Mechanism for modulation of nicotinic acetylcholine receptors that can influence synaptic transmission

Only recently has it been appreciated that neuronal nicotinic ACh receptors (NnAChRs) are highly permeable to Ca2+ and are modulated by Ca2+ in a dose-dependent manner. These findings suggest that Ca2+ could have roles in cholinergic synaptic plasticity. We report a possible mechanism for Ca(2+)-initiated synaptic plasticity that differs from the intracellular Ca2+ cascade associated with plasticity of glutamatergic synapses. Rapid changes in external Ca2+ modulate cholinergic spontaneous synaptic currents in superior cervical ganglionic sympathetic neurons. Inhibition of cholinergic currents by chlorisondamine, which blocks only open channels and becomes trapped in the pore, showed that the modulation is not by a mechanism that activates a previously unresponsive population of NnAChRs. Rather, single-channel recordings with ganglionic NnAChRs from chromaffin cells indicated that Ca2+ directly alters the probability of the channels being open. We hypothesize from the results that activity-dependent decreases in external Ca2+, which occur throughout the nervous system, could directly underlie a rapid negative-feedback mechanism that decreases the responsiveness of NnAChRs at synapses. When external Ca2+ is decreased, presynaptic Ca2+ currents and transmitter release also are diminished. Thus, several mechanisms could combine to potently and rapidly depress synaptic nicotinic receptors until the external Ca2+ concentration recovers.


Only recently
has it been appreciated that neuronal nicotinic ACh receptors (NnAChRs) are highly permeable to Ca*+ and are modulated by Ca*+ in a dose-dependent manner. These findings suggest that Ca2+ could have roles in cholinergic synaptic plasticity.
We report a possible mechanism for Ca*+-initiated synaptic plasticity that differs from the intracellular Ca*+ cascade associated with plasticity of glutamatergic synapses. Neuronal nicotinic ACh receptors (NnAChRs) are found throughout the CNS and PNS. It was recently shown that NnAChRs have a high Ca*+ permeability (Adams and Nutter, 1992;Mulle et al., 1992a;Vernino et al., 1992Vernino et al., , 1994. Permeability ratio measurements indicated that a NnAChR subtype composed of the (~7 subunit has an even higher Ca*+ permeability, comparable to the permeability of the NMDA subtype of glutamate receptors (SCguCla et al., 1993). At glutamatergic synapses, Ca*+ influx through NMDA receptors enhances protein kinase activity, which is an early step in the process of long-term potentiation (Malinow et al., 1988;Malenka et al., 1989;Madison et al., 1991). At highly active cholinergic synapses, Ca 2+ influx through NnAChRs could activate similar intracellular mechanisms, but such mechanisms have not yet been observed at central cholinergic synapses.
The present study focuses on the mechanism and potential for synaptic plasticity arising from another aspect of Caz+ modulation that is unique to neuronal nAChRs. External Caz+ enhances NnAChR responses in a dose-dependent manner (Mulle et al., 1992a;Vernino et al., 1992). Applications of nicotinic agonist to a cholinoceptive neuron induce progressively smaller responses as external Ca 2+ is decreased. The modulation is seen in a concentration range of Ca*+ from 0 to 10 mM and is strongest at the physiologic level of about 1 mM after correcting for the permeation properties of nAChRs for Ca2+ (Decker and Dani, 1990;Mulle et al., 1992b;Vernino et al., 1992). This modulation of NnAChRs by external Ca2+ can be of biological importance because high synaptic activity can produce millimolar reductions in external Ca2+ (Benninger et al., 1980;Pumain and Heinemann, 1985;Mody and Heinemann, 1986;Heinemann et al., 1990;Livsey et al., 1990). These activity-dependent changes in extracellular Ca*+ were measured using electrodes that averaged over a relatively large volume. Because the local density of current in the small cleft of an active synapse can be higher than the average current density in a large volume of neurons, the Ca*+ reduction within a synaptic cleft could be significantly greater than estimated using extracellular electrodes. We report here that changes in external Ca*+ act on NnAChRs to modify spontaneous cholinergic synaptic currents. Therefore, activity-dependent reductions in external Ca*+ could cause a rapid negative feedback onto cholinergic synaptic transmission. Our results indicate that Ca2+ does not activate a previously unresponsive population of Nn-AChRs. Rather, Ca2+ increases the likelihood that responsive NnAChRs will be open in the presence of agonist.

