Report

Corelease of Two Fast Neurotransmitters at a Central Synapse

See allHide authors and affiliations

Science  17 Jul 1998:
Vol. 281, Issue 5375, pp. 419-424
DOI: 10.1126/science.281.5375.419

Abstract

It is widely accepted that individual neurons in the central nervous system release only a single fast transmitter. The possibility of corelease of fast neurotransmitters was examined by making paired recordings from synaptically connected neurons in spinal cord slices. Unitary inhibitory postsynaptic currents generated at interneuron-motoneuron synapses consisted of a strychnine-sensitive, glycine receptor–mediated component and a bicuculline-sensitive, γ-aminobutyric acid (GABA)A receptor–mediated component. These results indicate that spinal interneurons release both glycine and GABA to activate functionally distinct receptors in their postsynaptic target cells. A subset of miniature synaptic currents also showed both components, consistent with corelease from individual synaptic vesicles.

Synaptic transmission in the central nervous system (CNS) is mediated by the release of neurotransmitters into the synaptic cleft and the subsequent activation of postsynaptic receptors. A single neurotransmitter can coactivate multiple ionotropic and metabotropic receptor types (1), and a fast neurotransmitter can be coreleased with neuropeptides (2). In the spinal cord and brainstem, both glycine and GABA mediate inhibitory synaptic transmission (3). It is not known, however, whether glycine and GABA are released from separate or overlapping populations of interneurons. Glycine- and GABA-like immunoreactivity coexist in the somata and boutons of subpopulations of spinal interneurons, and glycine receptor (GlyR) subunit, GABAA receptor (GABAAR) subunit, and gephyrin immunoreactivity are present in postsynaptic densities (4). This supports the hypothesis of a glycine- and GABA-mediated cotransmission in the mammalian spinal cord (5).

We examined the cotransmission hypothesis directly, using dual whole-cell patch-clamp recordings from synaptically coupled pairs of interneurons and putative motoneurons in slices from neonatal rats (Fig. 1) (6, 7). Unitary inhibitory postsynaptic currents (IPSCs) evoked in motoneurons by single action potentials in presynaptic interneurons showed a rapid rise (20 to 80% rise time of 0.46 ± 0.05 ms at 22° to 25°C,n = 11; Fig. 1A). The mean peak amplitude (including failures of transmission) was 330 ± 99 pA (−50 mV), corresponding to the opening of ∼100 channels (8). The short and constant latency (2.85 ± 0.19 ms; Fig. 1A) and the small percentage of failures (16.4 ± 6.7%) indicated that the IPSCs were monosynaptic in origin. Unitary IPSCs evoked by presynaptic action potential trains exhibited marked facilitation of the IPSC amplitude (Fig. 1A). These functional properties of unitary IPSCs were similar to those reported with minimal extracellular stimulation (9).

Figure 1

GlyR- and GABAAR-mediated components of unitary IPSCs in interneuron-motoneuron pairs. (A) Simultaneous whole-cell patch-clamp recording from a spinal interneuron (current clamp, upper traces) and a putative motoneuron (voltage clamp, lower traces). In part a, three individual IPSCs were evoked by single presynaptic action potentials (5-ms, 380-pA pulses). In part b, IPSCs were evoked by a train of presynaptic action potentials (55-ms, 380-pA pulse). The inset shows a schematic representation of the paired recording configuration. In part c, the locations of presynaptic interneuron somata are shown superimposed with a schematic drawing of the cytoarchitechtonic layers of the deep lumbar rat spinal cord. (B) Unitary IPSCs in a pair, part a, before antagonist application, part b, in the presence of 400 nM strychnine, and part c, in the presence of 400 nM strychnine plus 5 μM bicuculline. Lower traces in parts b and c are shown at an expanded amplitude scale. The traces represent averages from 3 (part a) or 10 (parts b and c) single sweeps, obtained at the end of the control period and the antagonist washin periods, respectively. (C) Peak amplitude of single unitary IPSCs plotted against time during the application of strychnine and bicuculline as indicated by the horizontal bars. Same pair as shown in (B). Inset shows a histogram of the relative contribution of the GABAAR-mediated component to the peak IPSC, estimated from the amplitude of the strychnine-resistant component and the strychnine block of GABA-activated currents (Fig. 4C).

