γ1 GABA_A Receptors in Spinal Nociceptive Circuits

Expression of GABA_AR γ subunits in mouse DRG and spinal cord

We first used qRT-PCR to quantify the expression of GABA_AR γ1, γ2, and γ3 subunits (encoded by Gabrg1, Gabrg2, and Gabrg3 genes) in lumbar spinal cord tissue and DRG, which harbor the neurons that give rise to the peripheral sensory nerve fibers. We analyzed the expression at different developmental stages ranging from E15 to P50 (Fig. 1). Gabrg2 was the most highly expressed GABA_AR γ subunit gene during all developmental stages investigated both in DRG and spinal cord. In both tissues, Gabrg1 expression was low at E15 but increased at birth and remained at relatively constant levels during postnatal development. Expression levels were ∼25 and 30% of those of Gabrg2 in the DRG and spinal cord, respectively. Expression of the Gabrg3 was generally very low with a small peak at P1 both in the DRG and spinal cord. These data show that mouse DRG and spinal cords contain not only high amounts of Gabrg2 mRNA but also considerable amounts of Gabrg1, while Gabrg3 mRNA is basically absent.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

GABA_AR γ subunit expression in mouse spinal cord and DRG. A, qRT-PCR measurements of the three GABA_AR γ subunits relative to β-actin in lumbar DRGs of naive mice. Developmental changes from E15 to adulthood (P50) (n = 5–7 mice). B, Same as A but lumbar spinal cord.

Cellular distribution of γ1 and γ2 GABA_AR in the mouse spinal cord

We next investigated Gabrg1 and Gabrg2 expression on a cellular level in the lumbar spinal cord of adult (7 week old) mice (Fig. 2). On a gross scale, multiplex fluorescent in situ hybridization (mFISH) revealed that both Gabrg1 and Gabrg2 transcripts were found across the entire spinal dorsal horn (Fig. 2A). Gabrg2 was present in virtually all GABAergic (vGAT positive) and glutamatergic (vGluT2 positive) neurons, both in the superficial and deep dorsal horn (Fig. 2B,C). The expression pattern of Gabrg1 was less uniform. Within the superficial dorsal horn, Gabrg1 was expressed in 44.2 ± 2.0% of GABAergic and in 29.0 ± 3.6% of glutamatergic neurons. Expression in the deep dorsal horn was lower, with 12.0 ± 1.2% of GABAergic and 9.8 ± 2.0% of glutamatergic neurons expressing Gabrg1. In total, 56.2 ± 2.5% of all Gabrg2-containing neurons were glutamatergic, while the remaining ones were GABAergic (37.2 ± 2.3%; Fig. 2B,C). Gabrg1 was slightly more prevalent in GABAergic (21.2 ± 2.2%) than that in glutamatergic dorsal horn neurons (15.6 ± 2.0%), in line with the results of a previous single-cell RNA sequencing study (Haring et al., 2018).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cellular expression pattern of the GABA_AR γ1 and γ2 subunits. A, mFISH experiments showing expression of Gabrg1 (left) and Gabrg2 (right) in transverse lumbar spinal cord sections costained with DAPI to visualize cell nuclei. Scale bar, 200 µm. B, Expression of Gabrg1 and Gabrg2 in glutamatergic (vGluT2) and GABAergic (vGAT) neurons in the superficial dorsal horn. Arrows indicate cells with coexpression. C, Statistical analysis: g1, Gabrg1; g2, Gabrg2; sdh, superficial dorsal horn; ddh, deep dorsal horn. Mean ± SD. D, Coexpression of Gabrg1 and Gabrg2 with Gabra2 and Gabra3. Arrows indicate cells with coexpression. Scale bar, 20 µm. E, Statistical analysis. g1, Gabrg1; g2, Gabrg2; a2, Gabra2; a3, Gabra3. Eight sections were analyzed per condition. Two sections were taken per mouse. Each dot represents a single section. F, G, Expression of Gabrg1 in different types of non-neuronal cells. F, mFISH using probers for Gabrg1, GFAP (astrocytes), Olig2 (oligodendrocytes), and Aif1 (microglia). Arrows indicate cells coexpressing the respective marker with Gabrg1. Scale bar, 20 µm. G, Statistical analysis. Mean ± SD.

