Glutamate and acetylcholine corelease at developing synapses W.-C. Li*, S. R. Soffe, and Alan Roberts School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, United Kingdom Edited by Roger A. Nicoll, University of California, San Francisco, CA, and approved September 16, 2004 (received for review July 7, 2004)
Most neurons release a single fast-acting low-molecular-weight transmitter at synapses to activate and open postsynaptic ion channels. We challenge this principle with evidence for corelease of the two major excitatory transmitters, glutamate and acetylcholine (ACh), from single identified neurons in the developing frog tadpole spinal cord. Whole-cell patch electrodes were used to record from single spinal neurons. When action potentials and inhibition were blocked, spontaneous miniature excitatory postsynaptic currents (mEPSCs) were recorded. These were fully blocked only by joint application of glutamate [␣-amino-3hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) and N-methyl-D-aspartate receptor (NMDAR)] and nicotinic ACh receptor (nAChR) antagonists. Fast nAChR and slow NMDAR mEPSCs were isolated pharmacologically. We then show that some mEPSCs have both the fast nAChR rise and slow NMDAR decay and conclude that some individual synaptic vesicles corelease glutamate and ACh. Whole-cell recordings from pairs of neurons were then used to identify the spinal interneurons coreleasing the two excitatory transmitters. One anatomical class of interneuron with descending axons was found to excite other spinal neurons by activating nAChR, AMPAR, and NMDAR simultaneously at its synapses. Although Jonas, Bischofberger, and Sandkuhler [Jonas, P., Bischofberger, J. & Sandkuhler, J. (1998) Science 281, 419 – 424] showed that the inhibitory transmitters GABA and glycine can be coreleased at spinal synapses, the Xenopus tadpole provides a case where the two main CNS excitatory transmitters are released from single vesicles, and where the presynaptic neuron coreleasing two transmitters has been identified.
T
here is now abundant evidence that neurons can release more than one type of transmitter molecule at their synapses. The most usual case is when a single low-molecular-weight transmitter that binds with postsynaptic receptors to open ionotropic channels is released with a peptide (1). Despite this type of corelease, typically of an amino acid and a peptide, the general rule remains that neurons release only a single low-molecularweight transmitter. The exceptions to this rule all involve the inhibitory transmitter GABA. Recordings from single neurons often show miniature postsynaptic potentials or currents that are thought to result when a single synaptic vesicle releases its contents spontaneously without presynaptic impulses (2, 3). In the neonatal rat spinal cord and brain stem, some miniature inhibitory postsynaptic currents (mIPSCs) result from the corelease of the inhibitory transmitters GABA and glycine from single presynaptic vesicles (4–6). The separate components of these mIPSCs can be distinguished by their different time courses and pharmacology. Similar recordings from cultured neonatal rat spinal dorsal horn neurons have shown corelease of GABA and the excitatory transmitter ATP at their synapses when stimulation evokes an action potential in a single presynaptic neuron (7). Finally, indirect evidence suggests that GABA and the excitatory transmitter ACh may both be released from amacrine cells in the retinas of mammals and chicks (8). The present evidence might suggest that corelease of lowmolecular-weight transmitters is a feature peculiar to GABA transmission and notably absent are cases involving the major CNS excitatory transmitter glutamate. However, recent evi15488 –15493 兩 PNAS 兩 October 26, 2004 兩 vol. 101 兩 no. 43