We next determined the contribution of GlyRs and GABAARs to the unitary IPSCs in pairs using the antagonists strychnine and bicuculline. The GlyR antagonist strychnine (10), added to the bath solution at a concentration of 400 nM (8 pairs) or 1 μM (3 pairs), blocked the unitary IPSC completely in only 2 of 11 interneuron-motoneuron pairs. In 9 of 11 pairs, a strychnine-resistant component remained that decayed more slowly than the strychnine-sensitive component (Fig. 1B) and, on average, contributed 15 ± 5% to the total peak current amplitude (range of 4 to 53%). In 9 of 9 pairs, the strychnine-resistant IPSC component was abolished by the GABAAR antagonist bicuculline (5 μM; Fig. 1, B and C). Thus, unitary IPSCs at spinal inhibitory synapses comprised GlyR- and GABAAR-mediated components that could be dissected pharmacologically.

A comparison of the functional properties of the GlyR- and the GABAAR-mediated component of the evoked unitary IPSC indicated that the two components were not significantly different in the 20 to 80% rise time (0.40 ± 0.06 ms versus 0.53 ± 0.09 ms; Fig. 2A; P > 0.1), the synaptic latency (2.8 ± 0.3 ms versus 3.3 ± 0.2 ms; Fig. 2B;P > 0.1), or the paired-pulse facilitation ratio (1.95 ± 0.29 versus 2.36 ± 0.30; Fig. 2C; P > 0.1). However, they differed markedly in the decay time course; the decay time constant of the GlyR-mediated component was, on average, a factor of 3.6 faster than that of the GABAAR-mediated component (16.3 ± 3.0 ms versus 58.9 ± 12.8 ms; Fig. 2D; P < 0.01).

Figure 2

Comparison of the functional properties of GlyR- and GABAAR-mediated IPSC components. Hatched bars, unitary IPSC in control conditions; open bars, GlyR-mediated component (in the presence of bicuculline in three pairs, and in control conditions in three additional pairs in which subsequent strych- nine application blocked >95% of the IPSC); filled bars, GABAAR-mediated component (in the presence of strychnine in the pairs that showed the largest GABAAR-mediated component). (A) The 20 to 80% rise time of the IPSC. (B) Synaptic latency, measured from the rising phase of the presynaptic action potential to the beginning of individual IPSCs. (C) Paired-pulse facilitation ratio (PPF; amplitude of the second IPSC relative to that of the first IPSC, both measured from the baseline preceding the first IPSC, 10-ms interval). (D) Decay time constants of the IPSC. For the unitary IPSC in control conditions, τ1 and τ2 from the bi-exponential fit are specified separately (amplitude contribution plotted within bar). For the isolated GlyR-mediated and GABAAR-mediated component the mean of τ1 and τ2, weighted with the respective amplitude contribution, is given. Number of pairs is indicated in parentheses.

To examine a possible differential modulation of GlyR- and GABAAR-mediated components by presynaptic GABABreceptors, we investigated the effects of the GABABreceptor agonist R-(−)-baclofen on composite IPSCs evoked by extracellular stimulation of presynaptic axons (Fig. 3) (6, 11). Baclofen reduced the peak amplitude of both the GlyR-mediated component (in the presence of bicuculline) and the GABAAR-mediated component (in the presence of strychnine) (Fig. 3A). Plots of the coefficient of variation to the −2 power against the mean value of the IPSC, both normalized to the respective control values (12), revealed that the data points were located close to the identity line, indicating that the inhibition was presynaptic in origin (Fig. 3B). The concentration dependence of the baclofen effects was almost indistinguishable for the GlyR- and the GABAAR-mediated component (Fig. 3C). Similarly, the metabotropic glutamate receptor agonist L-AP4 (25 μM) reduced the peak amplitude of both components to the same extent (to 55.4 ± 13.7% and 46.2 ± 5.8% of the control value, respectively, four motoneurons in each case). These results suggest comodulation of GlyR- and GABAAR-mediated components of the IPSC by the same set of presynaptic receptors.