We next analyzed whether GABA_AR α subunits were coexpressed with Gabrg1 and Gabrg2 (Fig. 2D,E). Since GABA_AR α2 and α3 subunits are the most prevalent α subunits in the spinal dorsal horn (Bohlhalter et al., 1996; Paul et al., 2012), we focused our analyses on these subunits. Most dorsal horn cells expressing Gabrg1 also contained Gabra2 and/or Gabra3 (87.0 ± 2.4% and 49.1 ± 3.6% for Gabra2 and Gabra3, respectively). Almost all Gabrg2-containing neurons (96.5 ± 1.0%) also contained Gabra3 and 79.3 ± 1.2% contained Gabra2 (Fig. 2E).

Interestingly, much of the Gabrg1 expression (∼63%) associated with cell bodies (DAPI-positive structures) could not be localized to GABAergic or glutamatergic neurons (compare Fig. 2C), suggesting significant expression in non-neuronal cells. Since Gabrg1 was often localized in thin elongated structures presumed axons (compare Fig. 2A), it seemed conceivable that the non-neuronal expression is in oligodendrocytes ensheathing neuronal axons, consistent with the report by Ordaz et al. (2021). Additional mFISH experiments (Fig. 2F,G) verified that Gabrg1 was expressed in oligodendrocytes (Olig2 positive cells), but also in astrocytes (Gfap positive cells) and microglia, identified by expression of the Aif1 gene, which encodes for the microglia marker IBA1.

Taken together, these experiments confirm that Gabrg1 is expressed both in inhibitory and excitatory neurons and in different glial cells of the dorsal horn and that neuronal Gabrg1 expression is more prevalent in the superficial than in the deep dorsal horn.

