In lamina II of the spinal dorsal horn, synaptic inhibition mediated by ionotropic GABAA and glycine receptors contributes to the integration of peripheral nociceptive messages. Whole-cell patch-clamp recordings were performed from lamina II neurons in spinal cord slices to study the properties of miniature IPSCs (mIPSCs) mediated by activation of GABAA and glycine receptors in immature (<30 d) and adult rats. Blockade of neurosteroidogenesis by 1-(2-chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline carboxamide (PK11195), an inhibitor of the peripheral benzodiazepine receptor (PBR), or finasteride, which blocks 5α-reductase, accelerated the decay kinetics of GABAA receptor-mediated mIPSCs in immature, but not in adult animals. Glycine receptor-mediated mIPSCs remained unaffected under these conditions. These results suggest the presence of a tonic production of 5α-reduced neurosteroids in young rats that confers slow decay kinetics to GABAA mIPSCs. At all of the ages, selective stimulation of PBR by diazepam in the presence of flumazenil prolonged GABAA mIPSCs in a PK11195- and finasteride-sensitive manner. This condition also increased the proportion of mixed GABAA/glycine mIPSCs in the immature animals and led to the reappearance of mixed GABAA/glycine mIPSCs in the adult. Our results might point to an original mechanism by which the strength of synaptic inhibition can be adjusted locally in the CNS during development and under physiological and/or pathological conditions by controlling the synthesis of endogenous 5α-reduced neurosteroids.
- substantia gelatinosa
- dorsal horn
In the dorsal horn of spinal cord, synaptic inhibition mediated by GABAA receptors (GABAARs) and glycine receptors (GlyRs) is of fundamental importance to prevent the generation of hyperalgesia or allodynia (Yaksh, 1989; Sherman and Loomis, 1996; Sorkin and Puig, 1996). In lamina II, local excitatory and inhibitory interneurons receive and integrate cutaneous nociceptive messages conveyed by nonmyelinated (type C) primary afferent fibers before transmitting them to supraspinal centers (Millan, 1999). Recently, it has been shown that, although GABA and glycine are coreleased at a subset of lamina I–II inhibitory synapses, postsynaptic codetection by GABAARs and GlyRs disappeared around postnatal day 23 (P23), leading to an overall reduction of synaptic inhibition (Chéry and De Koninck, 1999; Keller et al., 2001). This phenomenon could, however, be reversed by increasing the affinity of GABAARs with the benzodiazepine diazepam (DZP). Apart from a direct interaction with GABAARs [at a binding site referred to as central benzodiazepine receptor (CBR)], most benzodiazepines/endozepines stimulate neurosteroidogenesis by activating peripheral benzodiazepine receptors (PBRs) located on glial/neuronal mitochondrial membranes (Compagnone and Mellon, 2000). Although it is well established that neurosteroids and benzodiazepines can modulate GABAA receptor function (Harrison et al., 1987; Lambert et al., 1995; Cooper et al., 1999; Mehta and Ticku, 1999; Vicini et al., 2002; Koksma et al., 2003), a role for endogenously produced neurosteroids or endozepines in the acute functional modulation of synaptic transmission has not been clearly reported so far. We addressed this issue by evaluating the consequences of selective activation or blockade of the CBR and PBR on inhibitory synaptic transmission within lamina II of the spinal cord slices from immature and adult rats. Pharmacological manipulation of neurosteroidogenesis allowed us to demonstrate that 5α-reduced neurosteroids are continuously produced during immature stages, but not in adulthood. As a consequence, synaptic inhibition is sustained in the spinal cord of young rats and is mediated by prolonged GABAAR currents as well as by mixed GABA/glycine synaptic currents. In contrast, in the adult, the decay kinetics of GABAergic miniature IPSCs (mIPSCs) are much faster, and mixed GABA/glycine mIPSCs are no longer detected. At all of the ages examined, stimulation of neurosteroidogenesis remained possible and led to a significant increase in overall inhibition by prolonging GABAA receptor-mediated mIPSCs and by unmasking a GABA/glycine cotransmission. Our results suggest an important role of endogenously produced neurosteroids in the modulation of processing of nociceptive information at the spinal cord level via the control of the strength of inhibitory transmission among dorsal horn interneurons.
