Redistribution of GABAB(1) protein and atypical GABAB responses in GABAB(2)-deficient mice

Abstract

GABAB receptors mediate slow synaptic inhibition in the nervous system. In transfected cells, functional GABAB receptors are usually only observed after coexpression of GABAB(1) and GABAB(2) subunits, which established the concept of heteromerization for G-protein-coupled receptors. In the heteromeric receptor, GABAB(1) is responsible for binding of GABA, whereas GABAB(2) is necessary for surface trafficking and G-protein coupling. Consistent with these in vitro observations, the GABAB(1) subunit is also essential for all GABAB signaling in vivo. Mice lacking the GABAB(1) subunit do not exhibit detectable electrophysiological, biochemical, or behavioral responses to GABAB agonists. However, GABAB(1) exhibits a broader cellular expression pattern than GABAB(2), suggesting that GABAB(1) could be functional in the absence of GABAB(2). We now generated GABAB(2)-deficient mice to analyze whether GABAB(1) has the potential to signal without GABAB(2) in neurons. We show that GABAB(2)-/- mice suffer from spontaneous seizures, hyperalgesia, hyperlocomotor activity, and severe memory impairment, analogous to GABAB(1)-/- mice. This clearly demonstrates that the lack of heteromeric GABAB(1,2) receptors underlies these phenotypes. To our surprise and in contrast to GABAB(1)-/- mice, we still detect atypical electrophysiological GABAB responses in hippocampal slices of GABAB(2)-/- mice. Furthermore, in the absence of GABAB(2), the GABAB(1) protein relocates from distal neuronal sites to the soma and proximal dendrites. Our data suggest that association of GABAB(2) with GABAB(1) is essential for receptor localization in distal processes but is not absolutely necessary for signaling. It is therefore possible that functional GABAB receptors exist in neurons that naturally lack GABAB(2) subunits.

Figure 1.

Characterization of GABA_B(2)^-/- mice. A, Top, GABA_B(2) locus encompassing exons 8-11, encoding part of the N-terminal and the transmembrane (TM) domains 1 and 2. Bottom, GABA_B(2) allele after homologous recombination with a targeting construct containing a neomycin resistance cassette (neo) flanked by 4.5 and 1.8 kb of genomic DNA (bold lines). Exons 9 and 10 (3.5 kb, black boxes) are deleted. The Southern blot probe used in B is indicated. H, HindIII; E, EcoRI; B, BamHI. B, Southern blot analysis of EcoRI cut genomic DNA from wild-type (+/+), GABA_B(2)^+/- (+/-), and GABA_B(2)^-/- (-/-) mice. C, Top, Northern blot analysis of total brain RNA hybridized with GABA_B(2) cDNA probes upstream (5′ probe) and downstream of the deletion (3′ probe). The probes hybridize to a band just above the 28 S ribosomal RNA (arrow) in wild-type and GABA_B(2)^+/-, but not in GABA_B(2)^-/-, mice. Bottom, Blots from top panels stained with methylene blue, demonstrating equal loading of RNA. Ribosomal RNA bands (18 and 28 S) are labeled. D, In situ hybridization analysis of GABA_B(1) (1, top) and GABA_B(2) (2, bottom) transcripts of sagittal sections from adult wild-type and GABA_B(2)^-/- brains. E, Immunoblot analysis of brain extracts from adult mice using antibodies directed against C-terminal and N-terminal epitopes of GABA_B(2) and GABA_B(1). Antibodies to PSD-95 control for equal loading. GABA_B(1a) (1a) and GABA_B(1b) (1b) proteins are indicated. F, Immunoblot analysis demonstrating the presence of GABA_B(1) protein in synaptic plasma membranes (SPM) purified from the P2 pellet (P2) of brain extracts of wild-type and GABA_B(2)^-/- mice. Antibodies to calreticulin show that the synaptic plasma membrane fraction is free of ER proteins. Equal loading of samples was controlled with anti-syntaxin antibodies. To detect putative truncated GABA_B(2) proteins, we used 15% SDS-PAGE and N-terminal GABA_B(2) antibodies. In all other immunoblot experiments, we used 10% SDS-PAGE.

Figure 2.