Mice lacking γ2 subunits from the spinal cord

To further investigate the function of GABA_AR γ subunits, we decided to study the consequences of genetic ablation of γ2 GABA_AR subunit. Since most mice that lack GABA_AR γ2 subunits globally die early after birth (Essrich et al., 1998), we used a strategy that allowed us to delete GABA_AR γ2 subunits from the spinal cord and DRG but to retain expression in the brain. To this end, we crossed mice carrying a floxed Gabrg2 allele (Schweizer et al., 2003) with transgenic mice that express the Cre recombinase under the transcriptional control of the hoxB8 gene (Witschi et al., 2010, see also Paul et al., 2014). HoxB8-γ2^−/− mice were viable and showed no obvious anatomical or behavioral abnormalities. Using immunocytochemistry and qRT-PCR, we verified that γ2 GABA_ARs were completely absent from the spinal cord and DRG (Fig. 3A–C). FISH experiments demonstrate that the gross expression pattern of Gabrg1 was not altered in dorsal horn sections taken from hoxB8-γ2^−/− mice (Fig. 3A, right panels). Accordingly, neither Gabrg1 nor Gabrg3 mRNA were altered in the DRG or spinal cords of hoxB8-γ2^−/− mice (Fig. 3A, left panel). We found, however, an upregulation of Gabrg2 mRNA when probes were used that bind to mRNA outside the deleted region, indicating the presence of some homeostatic processes (Fig. 3B,C). The presence of Gabrg2 mRNA in hoxB8-γ2^−/− mice raises the possibility that a truncated protein might have be expressed in the hoxB8-γ2^−/− mice. However, the γ2 antibody used in this study, which was raised against the N-terminal 29 amino acids of the γ2 GABA_AR subunit (Fritschy and Mohler, 1995), did not detect any remaining γ2 GABA_AR subunit protein in the hoxB8-γ2^−/− mice indicating that the remaining mRNA was not translated into protein (compare Fig. 3A). Furthermore, any remaining γ2 GABA_AR subunit protein would lack the transmembrane segment (TM) 3, part of TM2, and part of the large intracellular loop (Günther et al., 1995) and would therefore be nonfunctional.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Nociceptive sensitivity of hoxB8γ2^−/− mice. A, B, Verification of the loss of γ2 subunit expression in hoxB8γ2^−/− mice and unchanged Gabrg1 expression in hoxB8γ2^−/− mice. A, Left, Transverse lumbar spinal cord sections of a γ2^fl/fl (left) and a hoxB8γ2^−/− mouse (right) stained for gephyrin and GABA_AR γ2 subunit protein. Right, Comparison of Gabrg1 expression (FISH) in γ2^fl/fl and hoxB8γ2^−/− mice. Scale bar, 100 µm. B, Left, Quantification of Gabrg2 mRNA relative to β-actin in lumbar DRGs of γ2^fl/fl (n = 4–6) and hoxB8γ2^−/− (n = 8–9) mice. Gabrg2 mRNA was detected with a probe binding to the sequence flanked by the two loxP sites. Unpaired test. Right, Quantification of the GABA_AR γ subunit expression in DRGs of γ2^fl/fl (n = 4–6) and hoxB8γ2^−/− mice (n = 6). Here, Gabrg2 mRNA was detected with a probe binding a sequence outside the region flanked by the two loxP sites. t tests followed by Bonferroni’s correction for multiple testing. C, Same as B, but lumbar spinal cord tissue (n = 8, for both γ2^fl/fl and hoxB8γ2^−/− mice). D, Nociceptive and somatic sensitivity. Mechanical sensitivity was tested in the pin prick, the von Frey test, and with a soft paint brush. Thermal sensitivity was assessed in the Hargreaves test, in the cold plantar test (dry ice) and the acetone test. Unpaired t tests, n = 10–14 and 10–16 for hoxB8γ2^−/− mice and γ2^fl/fl mice, respectively. Closed and open symbols indicate male and female mice, respectively. E, Muscle strength, motor coordination, and effects of TPA023B (1 mg/kg, p.o.) on locomotor activity assessed in the horizontal wire test, the rotarod test, and the actimeter test, respectively. Muscle relaxation, unpaired t test, n = 10 and 10 for hoxB8γ2^−/− mice and γ2^fl/fl mice. Motor coordination, unpaired t test, n = 8 and 8 for hoxB8γ2^−/− mice and γ2^fl/fl mice. Locomotor activity. Two-way ANOVA followed by Bonferroni’s post hoc test. Treatment * genotype interaction F_(1,18) = 0.014, n = 6–7 and 4–6 for hoxB8γ2^−/− mice and γ2^fl/fl mice. Circles and squares represent individual mice. Closed and open symbols indicate male and female mice, respectively. Mean ± SEM.

Spinal cord-specific deletion of γ2 GABA_ARs does not alter nociceptive sensitivity

Loss of synaptic inhibition in the spinal dorsal horn, for example through blockade of spinal GABA_ARs, induces exaggerated nociceptive reactions (for a review, see Zeilhofer et al., 2012). We therefore tested whether spinal cord-specific deletion of spinal GABA_AR γ2 subunits would alter the sensitivity of mice in a battery of sensory and nociceptive tests (Fig. 3D). Unexpectedly, sensitivity to noxious mechanical, heat, and cold stimuli was indistinguishable from that of wild-type (γ2^fl/fl) mice. We also found no differences in muscle strength, assessed in the horizontal wire test, and in the rotarod test, a measure of motor coordination (Fig. 3E). The only significant difference discovered was a decreased responsiveness to light dynamic touch. Unaltered sensitivity to noxious stimuli suggests that synaptic inhibition was sufficiently retained in the superficial layers of the dorsal horn, where nociceptive signals are processed. The observed change in responsiveness to dynamic touch stimuli may reflect a change in synaptic inhibition in the deep dorsal horn, where signals from innocuous mechanical stimulation are processed. The decreased rather than increased sensitivity may suggest the presence of a disinhibitory circuit involving γ1 GABA_ARs expressed on GABAergic neurons. To verify that supraspinal γ2 GABA_ARs were intact, we tested the effect of TPA023B, an α2/α3 GABA_AR subtype-selective BDZ site agonist, that increases spontaneous locomotion in mice probably via its anxiolytic activity (Ralvenius et al., 2018). No differences were found in TPA023B-induced increase in locomotion between hoxB8-γ2^−/− and γ2^fl/fl mice (Fig. 3E).