Figure 3

Comodulation of GlyR- and GABAAR-mediated IPSC components by presynaptic GABAB receptors. (A) Inhibition of composite IPSCs by 1, 3, and 10 μM R-(−)-baclofen. Part a shows the GlyR-mediated IPSC in the presence of 5 μM bicuculline, and part b shows the GABAAR-mediated IPSC in the presence of 400 nM strychnine, average of 10 single sweeps each. The bath solution contained 5 μM CNQX and 25 μM D-AP5 in both cases. The inset shows a schematic representation of the method of extracellular stimulation of presynaptic axons and the postsynaptic recording configuration. (B) The coefficient of variation to the −2 power (CV−2) of the IPSC peak amplitude in the presence of baclofen was plotted against the mean in the presence of baclofen, both normalized by the respective control values (12). The CV and mean were calculated from 40 evoked IPSCs. Data points for both GlyR- (open circles) and GABAAR-mediated (filled circles) component were adjacent to the identity line, confirming the presynaptic nature of the effect. (C) Inhibition of the GlyRmediated component (open circles) and the GABAAR-mediated component (filled circles) of the IPSC by baclofen, plotted against baclofen concentration. Data points were fit with the equationf = A/[1 + (IC50/c)n], where IC50 is the half-maximal inhibitory concentration,n the Hill coefficient, c the baclofen concentration, and A the maximal effect. The IC50 values were 0.41 μM (n = 0.70,A = 0.85; GlyR-mediated component, dashed curve) and 0.53 μM (n = 0.88, A = 0.89; GABAAR-mediated component, solid curve), respectively. Number of motoneurons indicated in parentheses.

Although these findings appeared to be consistent with a corelease of glycine and GABA at inhibitory synapses in the spinal cord, this conclusion relies on the assumptions that glycine and GABA activate separate receptors, and that strychnine and bicuculline block these receptors selectively. We thus applied brief (1-ms) pulses of the putative transmitters to outside-out patches isolated from motoneuron somata (Fig. 4) (6, 13). Both glycine and GABA evoked detectable currents in the majority of patches (19 of 23 patches); the peak amplitude of the current evoked by 1 mM glycine in these patches was on average a factor of 2.5 larger than that activated by 1 mM GABA (Fig. 4A). The current activated by coapplication of 1 mM glycine and 1 mM GABA was 97.5 ± 3.9% of the arithmetic sum of the glycine- and GABA-activated currents in the same patches (six patches), indicating that glycine and GABA activated molecularly distinct receptors (Fig. 4B).

Figure 4

Selective activation and inhibition of GlyRs and GABAARs in motoneurons. (A) Currents activated in an outside-out patch by 1-ms pulses of either 1 mM glycine or 1 mM GABA and by 1 mM glycine plus 1 mM GABA in the absence and presence of 300 nM strychnine. The inset shows a representation of the fast application method and the recording configuration. The traces are averages from 10 or 20 single sweeps. (B) Bar graphs of the peak currents activated by glycine alone, GABA alone, and coapplication of glycine and GABA in the same patches. The arithmetic sum of glycine- and GABA-activated peak currents is also shown. (C) Fraction of the glycine-activated (open circles, dashed curve) and the GABA-activated current (filled circles, solid curve) blocked by strychnine added to control barrel solution, plotted against strychnine concentration. Data points were fit with the equation given in the legend to Fig. 3. The IC50 values were 39 nM (n = 1.73, A constrained to 1) and 1.5 μM (n and A constrained to 1), respectively. (D) Fraction of GABA-activated (filled circles, solid curve) and glycine-activated current (open circles, dashed curve) blocked by bicuculline added to control barrel solution, plotted against bicuculline concentration. The IC50 values were 0.58 μM (n = 0.99, A constrained to 1) and 144 μM (n and A constrained to 1), respectively. (E) GlyR activation by glycine, β-alanine, taurine, and GABA. The peak amplitudes of the agonist-activated currents were normalized to the peak current activated by 10 mM glycine in the same patches and were plotted against agonist concentration. GABAARs were blocked by 10 μM bicuculline (both barrels). Estimated EC50 values were 441 μM (n = 1.4, A = 0.99), 359 μM (n = 1.7, A = 0.74), 2.46 mM (n = 1.3, A = 0.26), and 105 mM (n constrained to 1.4, A constrained to 1). (F) GABAAR activation by GABA, β-alanine, taurine, and glycine. The peak amplitudes of the agonist-activated currents were normalized to the peak current activated by 10 mM GABA in the same patches and were plotted against agonist concentration. GlyRs were blocked by 1 μM strychnine (both barrels). Estimated EC50 values were 472 μM (n = 1.8,A = 1.02), 15.1 mM (n = 1.9,A constrained to 1), 44.2 mM (n constrained to 1.8, A constrained to 1), and 236 mM (nconstrained to 1.8, A constrained to 1). Duration of agonist pulses was 1 ms in all experiments. Number of patches (sometimes with multiple measurements) is indicated in parentheses.