dence on the distribution of the proteins that load glutamate into synaptic vesicles suggests the possibility of glutamate corelease. Glutamate-loading proteins are expressed in cholinergic neurons in the rat brain, for example in the striatum (9, 10), and in esophageal motor nerve terminals in the mouse (11). To try to establish the occurrence of corelease of glutamate and ACh, we have used the hatchling frog tadpole, where there is already evidence that neurons in the spinal cord receive both glutamatergic and nicotinic cholinergic synaptic drive during swimming activity, and where the spinal cord neurons are limited in number and well defined anatomically (12–14). We have reexamined the synaptic excitation to spinal neurons by making whole-cell patch recordings and have obtained evidence for the spontaneous corelease of glutamate and ACh. By recording from pairs of neurons, we then identified the presynaptic neurons coreleasing these two excitatory transmitters. Materials and Methods Whole-Cell Patch Recording. Details of the recording methods have
been given recently (15). Briefly, Xenopus tadpoles at stage 37兾38 (Fig. 1a) were anaesthetized with 0.1% MS-222 (3aminobenzoic acid ester, Sigma), immobilized in 10 mM ␣-bungarotoxin saline, then pinned in a bath of saline (115 mM NaCl兾3 mM KCl兾3 mM CaCl2兾2.4 mM NaHCO3兾10 mM Hepes, adjusted with 5 M NaOH to pH 7.4). Saline with 0 mM Mg2⫹ was used so N-methyl-D-aspartate receptor (NMDAR)-mediated components could be seen (15). Skin and muscles over the right side of the spinal cord were removed and a middorsal cut made along the spinal cord to open the neurocoel. Small cuts were made in the wall of the neurocoel on the right side to expose more ventral neurons. The left side of the spinal cord (between the first and 10th muscle segments) was removed to allow better visibility of neurons on the right side. The tadpole was then repinned in a small 700-l recording chamber with saline flow of ⬇2 ml per minute. Exposed neuronal cell bodies were seen by using a ⫻40 water immersion lens with bright-field illumination on an upright Nikon E600FN microscope. Antagonists were applied close to the recorded neuron soma by using gentle pressure to solution in a pipette with tip diameter of 10–20 m or dropped into a 100-l well upstream of the recording chamber. Drugs used were 2,3-dihydroxy-6-nitro-7-sulfamoylbenzo[f]quinoxaline (NBQX), D-(-)-2-amino-5-phosphonopentanoic acid (D-AP5) (Tocris Cookson, Bristol, U.K.), bicuculline, str ychnine, tetrodotoxin, D -tubocurarine, mecamylamine (Sigma), and dihydro--erythroidine (DHE; Research Biochemicals International, Natick, MA). Patch pipettes were filled with 0.1% neurobiotin and 0.1% Alexa Fluor 488 (Molecular Probes) in intracellular solution This paper was submitted directly (Track II) to the PNAS office. Abbreviations: ACh, acetylcholine; nAChR, nicotinic ACh receptor; NBQX, 2,3-dihydroxy6-nitro-7-sulfamoylbenzo[f]quinoxaline; mEPSCs, miniature excitatory postsynaptic currents; DHE, dihydro--erythroidine; dINs, descending interneurons; NMDAR, N-methylD-aspartate receptor; AMPAR, ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor. *To whom correspondence should be addressed. E-mail:
[email protected]. © 2004 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0404864101
were corrected before making recordings. Signals were recorded with an Axoclamp 2B (Axon Instruments, Union City, CA) in a conventional bridge or continuous single-electrode voltage-clamp mode, acquired with SIGNAL software through a CED 1401 Plus interface with a sampling rate of 10 kHz (Cambridge Electronic Design, Cambridge, U.K.). Off line analyses were made with MINITAB, and MINIANAL software (Synaptosoft, Decatur, GA). To separate them into subgroups, miniature excitatory postsynaptic currents (mEPSCs) were selected individually as negative events crossing a threshold (usually 5 pA) set relative to a local baseline. A minimum area threshold was used to reject rapid transients such