Alteration of GABA_B receptor-IR in GABA_B(2)^-/- brains. A, Effect of GABA_B(2) gene deletion on the distribution of GABA_B(2)-IR (2) and GABA_B(1)-IR (1), as visualized in color-coded parasagittal sections from adult wild-type (+/+), GABA_B(2)^+/- (+/-), and GABA_B(2)^-/- (-/-) mice. The color scale is indicated. The reduced expression of GABA_B(2) in GABA_B(2)^+/- mice and the complete loss of expression in GABA_B(2)^-/- mice are evident throughout the brain (top). GABA_B(1)-IR is retained in GABA_B(2)^+/- mice and partly reduced in GABA_B(2)^-/- mice, in which it exhibits an altered cellular distribution, as seen in the hippocampus (bottom). The residual GABA_B(1)-IR in GABA_B(2)^-/- mice is not caused by nonspecific binding of the secondary antibodies, which are the same for GABA_B(1) and GABA_B(2). The specificity of the GABA_B(1) antiserum was also tested in GABA_B(1)^-/- mice, in which no specific staining was observed (data not shown). B, Color photomicrographs of the hippocampal formation stained for GABA_B(1) in adult wild-type and GABA_B(2)^-/- mice. The pronounced increase of IR in the CA1-CA3 pyramidal cell layer and in the dentate gyrus granule cell layer (DG) contrasts with the strong reduction in the dendritic layers [stratum oriens (so), stratum radiatum (sr), stratum lucidum (sl), and molecular layer (ml)]. C, Enlargement of the framed areas in B. Numerous interneurons, which are primarily hidden in sections from wild-type mice because of the homogeneous staining, appear more strongly labeled in GABA_B(2)^-/- mice but with a normal distribution and morphology. Scale bars: A, 2 mm; B, 200 μm.

Figure 6.

Baclofen inhibits a postsynaptic K⁺ conductance in CA1 pyramidal cells of GABA_B(2)^-/- mice. A, Holding current (at -50 mV) plotted versus time for wild-type (top, filled circles) and GABA_B(2)^-/- (bottom, open circles) mice. Whereas both baclofen (50 μm) and adenosine (100 μm) induce an outward current in wild-type mice, baclofen induces an inward current in GABA_B(2)^-/- mice. Baclofen-induced effects were blocked by application of the GABA_B receptor antagonist CGP55845A (2 μm) in wild-type as well as in GABA_B(2)^-/- mice. B, Summary graph illustrating the baclofen-induced inward current at -50 mV in GABA_B(2)^-/- mice. Baclofen-induced currents: wild-type, n = 5; GABA_B(2)^-/-, n = 9. Adenosine-induced currents: wild-type, n = 5; GABA_B(2)^-/-, n = 4. C, Current-voltage relationship of the baclofen-induced conductance in wild-type (black trace) and GABA_B(2)^-/- (gray trace) mice. Currents were obtained by calculating the difference between the I-V curves before and after addition of baclofen. Whereas a current with a positive slope conductance is induced by baclofen in wild-type mice, a current with negative slope conductance is induced in GABA_B(2)^-/- mice. D, Current-voltage relationship of the adenosine-induced conductance in wild-type mice (black trace) is not different from GABA_B(2)^-/- mice (gray trace). E, Baclofen induces the closure of K⁺ channels in GABA_B(2)^-/- mice. Raising extracellular [K⁺] concentration shifts the reversal potential of the baclofen-induced current. F, The baclofen-induced conductance change is mediated by G-protein activation. In the presence of intracellular GDPβS (1 mm for 25 min), both the baclofen-induced (control, n = 5; GDPβS, n = 5) and adenosine-induced (control, n = 4; GDPβS, n = 5) currents are inhibited in GABA_B(2)^-/- mice. G, Changes in the holding current (at -50 mV) in response to baclofen (Bacl.) after preincubation with adenosine. H, Summary graph illustrating that the effects of adenosine and baclofen are not fully additive. In wild-type neurons (+/+), the effect of a combined application of adenosine and baclofen is lower than the sum of the individual effects [Adenosine + Bacl. (calculated)]. In GABA_B(2)^-/- (-/-) neurons, the effects of adenosine and baclofen are not fully additive. Application of baclofen does not obliterate the adenosine response. I, Left, Summary graph illustrating postsynaptic conductance changes induced by baclofen in wild-type (n = 5), GABA_B(2)^+/- (n = 10), and GABA_B(1)^-/- (n = 4) mice. The conductance changes were blocked by application of the GABA_B(1) receptor antagonist CGP55845A (2 μm; wild-type, n = 4; GABA_B(2)^-/-, n = 8). Adenosine-induced conductance changes are not different between genotypes (wild-type, n = 4; GABA_B(2)^-/-, n = 4; GABA_B(1)^-/-, n = 3). ^*p < 0.05; ^**p < 0.01.

Figure 3.