Materials and Methods
Slice preparation and solutions. All of the procedures were performed in accordance with the guidelines of the European Community on animal experimentation. Transverse slices were prepared from very young (P6–P8), immature (P13–P23), and adult (>P30) Wistar rats as described previously (Keller et al., 2001). The spinal cord was removed and immediately immersed in cold (4°C) sucrose-based artificial CSF (ACSF) containing (in mm): 248 sucrose, 11 glucose, 26 NaHCO3, 2 KCl, 1.25 KH2PO4, 2 CaCl2, 2 MgSO4, 2 kynurenic acid (Fluka, Buchs, Switzerland) continuously bubbled with 95% O2–5% CO2, pH 7.35 ± 0.05. Transverse slices (250–300 μm thick) were cut from the lumbar segment with a Vibratome (Pelco International, Redding, CA) and stored at room temperature in regular ACSF containing 125 mm of NaCl instead of sucrose and no kynurenic acid. After at least 1 hr, slices were transferred to the recording chamber and continuously perfused with oxygenated ACSF containing 0.5 μm TTX (Latoxan, Rosans, France) and 2 mm kynurenic acid to record mIPSCs. All of the recordings were made at 20–22°C.
Electrophysiological recordings, data acquisition, and analysis. Lamina II neurons were identified under an upright microscope (Axioscope; Zeiss, Oberkochen, Germany) equipped with digital interference contrast optics and an infrared camera (Hamamatsu, Hamamatsu City, Japan). Whole-cell voltage-clamp recordings were obtained with an Axopatch 200B amplifier (Axon Instruments, Union City, CA) and borosilicate glass electrodes with an inner filament (1.2 outer diameter to 0.69 inner diameter; Clark Electromedical Instruments, Pangbourne, UK) filled with a intracellular solution containing (in mm): 130 CsCl, 2 MgCl2, and 10 HEPES, pH 7.3, CsOH.
Membrane currents were filtered (DC; 5 kHz; Bessel filter of the amplifier) and stored on videotape after digitization with a pulse-code modulator (40 kHz; Sony, Tokyo, Japan). For off-line analysis, current traces were filtered at 2 kHz (eight-pole Bessel filter; 9002; Frequency Devices) and digitized at 4 kHz on an Intel (Santa Clara, CA) Pentium-based personal computer. Data were acquired using the Strathclyde electrophysiology software CDR [courtesy of Dr. J. Dempster (University of Strathclyde, Glasgow, UK)] and analyzed with a custom software developed and kindly provided by Dr. Yves De Koninck (University of Laval, Québec, Québec, Canada). Mean values of the amplitudes, 10–90% rise times, decay time constants, and frequency of occurrence of mIPSCs were compared between groups using Student's t tests with a confidence interval of 0.95. Decay time constants of mIPSCs were fitted using a nonlinear least square methods, and goodness of fit was evaluated on the basis of fitting subsets of points drawn from the entire set of data points, from evaluation of the reduced χ2 values, and the change in the F values calculated from the sum of squared differences from the fitted line. The Kolmogorov–Smirnov test was used to compare the cumulative distributions. Distributions were considered different when p < 0.01. All of the numerical results are expressed as mean ± SEM.
Drug application. Selective blockade of glycine or GABAA receptors was achieved by adding strychnine hydrochloride (1 μm; Sigma, St. Louis, MO) or bicuculline methiodide (10 μm; Sigma), respectively. The effects of diazepam (Sigma) and allopregnanolone (AP) (5α-pregnan-3α-ol-20-one; Sigma) were determined after steady-state applications lasting at least 30 min. For experiments requiring >1 hr of drug exposure, slices were preincubated in ACSF containing the substance to be tested [e.g., 1-(2-chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline carboxamide (PK11195) (Tocris; Bioblock, Illkirch, France), flumazenil (FLU) (gift of Roche, Basel, Switzerland), diazepam (Sigma), or finasteride (FIN) (Sigma)]. In such experiments, the controls consisted of sister slices incubated under the same conditions but in the absence of the pharmacological agents. The GABAA receptor agonist isoguvacine (Peninsula Laboratories, San Carlos, CA) was applied locally by pressure ejection (duration, 100 msec) through a borosilicate glass pipette.