The antagonist strychnine blocked the glycine-activated current with an IC50 of 39 nM (Fig. 4, A and C), whereas the GABA-activated current was affected only at several-fold higher concentrations (IC50 = 1.5 μM). Conversely, bicuculline blocked the GABA-activated current with an IC50 of 0.58 μM (Fig. 4D), whereas the glycine-activated current was almost unaffected by concentrations up to 30 μM (estimated IC50 = 144 μM). Thus, strychnine and bicuculline blocked spinal GlyRs and GABAARs with high (but not absolute) selectivity.

Although synaptic corelease of glycine and GABA is the most likely interpretation of the above results, an alternative possibility would be that a third transmitter, such as β-alanine or taurine, coactivated postsynaptic GlyRs and GABAARs (14). We therefore compared the affinity of both types of receptors for glycine and GABA with that for β-alanine and taurine. Glycine and GABA activated GlyRs and GABAARs at low concentrations; the half-maximal activating concentrations (EC50's) were 441 μM and 472 μM, respectively, with 1-ms agonist pulses (Fig. 4, E and F). In contrast, GABA at concentrations as high as 10 mM was unable to activate GlyRs (Fig. 4E), and conversely, glycine was unable to activate GABAARs (Fig. 4F). β-Alanine appeared to be a partial agonist for GlyRs (Fig. 4E), but it hardly activated GABAARs (Fig. 4F). Finally, taurine showed very low affinity for both types of receptors (Fig. 4, E and F). These results further support the hypothesis that glycine and GABA are coreleased at inhibitory spinal synapses (15).

We next examined GlyR- and GABAAR-mediated components in spontaneous miniature IPSCs (Fig. 5) (6, 16), presumably generated by the release of the contents of single synaptic vesicles (17). To enhance the difference in decay time course between the two components, we added the benzodiazepine flunitrazepam (18). Flunitrazepam (2 μM) slowed selectively the decay of the GABAAR-mediated component of the evoked compound IPSC; the decay of the GABAAR-mediated component was, on average, prolonged by a factor of 1.65 ± 0.18 (six motoneurons), whereas that of the GlyR-mediated component was almost unchanged (factor of 0.95 ± 0.05 change, four cells). In the presence of 2 μM flunitrazepam, subsets of individual miniature IPSCs and the average miniature IPSC showed a bi-exponential decay (Fig. 5A). In the presence of 5 μM bicuculline, the decay time course of the miniature IPSCs was fast, similar to that of the first component in control conditions, whereas in the presence of 400 nM strychnine, the decay was slow, comparable to that of the second component in the absence of antagonists (Fig. 5, B and C).