GABA_B(1) binding sites in GABA_B(2)^-/- brains. A, Saturation isotherms for [¹²⁵I]CGP64213 antagonist binding to cortex membranes. No significant binding is detected in membranes from GABA_B(2)^-/- mice. The number of binding sites is reduced in GABA_B(2)^+/- versus wild-type mice. The maximal number of binding sites (B_max) for wild-type and GABA_B(2)^+/- mice are 1.4 ± 0.12 and 0.7 ± 0.05 pmol/mg protein, respectively; K_d values were 1.1 ± 0.06 and 0.9 ± 0.05 nm, respectively (mean ± SEM; n = 3). B, Autoradiograms of brain extracts from wild-type (+/+), GABA_B(2)^+/- (+/-), and GABA_B(2)^-/- (-/-) mice, labeled with the photoaffinity antagonist [¹²⁵I]CGP71872 (0.5 nm) and analyzed by SDS-PAGE. Exposure for 8 d (8d exp.) reveals low amounts of labeled GABA_B(1a) (1a) and GABA_B(1b) (1b) proteins in GABA_B(2)^-/- brains. C, GABA_B(1) subunit autoradiography. Sagittal cryostat sections were incubated with the GABA_B antagonist [³H]CGP62349. Nonspecific binding was determined in the presence of an excess of 100 μm unlabeled l-baclofen. Tritium-sensitive x-ray films were exposed for 24 hr and developed using a Cyclone Storage Phosphor screen (PerkinElmer Life Sciences, Boston, MA). D, Quantitative analysis of [³H]CGP62349 receptor autoradiography. Individual brain regions (n = 3) were counted using the MCID software package (Imaging Research, St. Catharines, Ontario, Canada). The differences in radioligand binding between the three genotypes are significant (two-sided Dunnett test; p < 0.001 for combined analysis of all brain regions). A-C, Representative experiments, which were repeated three times.

Figure 4.

[³⁵S]GTPγS binding to cortex membranes. No significant GABA-stimulated (filled symbols, filled lines) or baclofen-stimulated (Bac; open symbols, dotted lines) [³⁵S] GTPγS binding is detected in GABA_B(2)^-/- membranes. [³⁵S]GTPγS binding to membranes from GABA_B(2)^+/- mice is significantly reduced compared with wild-type mice. Values are normalized to the maximal response obtained with wild-type mice.

Figure 5.

Lack of baclofen-induced presynaptic inhibition in CA1 pyramidal cells of GABA_B(2)^-/- mice. A, Excitatory synaptic transmission. Monosynaptic EPSC peak amplitudes plotted versus time and representative traces from wild-type (top, filled circles) and GABA_B(2)^-/- (bottom, open circles) mice. Both baclofen (50 μm) and adenosine (100 μm) significantly depress the EPSC amplitude in wild-type mice, whereas baclofen and CGP55845A (2 μm) have no effect on the EPSC amplitude in GABA_B(2)^-/- mice. The effect of adenosine is similar in both genotypes. Traces are averages of 10 consecutive sweeps. Calibration: 40 msec, 100 pA. B, Summary graph showing the lack of baclofen-induced presynaptic inhibition of excitatory synaptic transmission in GABA_B(2)^-/- mice (wild-type, n = 4; GABA_B(2)^-/-, n = 8). Adenosine-induced inhibition is similar in both genotypes (wild-type, n = 4; GABA_B(2)^-/-, n = 6). C, Inhibitory synaptic transmission. Monosynaptic IPSC peak amplitudes plotted versus time and representative traces from wild-type (top, filled circles) and GABA_B(2)^-/- (bottom, open circles) mice. Both baclofen (50 μm) and the μ-opioid agonist DAMGO (1 μm) significantly depress the IPSC amplitude in wild-type mice, whereas baclofen and CGP55845A (2 μm) have no effect on the IPSC amplitude in GABA_B(2)^-/- mice. The effect of DAMGO was similar in both genotypes. Traces are averages of 10 consecutive sweeps. Calibration: 100 msec, 200 pA. D, Summary graph showing the lack of baclofen-induced presynaptic inhibition of inhibitory synaptic transmission in GABA_B(2)^-/- mice (wild-type, n = 7; GABA_B(2)^-/-, n = 6). DAMGO-induced inhibition was similar in both genotypes (wild-type, n = 7; GABA_B(2)^-/-, n = 5). ^**p < 0.01; ^***p < 0.001.

Figure 7.