Diazepam and allopregnanolone are potent positive modulators of synaptic GABAARs expressed by immature (<P23) lamina II neurons
In immature lamina II neurons, inhibitory synaptic transmission is complex and involves a significant proportion of GABA/glycine cotransmission (Chéry and De Koninck, 1999; Keller et al., 2001). Therefore, we first recorded pharmacologically isolated glycine receptor-mediated miniature IPSCs (GlyR mIPSCs) or GABAA receptor-mediated miniature IPSCs (GABAAR mIPSCs), and characterized their modulation by the benzodiazepine DZP (1 μm) and the neurosteroid AP (100 nm).
GABAAR and GlyR mIPSCs occurred at a low frequency [GABAAR mIPSCs, 1.34 ± 0.12 Hz (n = 9); GlyR mIPSCs, 1.35 ± 0.08 Hz (n = 9)] and exhibited amplitudes ranging from 8 to 100 pA [mean peak amplitude, GABAAR mIPSCs, 33.2 ± 2.4 pA (n = 9); GlyR mIPSCs, 33.3 ± 6.2 pA (n = 9)]. The decaying phase of pharmacologically isolated GABAAR and GlyR mIPSCs could be fitted by a monoexponential function. Decay time constants were of 30.8 ± 0.9 and 11.7 ± 1.5 msec for GABAAR(n = 12) and GlyR (n = 5) mIPSCs, respectively.
Figure 1 summarizes the effects of DZP and AP on the decay kinetics of GlyR and GABAAR mIPSCs. Exogenous application of DZP (1 μm) or of AP (100 nm) increased the decay time constant of GABAAR mIPSCs [DZP, 69.7 ± 8.4 msec (n = 6); AP, 60.9 ± 5.9 msec (n = 5); t test; p < 0.05 in both conditions) without affecting their mean amplitude or frequency of occurrence. Superfusion of FLU (10 μm), a potent antagonist of the benzodiazepine site on GABAARs, fully antagonized the rapid modulatory effect of diazepam on GABAAR mIPSC mean decay kinetics (Fig. 1A). Application of FLU alone had no significant effect on GABAAR mIPSCs (t test; p > 0.05; n = 11) (Table 1). GlyR mIPSC kinetics was not affected by DZP or AP (t test; p < 0.05) (Fig. 1).
Diazepam stimulates neurosteroidogenesis and increases synaptic GABAAR function
As shown above for immature lamina II neurons (Fig. 1, Table 1), DZP (1 μm) rapidly increased the mean decay time constant of GABAAR mIPSCs, and this effect was fully blocked when DZP was applied together with 10 μm FLU (n = 5). However, when the application of DZP in the presence of FLU exceeded 1 hr, we observed a progressive and delayed increase in GABAAR mIPSC decay time constants, reaching an average of 41.3 ± 1.7 msec (n = 4) after 80 min (Fig. 2, Table 1). This delayed effect of DZP on GABAA mIPSCs was blocked by the PBR antagonist PK11195 (DZP+FLU+PK, τ = 29.5 ± 1.5; n = 10) and by the 5α-reductase inhibitor finasteride (DZP+FLU+FIN, τ = 31.7 ± 1.6; n = 5). In some cases, we noted that the decay time constant of GABAA mIPSCs had a tendency to slightly decrease during the initial phase (<1 hr) of superfusion with DZP+FLU (Figs. 1 A, 2 A). However, it must be emphasized that this apparent reduction never reached statistical significance (t test; p > 0.05).