GABA_AR clusters in mice lacking γ2 subunits in the spinal cord

To better understand why nociceptive responses remained unchanged despite the loss of γ2 GABA_AR subunits in the spinal cord, we quantified postsynaptic GABA_AR clusters in the dorsal horn of γ2^fl/fl and hoxB8-γ2^−/− mice (Fig. 4). We expected that the loss of γ2 would reduce the number of GABA_AR clusters as γ2 subunits are essential for the association of GABA_ARs with the postsynaptic scaffold protein gephyrin (Essrich et al., 1998). To quantify GABA_AR clusters, we stained transverse sections of lumbar spinal cord of hoxb8-γ2^−/− and γ2^fl/fl mice with antisera against α2 and α3 GABA_AR subunits and against gephyrin. We started with analyses of the deep dorsal horn, where the γ1 GABA_AR subunit is only weakly expressed. As expected, GABA_AR clusters containing α2 or α3 GABA_AR were almost completely absent or greatly diminished (α2 GABA_ARs: 0.23 ± 0.009/µm² in γ2^fl/fl mice vs 0.005 ± 0.001/µm² in hoxb8-γ2^−/− mice, equivalent to a reduction by 97.8%; α3 GABA_ARs: 0.46 ± 0.01/µm² vs 0.18 ± 0.02, equivalent to a reduction by 60.9%; Fig. 4A). In contrast, in the superficial layers, where γ1 GABA_AR subunits were more abundant, the numbers of α2 and α3 subunit containing GABA_AR clusters were reduced to a lesser extent (α2 GABA_ARs: 0.64 ± 0.02/µm² in γ2^fl/fl mice vs 0.50 ± 0.03/µm² in hoxb8-γ2^−/− mice, equivalent to a reduction by 21.9%; α3 GABA_ARs: 0.40 ± 0.02/µm² vs 0.37 ± 0.01, equivalent to a reduction by only 7.5%; Fig. 4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

GABA_AR clustering in the dorsal horn of hoxB8γ2^−/− mice. A, Deep dorsal horn. Immunofluorescent staining of gephyrin (green) and GABA_AR α2 subunits (top) or GABA_AR α3 subunits (bottom) in γ2^fl/fl and hoxB8γ2^−/− mice. Statistics: Cluster density of gephyrin and GABA_AR α2 and α3 subunits. Unpaired t tests. Mean ± SEM. B, Same as A but superficial dorsal horn. C, High-resolution images illustrating colocalization of gephyrin with GABA_AR α2 and α3 subunits in the deep dorsal horn. Scale bars, 10 µm (left) and 3 µm (right). D, Same as C but superficial dorsal horn. Individual dots represent one section. In total 9–10 sections from three mice were analyzed per condition.