as electrical artifacts. Rise times of mEPSCs (20 – 80% peak) were calculated by using the MINIANAL software. Decay times (peak to 37% peak) were calculated by fitting exponential curves (single or double, as appropriate) to the decay of averaged mEPSCs and then by using the equations of these curves to reconstruct the falling phase. All values are given as mean ⫾ SEM. Unless stated otherwise, statistical analysis was by ANOVA, by using one-way ANOVA or a General Linear Model with Tukey’s multiple comparisons. Experiments comply with United Kingdom Home Office regulations. from the patch electrode. Tadpoles were fixed after experiments in 2% glutaraldehyde in 0.1 M PBS (pH 7.2 at ⬇4°C). After rinsing with 0.1 M PBS, the animals were: (i) washed twice in 1% Triton X-100 in PBS for 15 min; (ii) incubated in a 1:300 dilution of extravidin peroxidase conjugate (Sigma) in PBS containing 0.5% Triton X-100 for 2–3 h; (iii) washed again in PBS; (iv) presoaked in 0.08% diaminobenzidine in PBS (DAB solution) for 5 min; (v) moved to 0.075% hydrogen peroxide in DAB solution for 5 min; (vi) washed in running tap water. The brain and spinal cord were then dissected free with the notochord and some ventral muscles, dehydrated, cleared in methyl benzoate and xylene, and mounted whole between two coverslips. Neurons were observed by using a ⫻100 oil immersion lens and traced by using a drawing tube or photographed. To compensate for shrinkage during dehydration, all measurements in this paper have been corrected by multiplying by 1.28 (14). Results Properties of mEPSCs in Spinal Neurons. Using immobilized XenoFig. 1. Evidence for corelease of glutamate with ACh during spontaneous mEPSCs. (a) Diagrams of the hatchling Xenopus tadpole and CNS in lateral view to show the position of a recorded spinal neuron. (b) Slow time-base record to show spontaneous mEPSCs in 3 M tetrodotoxin, 1 M strychnine, and 20 M bicuculline. (c) In 2.5 M NBQX, fast nAChR mEPSCs (green) and slow NMDAR mEPSCs (orange) are separated pharmacologically by addition of 25 M D-AP5 and 1 M DHE, respectively. Right traces are averages to show the characteristic shapes of the nAChR (n ⫽ 19) and NMDAR (n ⫽ 45) mEPSCs. Scales are the same as similar records in d and e. (d) Slow record of mEPSCs in 2.5 M NBQX and, underneath, examples of three types (arrows) separated by using their rise and decay times. (e) Averages to show three types of mEPSCs in NBQX [presumed nAChR, n ⫽ 10; NMDAR, n ⫽ 32; and dual-component nAChR ⫹ NMDAR (brown), n ⫽ 21]. ( f) Plot of mEPSC rise time vs. decay time. The mEPSCs form three groups: nAChR (green), NMDAR (orange), and dualcomponent nAChR and NMDAR (brown). Open triangles represent mEPSCs separated pharmacologically as in c. Filled circles represent mEPSCs in the presence of NBQX, separated by using rise and decay times and averaged for each neuron (n ⫽ 8 neurons) as in e. Lines enclose members of the three groups revealed by cluster analysis. (g) Examples of dual-component mEPSCs in 1 M DHE (black, AMPAR ⫹ NMDAR) and in 2.5 M NBQX (brown, nAChR ⫹ NMDAR), recorded from the same neuron. Scales are the same as similar records in d.
(100 mM K-gluconate兾2 mM MgCl2兾10 mM EGTA兾1 mM Hepes兾3 mM Na2ATP兾0.5 mM NaGTP, adjusted to pH 7.3 with KOH) and had resistances ⬇10 M⍀. Junction potentials Li et al.