Figure 5

GlyR- and GABAAR-mediated components in spontaneous miniature IPSCs distinguishable in the presence of flunitrazepam. (A) Three examples of spontaneous miniature IPSCs in control conditions recorded from a putative motoneuron in the whole-cell configuration in the presence of 500 nM tetrodotoxin and 2 μM flunitrazepam, superimposed with the average from 49 events. The inset shows a representation of the recording configuration. (B) Average miniature IPSCs recorded in the absence (uppertrace) and presence of either 5 μM bicuculline or 400 nM strychnine (lower traces, from 17 and 18 events, respectively). Same cell as shown in (A). (C) Comparison of the decay time constants of the dual-component average miniature IPSCs in the absence of blockers (hatched bars, respective amplitude contributions plotted within bars) with those recorded in the presence of 5 μM bicuculline (open bars) or 400 nM strychnine (filled bars). Number of motoneurons is indicated in parentheses. (D and E) Scatter plots of the amplitude of the GABAAR-mediated component (IGABAAR) against that of the GlyR-mediated component (IGlyR) in miniature IPSCs obtained by template fit analysis in the absence of blockers [(D), 560 events] and in the presence of 5 μM bicuculline [(E), open circles, 200 events] or 400 nM strychnine [(E), filled circles, 125 events]. Dashed lines indicate the +2σ ranges of the data points in the presence of antagonists. Positive amplitude values indicate inward currents. Occasional negative amplitude values arose from noise superimposed on miniature events. Data pooled from five motoneurons.

To quantify the amplitude contributions of the two components in individual miniature IPSCs, we obtained average GlyR- and GABAAR-mediated miniature IPSC templates in the presence of either bicuculline or strychnine. Subsequently, individual miniature events recorded from the same neuron before antagonist application were fit with the sum of these two templates, multiplied by variable amplitude factors (16). Scatter plots of the amplitudes of the two components indicated that in control conditions a large proportion of data points fell into the first quadrant (Fig. 5D), indicating a high percentage of dual-component events. In contrast, data points in the presence of antagonists were clustered around the axes (Fig. 5E), showing that dual-component events were largely absent. Defining the criterion separating the dual- and monocomponent events at the ±2σ ranges of the data points in the presence of either bicuculline or strychnine, we classified 44% of miniature IPSCs in control conditions as dual-component, 41% as pure GlyR-mediated, and 15% as pure GABAAR-mediated events. If miniature IPSCs reflect the fusion of individual synaptic vesicles (17), these results would indicate both the costorage of glycine and GABA in the same synaptic vesicle and the colocalization of GlyRs and GABAARs in the same postsynaptic density.

In conclusion, we showed that unitary IPSCs and subsets of miniature IPSCs at inhibitory synapses in the spinal cord are dual-component events comprising GlyR- and GABAAR-mediated components. On the basis of the strychnine sensitivity (Fig. 4C) and the different decay time constants of GlyR- and GABAAR-mediated currents, we estimate that the GlyR-mediated component dominates the IPSC peak amplitude (81%), whereas the GABAAR-mediated component dominates the inhibitory postsynaptic charge (68%).

The most likely explanation for the dual-component nature of the unitary IPSC is the quantal corelease of glycine and GABA from a single spinal interneuron. None of the plausible transmitter candidates coactivates both types of receptors. The dual-component nature of a subset of miniature IPSCs suggests the strictest possible form of corelease, that is, from the same synaptic vesicle. This is compatible with the observation that vesicular transporters at inhibitory synapses accept both glycine and GABA as substrates (19). Whether glycine- and GABA-mediated cotransmission is a general principle of inhibition that also applies to subcortical neuronal circuitries (for example, the auditory pathway) (20) remains to be addressed.

Glycine- and GABA-mediated cotransmission could support the precise regulation of the time course of the postsynaptic conductance by the relative amount of glycine and GABA released from the presynaptic interneuron. This could be of critical importance for motor coordination (21) and the generation of locomotor patterns (22). Cotransmission would also enable feedback control of transmitter release by presynaptic GABAB receptors, which may not be possible at pure glycinergic synapses. Finally, cotransmission may enable compensatory mechanisms in genetic GlyR subunit defects (23).

  • * To whom correspondence should be addressed. E-mail: jonasp{at}ruf.uni-freiburg.de †On sabbatical leave from II. Physiologisches Institut, D-69120 Heidelberg, Germany.

REFERENCES AND NOTES

View Abstract

Navigate This Article