A tonic production of neurosteroids shapes synaptic GABAAR mIPSCs in immature, but not in adult, spinal cord slices
In these experiments, pharmacologically isolated GABAAR mIPSCs were recorded from immature lamina II neurons (<P23) in the presence of strychnine (1 μm).
As shown in Figure 3A and Table 1, the properties of GABAAR mIPSCs were unaltered by FLU (10 μm), an antagonist of the CBR. In contrast, incubation of the slices with PK11195 (10 μm), an antagonist of the PBR, significantly reduced the mean decay time constant [control, τ = 30.6 ± 0.9 msec (n = 9); PK11195, τ = 22.8 ± 1.8 msec (n = 9); t test; p < 0.05], whereas all of the other parameters remained unchanged (Fig. 3A, Table 1). Blockade of 5α-reductase, an enzyme involved in the synthesis of 5α-reduced neurosteroids, by FIN (50 μm), significantly reduced the mean decay time constant of GABAAR mIPSCs to 23.1 ± 2.4 msec (n = 7), a value comparable with that obtained during a prolonged (>1 hr) incubation with PK11195 (Fig. 3B, Table 1) (t test; p < 0.05). It should be emphasized that whole-cell GABAA currents, evoked by exogenous application of 100 μm isoguvacine, were unaffected by FLU, PK11195, or FIN (Fig. 3C). The lack of significant change in the peak current during bath perfusion of FLU (0.36 ± 1.00%; n = 5), PK11195 (0.09 ± 2.57%; n = 6), or FIN (0.66 ± 0.88%; n = 3) indicated that these substances did not directly modulate the activity of postsynaptic GABAA receptors. Together, these results suggested that GABAAR mIPSC kinetics in spinal cord slices from immature rats were tonically and positively modulated by endogenous 5α-reduced neurosteroids generated via the tonic activation of the PBR.
The situation was slightly different in lamina II neurons of adult spinal cord preparations. The decay time constants of GABAAR mIPSCs were significantly shorter (τ = 18.8 ± 1.1 msec; n = 11; t test; p < 0.05) than in young rats (τ = 30.8 ± 0.9 msec; n = 48), whereas all of the other kinetic characteristics were similar to that of mIPSCs recorded in neurons from <P23 animals (Tables 1 and 2). In the adult, GABAAR mIPSCs could still be positively modulated by exogenous application of 1 μm DZP (τ = 28.9 ± 3.0 msec; n = 6; t test; p < 0.05) or 100 nm AP (AP, τ = 35.2 ± 1.1 msec; n = 4; t test; p < 0.05), or long incubation (>3 hr) with DZP plus FLU (τ = 34.1 ± 3.6 msec; n = 6; t test; p < 0.05) (Table 2). In the latter experimental conditions aimed at stimulating neurosteroidogenesis (i.e., DZP+FLU) in adult slices, it should be noted that prolongation of GABAA mIPSC decays was sensitive to the 5α-reductase inhibitor finasteride (DZP+FLU+FIN, τ = 18.5 ± 2.5; n = 6). The mean decay time constant of the GABAAR mIPSCs was, however, not affected by the steady-state perfusion of PK11195 (τ = 22.8 ± 1.8 msec; n = 7) or FIN (τ = 20.5 ± 0.7 msec; n = 3) alone. Together, these results indicated that, in the adult, synaptic GABAA receptors were apparently not tonically modulated by endogenously produced neurosteroids under resting conditions, but that such a modulation remained possible after stimulation of PBR.
Neurosteroidogenesis favors GABA/glycine cotransmission in immature and adult rats
As described previously, lamina II neurons of young rats (<p23) recorded in the absence of ionotropic inhibitory amino acid receptor antagonist displayed a significant proportion of mixed GABAA/GlyR mIPSCs characterized by a fast glycine-receptor-mediated and a slow GABAA receptor-mediated kinetic component (Keller et al., 2001).