pus tadpoles (Fig. 1a), we made whole-cell patch recordings from 78 ventral spinal neurons. Neurobiotin filling showed that they were types of neuron that are normally active during swimming locomotion (16). When 3 M tetrodotoxin blocked action potentials and 1 M strychnine and 20 M bicuculline blocked synaptic inhibition, we recorded spontaneous mEPSCs at a holding potential of ⫺60 mV (Fig. 1b). These were fully blocked only by the joint application of glutamate and nicotinic ACh receptor antagonists (2.5 M NBQX, 25 M D-AP5, and 1 M DHE). We therefore looked for evidence of coactivation of NMDAR and nicotinic ACh receptor (nAChR) in a sample of 1,656 mEPSCs in a sample of eight neurons. To see the time course of nAChR-mediated mEPSCs, these were isolated pharmacologically by using 2.5 M NBQX and 25 M D-AP5 to block the ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) and NMDAR EPSCs and averaged to the rising phase (Fig. 1c, green traces). The fast nAChR-mediated mEPSCs had a fast rise and decay (Table 1) and were blocked by the nAChR antagonist DHE (1 M). Slow NMDAR mEPSCs isolated in NBQX and DHE (Fig. 1c, orange traces) had longer rise and decay times and were blocked by D-AP5 (25 M). These observations showed that nAChR and NMDAR mEPSCs have distinctive waveforms (Fig. 1c). If corelease of ACh and glutamate was indeed occurring, we PNAS 兩 October 26, 2004 兩 vol. 101 兩 no. 43 兩 15489
NEUROSCIENCE
Processing for Neuron Anatomy. Dye filling was by passive diffusion
Table 1. Analysis of mEPSCs Separated by
Transmitter receptor
n
% occurrence
Rise 20–80% peak, ms
Decay peak to 37%, ms
Pharmacology Pharmacology Time course Time course Time course
nAChR NMDAR nAChR NMDAR nAChR ⫹ NMDAR
193 246 115 261 182
— — 9–38 14–80 12–50
0.8 ⫾ 0.1 4.3 ⫾ 1.1 0.9 ⫾ 0.1 4.1 ⫾ 0.7 1.1 ⫾ 0.1
4.3 ⫾ 0.5 114.8 ⫾ 28.5 5.7 ⫾ 0.5 67.9 ⫾ 15.4 83.9 ⫾ 23.9
n, the total number of mEPSCs analyzed. For each neuron, the percentage occurrence of the three types of mEPSCs separated from the mixed population in NBQX was measured; % occurrence gives the range across the eight neurons. Rise and decay times were analyzed for all five groups of mEPSCs: nAChR and NMDAR isolated pharmacologically and putative nAChR, NMDAR, and dual-component nAChR ⫹ NMDAR mEPSCs separated from the mixed population recorded in NBQX (as above). Overall, rise and decay times differed significantly among the five groups (both P ⬍ 0.001). There was no significant difference between either the rise or decay times for the two nAChR groups (both P ⬎ 0.99) or for the two NMDAR groups (P ⫽ 0.99 and 0.34). The rise time of the dual-component nAChR ⫹ NMDAR group also showed no significant difference from either nAChR groups (both P ⬎ 0.98) but was significantly shorter than for the two NMDAR groups (both P ⬍ 0.002). The decay time for the dual-component nAChR ⫹ NMDAR group was significantly longer than for the nAChR groups (both P ⬍ 0.004) but not significantly different to either NMDAR group (both P ⬎ 0.78).
expected to record ‘‘dual-component’’ mEPSCs combining the fast nAChR-mediated rise and a slow NMDAR-mediated decay. Because fast AMPAR mEPSCs are very similar in waveform to those mediated by nAChRs, we first blocked AMPAR with 2.5 M NBQX to simplify analysis. Dual-component mEPSCs were still present. At least 50 mEPSCs were recorded from each of eight neurons. The mEPSCs were then sorted manually on the basis of their time course by using the nAChR and NMDAR averages obtained pharmacologically, as templates. For each neuron, the mEPSCs fell into three groups that were separately averaged to their rising phases. The characteristics of the three groups were defined by using the averages obtained from each neuron (Fig. 1 d, e, and g; Table 1): (i) nAChR mEPSCs with a fast rise and decay, (ii) NMDAR mEPSCs with a slower rise and decay, (iii) dual-component mEPSCs with a fast rise like the nAChR mEPSCs and a slow decay like the NMDAR mEPSCs. When the mEPSCs in NBQX were plotted on a rise-time vs. decay-time graph, together with the data for pharmacologically isolated mEPSCs, the three main groups were clear (Fig. 1f ). With the exception of a single pharmacologically isolated NMDAR point, these groups were endorsed by cluster analysis (K means clustering). The simplest explanation for the dualcomponent mEPSCs with fast rise and slow decay in NBQX is that they result from simultaneous release of glutamate and ACh from single vesicles. The first step to reveal dual-component nAChR