the insertion of GIRK1 subunits into GIRK2-containing channels (Slesinger et al., 1996). Therefore a developmental regulation of channel subunit expression may contribute to transforming G–I_IR intoC–I_IR. On the other hand, it seems unlikely that constitutive C–I_IR activation results from a G_βγ-independent channel regulation by intracellular Na⁺ (Silverman et al., 1996), because the pipette solution contained a low intracellular Na⁺ concentration (4 mm) in granule cells expressing both G–I_IR andC–I_IR. However, we cannot rule out that changes in G-protein-coupled receptor activation regulate inward rectifier channels (Wickman and Clapham, 1995). Among several neurotransmitters and modulators, glutamate has been shown to regulate granule cell GIRK channels by activating metabotropic receptors (Surmeier et al., 1996).

G–I_IR and C–I_IR were probably both involved in setting the granule cell membrane potential, although in quite different ways. G–I_IRactivation extended into the high-voltage range, being well suited to hyperpolarizing the membrane of immature granule cells from its quiescent state at around −30 mV in response to exogenous modulatory actions. Conversely, C–I_IR was constitutively active in a narrow negative voltage range, tending to clamp the membrane within a negative potential region. In addition to changes in the inward rectifiers, other factors may contribute to stabilizing the resting membrane potential at more negative values in mature rather than immature granule cells, including a lowering in the GABA reversal potential (Brickley et al., 1996) and increased activation of K⁺ currents under muscarinic receptor control (Watkins and Mathie, 1996).

Inward rectifier currents in weaver mice

Unlike the wild-type cells, premigratory weaver granule cells did not show any inward rectifier currents induced by G-protein stimulation. This probably indicates a loss of G-protein modulation rather than a lack of inward rectifier channel expression, as suggested by the observation that GIRK1, which is expressed in the immature cerebellum, inhibits wv–GIRK2-containing channels (Navarro et al., 1996; Slesinger et al., 1996). A loss of G-protein-dependent inward rectifier currents has also been reported in weaver granule cell cultures by Surmeier et al. (1996). It should be noted that no increase in the leakage current was detectable in our experiments (Table 1), as in those of Surmeier et al. (1996) but unlike those of Kofuji et al. (1996).

By recording from putative granule cells that escaped degeneration at the premigratory stage (Sotelo and Changeux, 1974), we found an inward rectifier current with slow kinetics dubbedwv–I_IR that showed constitutive activation and a marked reduction after the extracellular application of low concentrations of QX-314. As well as being sensitive to QX-314, wv–I_IR showed a positive shift in its ionic reversal potential as compared with wild-type currents, consistent with reduced selectivity for K⁺in favor of Na⁺ and Ca²⁺ ions. These properties are compatible with the formation ofwv–GIRK2 homomultimers (Navarro et al., 1996; Kofuji et al., 1996; Slesinger et al., 1996). However, the identification ofwv–I_IR as the native counterpart ofwv–GIRK2-containing channels is not definitive. First, unlike wv–GIRK2 homomultimers, the neurons expressingwv–I_IR did not show any net increase in leakage current (Table 1). Second,wv–I_IR shares withI_h submillimolar Cs⁺sensitivity and slow kinetics associated with sagging voltage responses to step current injection. Although thewv–I_IR reversal potential was not as high as expected from I_h, the distinction based on QX-314 sensitivity cannot be considered as absolute, because a high intracellular QX-314 concentration blocked neuronalI_h (Perkins and Wong, 1995). Finally, although neurons expressing wv–I_IR did not show intrinsic firing discharge or complex spikes, as occurs inI_h-expressing Purkinje neurons (Crepel and Penit-Soria, 1986), it is possible that neurons like stellate or basket cells, which have unknown inward rectifier currents, could also have been recorded. Therefore, a direct identification of cell types associated with an extended analysis of inward rectifier currents in cerebellar neurons is needed to confirm the suggestion thatwv–I_IR is the native counterpart of mutated wv–GIRK2-containing channels.

Pathogenetic implications

The most important finding in seeking to identify the pathogenetic role of the GIRK mutation (Patil et al., 1995) is that noinward rectifier currents could be induced by G-protein stimulation in weaver premigratory granule cells. These cells cannot use their inward rectifier current (G–I_IR), which is presumably under the control of specific neuromodulators (Surmeier et al., 1996), to hyperpolarize the membrane from rest. A failure to regulate membrane potential may cause uncontrolled activation of voltage-sensitive N-type Ca²⁺ channels and NMDA receptors, which in normal conditions drive granule cell motility and differentiation (Komuro and Rakic, 1993a,b), leading to Ca²⁺ accumulation, neurotoxicity, and death (Garthwaite, 1994; Choi, 1995). The role of NMDA receptors is particularly interesting, because they are stimulated by ambient glutamate in the EGL (Slater and Rossi, 1996) and membrane depolarization unblocks the channel, causing Ca²⁺permeation. The absence of G–I_IR activation seems likely to be the major pathogenetic mechanism, because most granule cells degenerate when they are postmitotic in the premigratory region of the EGL (Smeyne and Goldowitz, 1989; Surmeier et al., 1996).

The few granule cells that pass the premigratory stage (Rakic and Sidman, 1973b; Sotelo and Changeux, 1974) might suffer further neurodegenerative damage. To the extent thatwv–I_IR is the native counterpart of the mutated GIRK-2 channel expressed in oocytes, it might promote further neurodegeneration through an anomalous Na⁺and Ca²⁺ influx (Navarro et al., 1996; Silverman et al., 1996). We noted, however, that the reversal potential remained rather negative and the voltage dependence remained steep, so thatwv–I_IR activation was confined to low potentials. Consistently, wv–I_IRwas not associated with any marked resting depolarization. We would also point out that constitutivewv–I_IR activation was not specifically related to the pathogenetic process, because a similar property was typical of IGL wild-type granule cells (also see D’Angelo et al., 1995, 1997).

Conclusions

The inward rectifier in cerebellar granule cells is developmentally regulated, changing from G-protein-dependent to constitutive activation as the granule cells develop from premigratory to postmigratory stages. The G-protein-dependent form probably serves to hyperpolarize the granule cell membrane under the control of specific neurotransmitters and hormones (Kofuji et al., 1996; Surmeier et al., 1996). The failure of inward rectifier G-protein activation identifies an important correlate of the weaver mutation, localizing its functional expression at the premigratory stage. This is probably just the first step in a more complex neurodegenerative process (Hess, 1996), because granule cell development involves cell–cell interactions and the weaver mutation acts nonautonomously (Gao et al., 1992; Gao and Hatten, 1993; Hatten and Heintz, 1995). An altered constitutive inward rectifier current may be expressed in weaver granule cells that develop mature excitable properties (Kofuji et al., 1996; Silverman et al., 1996; Slesinger et al., 1996), causing further cellular damage in a two-stage pathogenetic process. Our results support the concept that developmental regulation of GIRK channel subunit expression, and modulation of inward rectifier channel activation, may provide important signals for neuronal survival and differentiation.