In most CNS areas, GABA_AR cluster colocalizes with gephyrin. However, at certain sites, clustering apparently occurs in its absence (Kneussel et al., 2001; Levi et al., 2004; Panzanelli et al., 2011). We therefore analyzed whether the loss or retention of GABA_AR clusters in the deep and superficial dorsal horn parallel with the changes in the number of GABA_AR clusters containing gephyrin (Fig. 4C,D). We defined colocalization as points of spatial overlap between the signals generated by α2 or α3 GABA_AR subunit markers with gephyrin markers. In the deep dorsal horn colocalization of gephyrin and α2 GABA_AR subunits were virtually absent (0.003 ± 0.001 clusters/µm² in hoxb8-γ2^−/− mice compared with 0.175 ± 0.004/µm² in γ2^fl/fl mice, equivalent to a reduction by 92.3%), and colocalization between gephyrin and α3 GABA_ARs was reduced from 0.34 ± 0.01/µm² in γ2^fl/fl mice to 0.10 ± 0.01/µm² in hoxb8-γ2^−/− mice, equivalent to a reduction by 70.6% (Fig. 4C). In contrast, in the superficial dorsal horn of hoxb8-γ2^−/− mice, colocalization was reduced only by 30.2 and 25.0% for α2 and α3 GABA_ARs, respectively. These results demonstrate that the reduction in the number of GABA_AR α2 and α3 clusters parallels the reduction of clusters containing GABA_AR α2 or α3 subunits together with gephyrin (Fig. 4D). They hence suggest that neuronal Gabrg1 contributes to the retention of GABA_AR clusters in the absence of γ2 subunits. The colocalization of α2 and α3 GABA_ARs with gephyrin in the superficial dorsal horn indicates in addition that these clusters resided on intrinsic dorsal horn neurons rather than on sensory nerve terminals, which mostly lack gephyrin (Lorenzo et al., 2014).

Agonistic activity of HZ-166 at γ1 GABA_ARs and retained antihyperalgesia by HZ-166 in hoxB8-γ2^−/− mice

The GABA_AR γ2 subunit not only mediates synaptic clustering of GABA_ARs but, together with an α subunit, also forms the BDZ binding site. The majority of tested BDZ site agonists potentiate only γ2 GABA_ARs (Ymer et al., 1990; Wafford et al., 1993; Baburin et al., 2008). Some BDZ agonists, such as diazepam, flunitrazepam, and triazolam also potentiate γ1 GABA_ARs although with considerably lower potencies (Khom et al., 2006; Atack et al., 2011). We therefore asked whether BDZ site agonists with activity at γ1 GABA_ARs would exert at least part of their antihyperalgesic action through the potentiation of γ1 GABA_ARs. To avoid confounding sedative effects in these in vivo experiments, we tested whether the nonsedative BDZ site agonist HZ-166 (Rivas et al., 2009) would potentiate γ1 GABA_ARs (Fig. 5). We have previously shown that HZ-166 reduces inflammatory and neuropathic hyperalgesia without inducing sedation at antihyperalgesic doses (Di Lio et al., 2011). To test whether HZ-166 potentiates γ1 GABA_ARs, we compared the GABA_AR current potentiation by HZ-166 in HEK 293 cells transiently transfected with either α2, β3, and γ1 or with α2, β3, and γ2 subunits. HZ-166 potentiated both subtypes of GABA_ARs. γ2 GABA_ARs were potentiated with an EC₅₀ of 0.15 ± 0.01 µM and an E_max of 162.7 ± 19.5%. γ1 GABA_ARs were potentiated with lower potency (EC₅₀: 8.8 ± 2.3 µM) but higher efficacy (E_max: 375 ± 108%; Fig. 5B,C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Activity of HZ-166 at γ1 and γ2GABA_ARs. A, Schematic representation of the experiment. Potentiation of recombinant γ2 (α2β3γ2) and γ1 (α2β3γ1) GABA_AR currents by HZ-166 was assessed in HEK 293 cells. GABA concentration was EC₅ (1 μM for α2β3 γ2 and 10 μM for α2β3γ1). B, Example trances of GABA evoked membrane currents in the presence and absence of a saturating concentration of HZ-166 (10 and 100 µM for γ2 and γ1 containing GABA_ARs. C, Concentration response curve fitted to the Hill equation with a baseline fixed to 0. Number of cells, n = 6–11. Data are mean ± SEM.