and NMDAR mEPSCs was to block AMPARs (see above). However, to check that dual-component AMPAR and NMDAR mEPSCs were also present, we used a second sample of six neurons where fast nAChR mEPSCs were present in 2.5 M NBQX and 25 M D-AP5. We applied 1 M DHE alone to these neurons and found dual-component mEPSCs with fast rise (0.36 ⫾ 0.02 ms) and slow decay (75.5 ⫾ 54.2 ms), presumably mediated by simultaneous AMPAR and NMDAR activation (Fig. 1g). We applied 2.5 M NBQX alone to four of these neurons; they all showed fast rise, slow decay mEPSCs, in this case resulting from nAChR and NMDAR activation as described above. The Neurons Releasing ACh and Glutamate. In the hatchling frog tadpole, some spinal neurons receive both glutamatergic and cholinergic excitation during swimming (12, 13). Could some of this result from corelease of both transmitters from individual neurons? To answer this question, we made whole-cell recordings from pairs of spinal neurons to examine the synaptic connections of spinal interneurons with descending axonal projections. As in other vertebrates, tadpole descending interneu15490 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0404864101
rons (dINs) are thought to provide the glutamatergic excitation that drives spinal neuron activity during locomotion (17, 18). In 48 paired recordings, anatomically identified dINs (Fig. 2a) produced monosynaptic excitation of more caudal ventral spinal neurons. These were identified from their anatomy as motoneurons and premotor interneurons, such as reciprocal inhibitory commissural interneurons (16), and were all active during fictive swimming. Current-clamp recordings showed that dIN impulses evoked large EPSPs (mean maximum amplitude, 13.1 ⫾ 1.7 mV; n ⫽ 10 pairs) at short and constant delays (1.32 ⫾ 0.10 ms; n ⫽ 10 pairs) in postsynaptic neurons (Fig. 2b). In two cases, these EPSPs were blocked by the joint application of 25 M D-AP5 and 2.5 M NBQX, but in the other seven cases tested, the EPSPs were abolished only when the glutamate antagonists were applied with 1–5 M DHE. This suggests the corelease of glutamate and ACh at these synapses. Details of evoked glutamate and ACh corelease were analyzed in paired recordings from dINs with the postsynaptic neuron in current clamp (n ⫽ 14) or voltage clamp (n ⫽ 18; at a holding potential of ⫺60 mV; Fig. 2c). Seven interactions were completely blocked simply by joint application of glutamate antagonists (2–5 M NBQX ⫹ 20–50 M D-AP5), but in 25 interactions, a fast component remained (Fig. 2 c and d, green trace, Fig. 2e). This was blocked when an nAChR antagonist was added (Fig. 2 c and d, gray trace, Fig. 2e; 1–10 M DHE,: n ⫽ 18). In the remaining cases, we confirmed that the fast component was also blocked by other nAChR antagonists (20 – 40 M dtubocurarine, n ⫽ 6; 10 M mecamylamine, n ⫽ 1). The composition of the evoked EPSCs was analyzed in detail for eight of the interactions. As with the spontaneous mEPSCs, this was done by applying antagonists in different combinations and revealed three clear components (Fig. 2d). Of these, the nAChR component (isolated by using 25 M D-AP5 ⫹ 2.5 M NBQX) and the AMPAR component (isolated by using 25 M D-AP5 ⫹ 1 M DHE) both had a fast rise and fast decay (nAChR, 1.41 ⫾ 0.20 ms and 8.39 ⫾ 1.16 ms; AMPAR, 0.89 ⫾ 0.25 ms and 4.79 ⫾ 0.81 ms), whereas the NMDAR component (isolated by using 1 M DHE ⫹ 2.5 M NBQX) had a slow rise and slow decay (4.83 ⫾ 0.72 and 100.2 ⫾ 19.9 ms). To evaluate whether the connections were monosynaptic, delays from the presynaptic dIN spike to the onset of the control EPSC were measured for the same eight interactions by using averages of 15–20 EPSCs in each case. Delays ranged from 0.94 to 1.78 ms (mean, 1.30 ⫾ 0.10 ms; n ⫽ 8 neuron pairs; see measures above for EPSPs). Because both neurons were filled with neurobiotin, the distances between pre- and postsynaptic Li et al.