In line with this finding, the decay phases of the mixed mIPSCs recorded in the present experiments were fitted by a biexponential function (n = 51) with a fast glycine receptor-mediated component (τ1 = 9.2 ± 0.3 msec) and a slow GABAA receptor-mediated component (τ2 = 34.3 ± 1.2 msec) (Fig. 4A). Exogenous application of DZP (1 μm) or of the neurosteroid AP (100 nm) selectively increased the slow component of mixed GABAA/GlyR mIPSCs [DZP, τ2 = 64.7 ± 6.3 msec (n = 7); AP, τ2 = 70.6 ± 7.1 msec (n = 5)]. This effect was similar to that observed in the case of pharmacologically isolated GABAAR mIPSCs described in the previous section (Figs. 1, 4A). The fast decaying phase of mixed mIPSCs (i.e., corresponding to the GlyR current component) was unaffected by DZP or AP (Fig. 4A). Application of AP did not change the overall frequency of occurrence of mIPSCs (regardless of their nature) during AP application (0.247 ± 0.032 vs 0.251 ± 0.033 Hz; n = 6). However, the relative contribution of different mIPSC subtypes was modified by AP. The proportion of GABAAR mIPSCs remained stable (0.131 ± 0.027 vs 0.129 ± 0.027 Hz; n = 6) whereas that of mixed mIPSCs increased from 0.061 ± 0.005 to 0.090 ± 0.01 Hz (n = 6), and that of GlyR mIPSCs decreased in a symmetrical manner (from 0.055 ± 0.017 to 0.033 ± 0.017 Hz; n = 6). As a consequence, the contribution of mixed GABAA/GlyR mIPSCs to the overall mIPSCs population increased from 26.8 ± 3.1 to 40.2 ± 4.7% (n = 6). Application of bicuculline (10 μm) reduced the frequency of synaptic events. This phenomenon could be completely accounted for by the suppression of GABAAR mIPSCs and of the slow component of mixed GABAA/GlyR mIPSCs. As a consequence, mixed mIPSCs appeared to be converted into GlyR mIPSCs. Indeed, the frequency of GlyR mIPSCs remaining in the presence of bicuculline corresponded to the sum of that of mixed mIPSCs and GlyR mIPSCs recorded in the absence of bicuculline.
The effect of AP on the proportion of mixed mIPSCs was mimicked by DZP. After 15–30 min of exogenous application of DZP or AP (Fig. 5A), the fractional contribution of mixed GABAA/GlyR mIPSCs (with respect to control) to the overall population of mIPSCs was strongly increased [DZP, from 22.1 ± 1.4 to 30.5 ± 2.5% (n = 7); t test; p < 0.05; AP, from 22.1 ± 1.4 to 35.0 ± 6.2% (n = 6); t test; p < 0.05]. Perfusion of flumazenil alone did not significantly affect the proportion of mixed events (22.0 ± 3.9%; n = 8; t test; p > 0.05). Neither did acute and short-time (<1 hr) applications of DZP in the presence of flumazenil (22.9 ± 3.3%; n = 16; t test; p > 0.05) or of DZP in the presence of flumazenil plus PK11195 (19.8 ± 2.7%; n = 3; t test; p > 0.05). However, stimulation of neurosteroidogenesis by long treatment (>1 hr) with DZP in the presence of flumazenil (Fig. 5A) increased the fractional contribution of mixed mIPSCs to 27.5 ± 2.7% (n = 4; t test; p < 0.05). Accordingly, pharmacological treatments designed to inhibit the production of 5α-reduced neurosteroids dramatically decreased the proportion of mixed mIPSCs recorded [FLU+PK, 13.2 ± 1.27% (n = 6); t test; p < 0.05; PK11195, 9.9 ± 2.6% (n = 4); t test; p < 0.05; finasteride, 8.9 ± 1.0% (n = 3); t test; p < 0.05]. These results indicate that endogenous neurosteroids potentiate GABAA receptor function and thereby favor the occurrence of mixed GABAA/GlyR mIPSCs.