Lack of baclofen-induced delta waves in GABA_B(2)^-/- mice. A, Effect of l-baclofen (10 mg/kg, i.p.) on the EEG of freely moving wild-type (+/+) and GABA_B(2)^-/- (-/-) mice. The EEG of wild-type and GABA_B(2)^-/- mice were similar 10 min before baclofen application (-10 min). Twenty minutes after baclofen application, delta waves were observed in the EEG of wild-type, but not of GABA_B(2)^-/-, mice (+20 min). Single spikes appeared sporadically in the EEG of wild-type mice (+40 min), followed by delta waves that lasted for several hours (+7 hr). Ten hours after baclofen application, the EEG traces of wild-type and GABA_B(2)^-/- mice were again similar (+10 hr). B, Quantification of baclofen-induced delta waves in the EEG of wild-type and GABA_B(2)^-/- mice. The percentage of delta waves of the total power amplitude was calculated over periods of 10 min. Three to four mice per genotype were analyzed.

Figure 8.

Lack of baclofen-induced motor impairment and hypothermia in GABA_B(2)^-/- mice. A, No baclofen-induced impairment of rotarod endurance is observed in GABA_B(2)^-/- (-/-) mice (n = 7-10). In contrast, wild-type mice (+/+) show a marked reduction in rotarod performance after baclofen application (p < 0.05; Fisher's post hoc tests). The vehicle-treated control groups stayed on the rotarod during the entire experiment (300 sec) at all time points examined. Thus, in the graph, the data points for the wild-type vehicle control (black dots) are hidden behind the data points for the GABA_B(2)^-/- vehicle control (white dots). At all time points after baclofen application (1, 2, and 4 hr), the GABA_B(2)^-/- group (white triangles) differed significantly from the wild-type control group (black triangles) (p < 0.05; Fisher's post hoc tests). All data points represent mean ± SEM values. B, Baclofen induces a potent reduction in body temperature in wild-type mice (black triangles) compared with the vehicle control group (black dots) (p < 0.05; Fisher's post hoc tests), whereas it is without effect on basal temperature in GABA_B(2)^-/- mice (n = 7-10). However, GABA_B(2)^-/- mice (white dots) exhibit a slight but significantly reduced basal temperature compared with wild-type littermates (black dots) (p < 0.05; Fisher's post hoc tests). All data points represent mean ± SEM values.

Figure 9.

Behavioral analysis of GABA_B(2)^-/- mice. A, Hypolocomotor activity in GABA_B(2)^-/- mice. During a 1 hr observation period, GABA_B(2) knock-out mice (-/-) moved over significantly longer distances (left histogram) with significantly greater speed (right histogram) than heterozygous (+/-) and wild-type (+/+) control mice. n = 7-8 per genotype; mean ± SEM; ^*p < 0.05. B, Response latencies of wild-type (+/+), heterozygous (+/-), and GABA_B(2) knock-out (-/-) mice in the hotplate test assessed at 55°C. GABA_B(2)^-/- mice show significantly reduced paw-lick latencies compared with wild-type and heterozygous control groups. n = 19-20 per genotype; mean ± SEM; ^***p < 0.001. C, Response latencies of wild-type (+/+), heterozygous (+/-), and GABA_B(2) knock-out (-/-) mice in the tail-flick test assessed at infrared intensity 14. GABA_B(2)^-/- mice show significantly reduced tail-flick latencies compared with wild-type and heterozygous control groups. n = 19-21 per group; mean ± SEM; ^*p < 0.05. D, Paw-withdrawal thresholds for wild-type (+/+), heterozygous (+/-), and GABA_B(2) knock-out (-/-) mice in response to a mechanical stimulus. Withdrawal thresholds of the left hindpaw were assessed for each genotype. GABA_B(2)^-/- mice show a significantly reduced withdrawal threshold compared with wild-type and heterozygous control groups. n = 19-21 per group; ^***p < 0.001. Nociception tests were analyzed with Tukey's honestly significant difference test. In all tests, there were no significant differences in the behavior of wild-type or heterozygous mice. E, Impaired passive avoidance learning in GABA_B(2)^-/- mice. Step-through latencies of wild-type (+/+) and GABA_B(2) knock-out (-/-) mice into the dark (shock) compartment on the training day (white bars) and in the retention test (black bars). GABA_B(2)^-/- mice were slower to enter on training day but faster in the retention test compared with the wild-type control mice. Wild-type, but not GABA_B(2)^-/-, mice show significantly longer latencies to enter the dark compartment in the retention test compared with the training trial, which is taken as an index of memory of the initial experience. n = 6-11 per group; mean ± SEM; ^***p < 0.001 versus training; #p < 0.05 versus genotype; ##p < 0.01 versus genotype.