NEUROSCIENCE
Fig. 2. Spinal neurons corelease glutamate with ACh. (a) Diagram of the tadpole CNS in lateral view with two recorded spinal neurons seen below in the photograph. The dIN descending axon (arrowhead) contacts a more caudal neuron (cN at arrow). (b) Records in current– clamp show that current induced spikes in the dIN lead to EPSPs in the cN (black) with a possible nAChR component revealed by 2.5 M NBQX ⫹ 25 M D-AP5 (green). The proximity of the two electrodes leads to direct electrical coupling between them to produce a dIN spike artifact in the cN trace (at arrow). (c) Applying antagonists in sequence (as shown) reveals the components of EPSCs evoked in the caudal neuron by current evoked dIN impulses (at the top). A small fast current remains in 2.5 M NBQX and 25 M D-AP5 (green). This is blocked by adding 1 M DHE (gray). A fast current remains in 25 M D-AP5 ⫹ 1 M DHE (red) and a slow current in 2.5 M NBQX ⫹ 1 M DHE (orange). (d) In the same caudal neuron, averaged records show the control EPSC (black; 20 per trace), and individual components separated pharmacologically: nAChR component (green) in NBQX and D-AP5; AMPAR component (red) in DHE and D-AP5; NMDAR component (orange) in NBQX and DHE. In all three antagonists, the EPSC is abolished (gray). (e) EPSCs evoked in caudal neurons (n ⫽ 8) by dIN spikes are significantly reduced in amplitude but not blocked by the glutamate antagonists NBQX (2.5 M) and D-AP5 (25 M). They are blocked by addition of 1 M DHE. ( f) In spinal sensory interneurons, the glutamate antagonists NBQX (2.5 M) and D-AP5 (25 M) together completely block EPSCs evoked by spikes in skin sensory neurons, whereas DHE (10 M) has no significant effect (n ⫽ 6 interactions; ns, not significant; ** significant at P ⬍ 0.001).
neurons were measured after processing (range, 0.1–0.4 mm; mean, 0.22 ⫾ 0.04 mm). Assuming a synaptic delay of 0.5 ms, the calculated conduction velocities were 0.10–0.76 m䡠s⫺1 (mean, 0.32 ⫾ 0.08 m䡠s⫺1). These values are typical of estimates for the tadpole at this stage of development (15, 19). We therefore conclude that the connections recorded between dINs and other spinal neurons were monosynaptic. Delays were then measured for individual fast EPSC components (AMPAR and nAChR) after isolating them pharmacologically (as above). If they shared the same presynaptic source, Li et al.
we should expect the delays to be the same. For each neuron, the delays to the AMPAR component (1.28 ⫾ 0.10 ms) and to the nAChR component (1.38 ⫾ 0.11 ms) were not significantly different (one-way ANOVA; P ⫽ 0.329). We therefore conclude that the glutamatergic and cholinergic components of evoked EPSCs do indeed come from the same presynaptic source. The relatively low amplitude and slow rise of the NMDAR component in combination with coupling artifacts from the presynaptic recording electrode made it impossible to accurately judge the onset of the NMDAR current. Delays for the NMDAR comPNAS 兩 October 26, 2004 兩 vol. 101 兩 no. 43 兩 15491
ponents could not therefore be determined with sufficient accuracy so were not analyzed. The synapses that we have just described from dINs to more caudal spinal neurons contrast with other connections in the tadpole where skin afferents monosynaptically excite spinal sensory interneurons (Fig. 2f; ref. 20). Here, DHE had no measurable effect on unitary EPSCs (mean amplitudes: 48 – 142 pA) evoked by presynaptic skin afferent spikes (n ⫽ 6), whereas NBQX and D-AP5 totally blocked both these unitary EPSCs (n ⫽ 6) and the compound EPSCs evoked by stimulating the sensory processes of the afferents in the skin (n ⫽ 3), indicating that the excitation here is mediated by glutamate alone.