In adult lamina II neurons, mixed GABAA/GlyR mIPSCs were not detected (Fig. 5B), although these neurons (n = 14) displayed GABAAR and GlyR mIPSCs in an equivalent proportion. Application of DZP or AP led to the reappearance of mixed GABAA/GlyR mIPSCs, which represented respectively 25.6 ± 6.0% (n = 3) and 16.3 ± 7.3% (n = 7) of all of the mIPSCs (Fig. 5B). Similarly, after long incubation with DZP plus FLU (>3 hr), the duration of GABAAR-mediated current components was increased, and mixed mIPSCs represented 16.0 ± 5.3% (n = 6) of the total population of mIPSCs (Fig. 5B). The mean value of their slow GABAA receptor-mediated component was 35.0 ± 1.1 msec (n = 6), which was close to that of isolated GABAAR mIPSCs recorded under the same experimental conditions. In slices treated with finasteride for at least 6 hr, mixed GABAA/GlyR mIPSCs were never detected.
Developmental regulation of GABA/glycine cotransmission by neurosteroidogenesis
Figure 6 summarizes the role of neurosteroidogenesis in the shaping of mIPSCs during postnatal maturation of inhibitory synaptic transmission in lamina II.
The proportion of GABAAR mIPSCs increased with age to reach 50% of overall mIPSCs in the adult. This was accompanied by a marked decrease in the mean time constant of their decaying phase. Time constant values similar to that observed in P > 30 could be reached in slices from P6 and P15 animals after incubation with finasteride for 3 hr (Fig. 6A1, Table 2). Contribution of GlyR mIPSCs to the total number of mIPSCs increased from ∼30% at P6 to 50% in the adult, but the decay time constant of GlyR mIPSCs remained stable over the period examined. GlyR mIPSCs kinetics was unaffected by finasteride treatment at any age examined, although the relative proportion of GlyR mIPSCs increased in P6 and P15 animals to reach values close to that observed in the adult (Fig. 6A2). At P6, mixed GABAA/GlyR mIPSCs represented 40% of the events recorded. This proportion decreased to 30% at P15, and no mixed currents were detected in the adult. At immature stages (P6 and P15), the decay time constant of the fast component of mixed mIPSCs remained stable as noted for GlyR mIPSCs, whereas that of the slow component decreased as observed for GABAA mIPSCs (Fig. 6B). At P6 and P15, finasteride treatment strongly reduced the proportion of mixed mIPSCs to ∼10% as well as the mean decay time constant of the slow component of mixed mIPSCs. This effect was similar to that observed for GABAAR mIPSCs.
The present study indicates a role for endogenously produced 5α-reduced neurosteroids in the functional modulation of inhibitory synapses. Moreover, our results on inhibitory synaptic transmission in lamina II of the spinal cord suggest that the acceleration of GABAAR mIPSCs and the disappearance of mixed GABAA/GlyR mIPSCs (Keller et al., 2001) observed between immature and adult developmental stages were primarily associated with a reduction in the endogenous and tonic production of 5α-reduced neurosteroids. Although a tonic neurosteroidogenesis was not detected in adult animals, activation of neurosteroid synthesis by stimulation of PBR remained possible and led to an increase in synaptic GABAAR-mediated transmission and to the reappearance of mixed GABAA/GlyR mIPSCs, a phenomenon that might have important physiological consequences.