We then analyzed the antihyperalgesic effects of intrathecally injected HZ-166 in mice with neuropathic sensitization induced by a CCI surgery of the sciatic nerve (Fig. 6). Before starting with these behavioral experiments, we analyzed whether the expression of any of the γ GABA_AR subunit would change in response to peripheral nerve injury. We found a significant upregulation of Gabrg1 transcript numbers in DRG (0.0097 ± 0.0022 vs 0.020 ± 0.002; pre- vs post-CCI surgery; t test; p = 0.025 corrected for three independent tests) and a trend toward reduced expression of Gabrg2. Gabrg3 remained at very low levels. In the spinal cord, we detected a significant upregulation of Gabrg3 (0.0036 ± 0.0004 vs 0.092 ± 0.0003; pre- vs post-CCI surgery; t test; p < 0.001 corrected for threes independent test), but its expression level remained well below that of Gabrg1 and Gabrg2 (Fig. 6A). For pharmacological analyses in neuropathic mice, we chose an intrathecal delivery route to further rule out confounding effects resulting from supraspinal sites. In γ2^fl/fl mice, HZ-166 exerted pronounced dose-dependent antihyperalgesia (Fig. 6B), as previously reported (Di Lio et al., 2011). We next compared the antihyperalgesic effects obtained with HZ-166 at a dose of 0.3 mg/kg with those in hoxB8-γ2^−/− mice. HZ-166 was still antihyperalgesic in hoxB8-γ2^−/− mice albeit with reduced efficacy (Fig. 6C). In additional qRT-PCR experiments, we ruled out that the deletion of the γ2 subunit might have led to a differential regulation of Gabrg1 or Gabrg3 in DRG or spinal cords of mice after CCI surgery (Fig. 6D). GABA_AR independent off-target effects of HZ-166 through receptors different from GABA_ARs can also be excluded since the antihyperalgesic effect of HZ-166 is absent from mice carrying BDZ-insensitive α2 GABA_ARs (Ralvenius et al., 2015). This result suggests that in addition to γ2 GABA_ARs, γ1 GABA_ARs contribute to HZ-166-induced antihyperalgesia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Antihyperalgesic actions of HZ-166 in mice with neuropathic hyperalgesia. A, Changes in GABA_AR γ subunit expression after peripheral nerve damage. qRT-PCR measurements of mRNA encoding for the three GABA_AR γ subunits in lumbar DRG (left) and lumbar spinal cords (right), before and 7 d after CCI surgery (n = 8–10 mice per group). mRNA expression is expressed relative to β-actin expression. Statistics, DRG: ANOVA followed by Bonferroni’s post hoc test. F_(2,39) = 6.25. Spinal cord: F_(2,33) = 1.44. Error bars indicate mean ± SEM. B, Dose-dependent reversal of mechanical hyperalgesia by HZ-166 7 d after CCI surgery. n = 9, 6, 7, 6, and 6 mice for vehicle, 0.01, 0.03, 0.1, and 0.3 mg/kg HZ-166 intrathecally. C, Left panel (paw withdrawal threshold vs time), Partially retained antihyperalgesia by HZ-166 (0.3 mg/kg, i.t.) in hoxB8γ2^−/− mice. Repeated-measures ANOVA followed by Dunnett's post hoc test with predrug baseline as reference F_(2,12) = 73.5. Right panel (statistical analysis). Percent maximum possible effect averaged for time points 0.5 and 1 h. Unpaired t test. n = 7, for both hoxB8γ2^−/− mice and γ2^fl/fl mice, respectively. Closed and open symbols indicate male and female mice. D, Expression of Gabrg1 and Gabrg3 in DRGs and spinal cords of γ2^fl/fl and hoxB8γ2^−/− mice 7 d after CCI surgery (mean ± SEM). t tests followed by Bonferroni’s correction for multiple testing. Data are mean ± SEM.