It remains possible that it may also occur at synapses made by the descending axons of brain stem neurons. It is also possible that corelease may occur at other excitatory spinal synapses at particular stages of their development, as is true for inhibitory synapses (6). The discovery of glutamate and ACh corelease raises some intriguing possibilities. It could play an important developmental role, such as the induction of particular postsynaptic changes. At many developing glutamatergic synapses, although not at some Xenopus tadpole spinal cord synapses (21), NMDARs are present early in development, and AMPARs appear only later (23). If such silent synapses (24) were present in the spinal cord, then corelease of ACh could lead to effective fast excitation to unblock voltage-gated NMDARs before the insertion of AMPARs. However, it is important to remember that, by this stage of tadpole development, the spinal networks controlling swimming are fully functional, and dINs were already thought to provide glutamatergic drive for swimming (18). The cholinergic component of excitation was missed prev iously, because an AChR ant agon ist, d tubocurarine, was used in all experiments to immobilize the tadpoles. Because the time course of the AMPAR and nAChR-mediated components of excitation are similar (Fig. 2d), it seems unlikely that these two components will complement each other like the fast and slower inhibition produced by glycine and GABA when these are coreleased (25). Glutamate and ACh do both contribute excitatory drive to spinal neurons during swimming in the hatchling tadpole (13), but the respective roles of AMPAR and nAChR-mediated components are not yet clear. It is clear that the nAChR-mediated mEPSCs are significantly faster (rise and decay times, 0.8 and 4.3 ms) than the dIN impulse-evoked nAChR-mediated EPSCs (1.4 and 8.4 ms; t test P ⫽ 0.02 and 0.01). The dual-component AMPAR ⫹ NMDAR mEPSCs similarly had a shorter rise time than evoked AMPAR EPSCs (0.34 and 0.89 ms, respectively). This would be expected if these central synapses operated like the frog neuromuscular synapse, where impulse-induced vesicular release is a stochastic process with variable latency (26). We suggest that corelease of fast-acting low-molecular-weight transmitters may be a more general feature during CNS development, not confined to mammals and to inhibitory synapses releasing GABA and glycine (4–7). However, it may be significant that reported cases of corelease from single vesicles are in developing brain stem or spinal cord rather than in higher brain centers, even though nicotinic receptors are important in the brain (27). It is interesting that, at this early stage of frog development, excitatory spinal interneurons (dINs) and motoneurons share a similar ventral location, many anatomical features, activity during swimming (16), and now even transmitters. Is this indicative of an early stage in the differentiation of these two groups of neurons from a common precursor (28, 29)?
Discussion We have looked for evidence of the corelease of glutamate and ACh at synapses in the spinal cord of the developing frog tadpole. In interpreting the first series of experiments, we assume that miniature EPSCs recorded when action potentials are blocked by tetrodotoxin result from the spontaneous release of transmitter from single synaptic vesicles (2, 3). The critical observation is that some of these individual mEPSCs, recorded when AMPARs are blocked, have both the fast rise characteristic of AChR-mediated EPSCs and the slow decay of NMDAR EPSCs. On this basis, we conclude that ventral neurons in the spinal cord of the hatchling frog tadpole receive some excitation where glutamate and ACh are coreleased from the same vesicles. The dual-component mEPSCs resulting from this corelease constitute as many as half of those recorded in ventral neurons and indicate that NMDA and nACh receptors are colocalized subsynaptically. Because all of the recorded neurons also have AMPAR mEPSCs, it is very likely that corelease of glutamate and ACh will activate three different postsynaptic ligand gated ion channels (AMPAR, NMDAR, and nAChR). By recording from pairs of spinal neurons, we have established that one source for the corelease of glutamate and ACh is a class of spinal interneurons with descending ipsilateral axonal projections (dINs; ref. 16). In 78% of dINs, intracellular current injection to evoke a single action potential led to unitary EPSPs or EPSCs where AMPARs, NMDARs and nAChRs were coactivated. Again, the most plausible explanation for the multicomponent nature of these unitary postsynaptic responses is the corelease of glutamate and ACh at single synapses, and at least in part from single vesicles as we have demonstrated for spontaneous mEPSCs. Because there are so few classes of neuron in the tadpole spinal cord at this early stage of development (16), it is likely that dINs are also the source of these spontaneous mEPSCs. The other main sources of excitation from within the spinal cord can be eliminated because they appear to involve only one transmitter. Primary skin afferent synapses onto spinal neurons are fully blocked by glutamate antagonists (Fig. 2f; ref. 21), as are the synapses from sensory pathway interneurons (20). Central motoneuron synapses are fully blocked by ACh antagonists (22). Corelease is therefore not a general feature of excitatory synapses made by spinal neurons at this stage of development.
We thank Ted Bullock, John Isaac, George Mackie, Rob Meech, Moo-Ming Poo, and Keith Sillar for helpful comments on drafts of this paper; Tim Colborn and Jenny Maxwell for technical help; and the Wellcome Trust for support.