A developmentally regulated acceleration of GABAAR synaptic currents has been described in various structures of the CNS (Brickley et al., 1996; Draguhn and Heinemann, 1996; Tia et al., 1996; Brussaard et al., 1997; Hollrigel and Soltesz, 1997; Pouzat and Hestrin, 1997; Dunning et al., 1999; Okada et al., 2000) including the spinal cord (Keller et al., 2001). In supraspinal regions of the CNS, there is clear evidence that changes in the decay kinetics of GABAergic mIPSCs, as well as changes in the biophysical and/or pharmacological properties of postsynaptic GABAARs (Cherubini and Conti, 2001) are correlated with newly expressed α1-containing (Okada et al., 2000; Vicini et al., 2001) or α6-containing (Tia et al., 1996) GABAARs in adult synapses. This is supported by studies of recombinant GABAARs (Verdoorn, 1994; Gingrich et al., 1995; Lavoie et al., 1997; McClellan and Twyman, 1999) and has been recently confirmed in the cerebellum of α1-deficient mice (Vicini et al., 2001). However, similar developmental changes in the decay kinetics of GABAergic mIPSCs occur at GABAergic synapses in lamina II of the spinal cord (Ma et al., 1993; Bohlhalter et al., 1996; Keller et al., 2001) that express exclusively α2, α3, β3, and γ2 subunits throughout postnatal development (Poulter et al., 1992; Paysan and Fritschy, 1998). Therefore, one possible explanation of the changes in mIPSC kinetics of lamina II neurons could be that this phenomenon involves a change in the properties of synaptic GABAARs mediated by endogenous modulators such as endozepines or neurosteroids rather than a change in GABAAR subunit composition.
In our experiments, the blockade of the benzodiazepine site of GABAA receptors by the specific antagonist flumazenil did not significantly affect the kinetics of GABAA mIPSCs, although, in a few cases, the application of DZP together with FLU for durations of <1 hr tended to accelerate mIPSC kinetics. This effect, which never reached statistical significance, (1) might point to the possible role of endozepines, which might not have been fully revealed under our experimental conditions, or (2) might indicate a complex effect of FLU in the presence of DZP. Additional experiments will be required to clearly distinguish between these possibilities. In contrast, during the early postnatal period (<P23), the occurrence of slowly deactivating GABAAR mIPSCs and of mixed GABAA/GlyR mIPSCs in lamina II seems to be clearly correlated with the tonic presence of significant extracellular levels of 5α-reduced neurosteroids, because inhibition of the metabolism of 5α-reduced neurosteroids by finasteride (Concas et al., 1998) conferred to mIPSCs an adult-like phenotype. PBR is implicated in this phenomenon, because its blockade had effects similar to that of finasteride. In the adult, no such tonic neurosteroidogenesis could be detected, although neurosteroid production was still inducible via PBR stimulation with diazepam. Stimulation of neurosteroidogenesis obtained by co-application of DZP+FLU resulted in a delayed (>1 hr) but apparently sudden increase in the duration of GABAAR mIPSCs. Several phenomena might account for this observation. The precise cellular and subcellular distributions of the key neurosteroidogenic enzymes in the dorsal horn spinal cord are still unknown. In addition, the enzymatic process leading to the synthesis of 5α-reduced neurosteroids is slow, and the site of production might be at some distance from the synaptic cleft. These points might account for the long delay between stimulation of PBR and the observed effect on mIPSCs. In addition, despite a constant speed of production of neurosteroids, their effect on mIPSC kinetics might be detectable only after having crossed a threshold concentration. In summary, a relatively long time might be required to reach this threshold, but once it is reached, the modulation of mIPSC kinetics might reflect the actual speed of production of neurosteroids. Because the expression of 5α-reductase appears to be stable throughout postnatal development (Melcangi et al., 1998; Compagnone and Mellon, 2000; Lephart et al., 2001), the observed developmental changes in tonic neurosteroidogenesis are likely to involve a plastic mechanism upstream of 5α-reductase possibly in relation with the expression of PBR and/or of its endogenous ligands and/or a downregulation of the activity of 5α-reductase in the adult.
Little is known about the neurosteroidogenic activity of the spinal cord (Pomata et al., 2000), and only a single study has reported a high activity of 5α-reductase (Celotti et al., 1997) in this structure of the CNS. In this respect, the cellular origin of neurosteroids is an important issue in our experiments. Neurosteroidogenesis has been first described to occur in glial cells of the central and peripheral nervous system (Compagnone and Mellon, 2000), but recently, some key enzymes have also been localized to neurons (Zwain and Yen, 1999; Follesa et al., 2000; Tsutsui et al., 2000; Patte-Mensah et al., 2003). Stimulation of PBR was shown to increase the concentration of pregnenolone, progesterone, and related steroids in brain tissue and in the plasma (Korneyev et al., 1993; Zwain and Yen, 1999; Serra et al., 2001). Here, we clearly show that stimulation of the PBR induces a modulation of synaptic GABAAR function that is independent of the activation of central benzodiazepine sites on GABAARs, because it persisted in the presence of flumazenil, an antagonist of the CBR. Our results indicate that activation of the PBR lead to the production of 5α-reduced neurosteroids at a concentration sufficient to potentiate the function of synaptic GABAA receptors. These results are in good agreement with our previous in vitro study (Schlichter et al., 2000), realized on cultured hypothalamic neurons, which showed that the nonbenzodiazepine anxiolytic etifoxine increased GABAA receptor-mediated synaptic transmission in part by stimulation of the activity of PBR. The hypothesis of a neosynthesis of neurosteroid is further supported by the long delay (>1 hr) required to observe a modulation when DZP was perfused in the presence of the CBR antagonist FLU.
The superficial layers of the dorsal horn spinal cord play a crucial role in the initial processing of peripheral nociceptive information (Millan, 1999). In the adult spinal cord, antagonists of GlyR or GABAAR induce hyperexcitability of dorsal horn neurons and behavioral states characteristic of hyperalgesia and allodynia (Yaksh, 1989; Sherman and Loomis, 1996; Sorkin and Puig, 1996; Ishikawa et al., 2000). As shown in our experiments, the adult spinal cord has preserved a neurosteroidogenic potential and endogenously produced 5α-reduced neurosteroids are able to efficiently increase the level of synaptic inhibition. It is therefore tempting to speculate that neurosteroids, if produced in physiological/pathological states, might increase the efficacy of inhibitory synaptic transmission and could act as a compensatory antinociceptive local control. Until now, no demonstration of such a role has been reported, although some indications may support this hypothesis. First, fluctuations in the concentrations of 5α-reduced neurosteroids in the brain and plasma have been shown to occur under different physiological (pregnancy, stress) and pathological (depression/anxiety) situations (Uzunova et al., 1998; Dong et al., 1999; Strohle et al., 1999; Reddy and Rogawski, 2002). Second, an antinociceptive effect of allopregnanolone to aversive thermal stimuli has been reported in rodents (Kavaliers and Wiebe, 1987; Frye and Duncan, 1994). Although it seems that GABAARs are involved in this process, it is not possible to clearly discriminate between spinal versus supraspinal effects of 5α-reduced neurosteroids in these experiments, because neurosteroids were injected in the peripheral circulation or in the CSF.
In conclusion, we show that spinal inhibition mediated by GABAergic synapses or mixed GABA/glycine cosynapses is controlled by endogenous 5α-reduced neurosteroids. A tonic production of such 5α-reduced neurosteroids was detected in immature animals and then progressively decreased during postnatal development. However, in the adult spinal cord, activation of neurosteroidogenesis was still possible after stimulation of the PBR. This situation leads to an increased inhibitory synaptic drive within lamina II, which could be part of an endogenous modulatory/compensatory mechanism in response to a strong and/or sustained activation of the spinal nociceptive system. Finally, the PBR, which controls the local production of 5α-reduced neurosteroids, might represent an interesting therapeutic target to limit the consequences of excessive activation of spinal pain nociceptive pathways during inflammatory or neuropathic pain situations.
This work was supported by Centre National de la Recherche Scientifique/Université Louis Pasteur, Institut UPSA de la Douleur, and a grant to R.S. from Institut Universitaire de France. A.F.K. is a fellow of the Fondation pour la Recherche Médicale. We thank Francine Herzog for excellent technical assistance. We also thank Roche (Basel, Switzerland) for the kind gift of flumazenil.
Correspondence should be addressed to Dr. Pierrick Poisbeau, Unité Mixte de Recherche 7519, Centre National de la Recherche Scientifique/Université Louis Pasteur, 21, rue René Descartes, F-67084 Strasbourg, France. E-mail:.
Copyright © 2004 Society for Neuroscience 0270-6474/04/240907-09$15.00/0