1. Kupfermann, I. (1991) Physiol. Rev. 71, 683–732. 2. Katz, B. (1969) The Release of Neural Transmitter Substances (Liverpool Univ. Press, Liverpool, U.K.). 3. Bekkers, J. M. & Stevens, C. F. (1989) Nature 341, 230–233. 4. Jonas, P., Bischofberger, J. & Sandkuhler, J. (1998) Science 281, 419–424. 5. O’Brien, J. A. & Berger, A. J. (1999) J. Neurophysiol. 82, 1638–1641. 6. Nabekura, J., Katsurabayashi, S., Kakazu, Y., Shibata, S., Matsubara, A., Jinno, S., Mizoguchi, Y., Sasaki, A. & Ishibashi, H. (2004) Nat. Neurosci. 7, 17–23. 7. Jo, Y. H. & Schlichter, R. (1999) Nat. Neurosci. 2, 241–245. 8. Duarte, C. B., Santos, P. F. & Carvalho, A. P. (1999) J. Neurosci. Res. 58, 475–479.
9. Fremeau, R. T., Jr., Burman, J., Qureshi, T., Tran, C. H., Proctor, J., Johnson, J., Zhang, H., Sulzer, D., Copenhagen, D. R., Storm-Mathisen, J., et al. (2002) Proc. Natl. Acad. Sci. USA 99, 14488–14493. 10. Gras, C., Herzog, E., Bellenchi, G. C., Bernard, V., Ravassard, P., Pohl, M., Gasnier, B., Giros, B. & El Mestikawy, S. (2002) J. Neurosci. 22, 5442–5451. 11. Kraus, T., Neuhuber, W. L. & Raab, M. (2004) Neurosci. Lett. 360, 53–56. 12. Perrins, R. & Roberts, A. (1995) J. Neurophysiol. 73, 1013–1019. 13. Zhao, F. Y., Wolf, E. & Roberts, A. (1998) J. Physiol. 511, 887–901. 14. Li, W. C., Perrins, R., Soffe, S. R., Yoshida, M., Walford, A. & Roberts, A. (2001) J. Comp. Neurol. 441, 248–265. 15. Li, W. C., Soffe, S. R. & Roberts, A. (2002) J. Neurosci. 22, 10924–10934.
15492 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0404864101
Li et al.
22. 23. 24. 25. 26. 27. 28. 29.
Perrins, R. & Roberts, A. (1995) J. Physiol. 485, 135–144. Wu, G., Malinow, R. & Cline, H. T. (1996) Science 274, 972–976. Isaac, J. T. (2003) Neuropharmacology 45, 450–460. Russier, M., Kopysova, I. L., Ankri, N., Ferrand, N. & Debanne, D. (2002) J. Physiol. 541, 123–137. Katz, B. & Miledi, R. (1965) Proc. R. Soc. London Ser. B 161, 483–495. Jones, S., Sudweeks, S. & Yakel, J. L. (1999) Trends Neurosci. 22, 555–561. Hartenstein, V. (1989) Neuron 3, 399–411. Jessell, T. M. (2000) Nat. Rev. Genet. 1, 20–29.
NEUROSCIENCE
16. Roberts, A. (2000) Brain Res. Bull. 53, 585–593. 17. Kiehn, O., Harris-Warick, R. M., Jordan, L. M., Hultborn, H. & Kudo, N. (1998) Neuronal Mechanisms for Generating Locomotor Activity (New York Academy of Sciences, New York), Vol. 860. 18. Dale, N. & Roberts, A. (1985) J. Physiol. 363, 35–59. 19. Clarke, J. D. W., Hayes, B. P., Hunt, S. P. & Roberts, A. (1984) J. Physiol. 348, 511–525. 20. Li, W. C., Soffe, S. R. & Roberts, A. (2003) J. Neurosci. 23, 9068–9077. 21. Rohrbough, J. & Spitzer, N. C. (1999) J. Neurosci. 19, 8528–8541.
Li et al.
PNAS 兩 October 26, 2004 兩 vol. 101 兩 no. 43 兩 15493