Materials and Methods
Cell culture. Sympathetic neurons were isolated from the superior cervical ganglia (SCG) of neonatal rats. During the first few days in culture, these neurons are adrenergic. When grown in coculture with cardiac muscle cells or in conditioned media, the neurons become cholinergic (Hawrot and Patterson, 1979). We grew cardiac cells on the bottom of culture dishes. Hearts from newborn rats were minced, dissociated with divalent-free solutions in 1 mg/ml collagenase, triturated, washed, and plated. The cardiac cells matured and covered the bottom of the dish in a few days. Then, poly-o-lysine/collagen coated cover-glass slips were placed on top of the cardiac cells. SCG neurons were plated onto the cover glass. After 7 or more days in culture, cholinergiclcholinoceptive synapses formed among the SCG neurons on the cover glass. The neurons were obtained from a pair of SCG by  Ca'+ were blocked. C, After washing the chlorisondamine away for I5 set, agonist was applied again, first in 5 mu then in 1 mu Ca'+. The current is three times larger in 5 mM Ca'+. After 10 agonist applications and 25 min of washing, the currents have recovered more from chlorisondamine blockade, but the enhancement by 5 mM Ca"+ is still threefold. The holding potential was always -70 mV.
modulate NnAChRs. First, Ca'+ inhux into the cell through nAChRs is strongly voltage dependent (Decker and Dani, 1990;Vernino et al., 1994), but the modulation by Ca'-is not (Fig. I C). At -70 mV there is a large Cal' influx into the cell, but at +70 mV very little Ca'+ enters. If the currents from different chromaffin cells are normalized to their amplitudes in 1 tnM Ca", then the enhancement of the currents in IO ITIM Ca" is 2.6 2 0.3 (n = 28) at -70 mV, 3.1 + 0.2 (II = 32) at -50 mV, and 2.8 + 0.3 (n = 15) at +70 mV. SCG neurons are similarly modulated at -70 mV: 2.7 2 0.3 (II = 3). Second, the modulation by Ca'+ is unaffected by internal perfusion of the cell with Ca" chelators. In Figure I and in most cases. the cells were perfused with IO mM BAPTA, but the same enhancement and lack of voltage dependence were seen with 20 mM BAPTA (n = 3) 20 mM EDTA (n = 3), and IO mM EDTA + 10 mM EGTA (n = 2) with or without 4 mM ATP-Mg. These data were all combined to give the averages listed above. In summary, voltage and intracellular CaL + buffering do not influence the modulation, indicating that Ca" is acting externally and not through an intracellular enzyme cascade (Mulle et al., 1992b;Vernino et al., 1992).
Figure I also indicates other types of modulation and variability that are seen superimposed on the enhancement of NnAChR currents by external CaL+. Although the current increase caused by Ca'+ is similar in Figure I, C and D, the rate and degree of desensitization are much different. Desensitization is thought to arise from intrinsic properties of the receptor channel. Figure I shows that for each cell desensitization is greater with increased Ca?+ and at more negative potentials. as has been found for muscle nAChRs (Fiekers et al., 1980;LCna and Changeux, 1993). There is, however, much variability from cell to cell. This variability is thought to arise from intercellular enzyme activity, such as phosphorylation speeding the rate of muscle nAChR desensitization (Huganir, 1988). Thus, the currents can differ significantly in appearance while still giving comparable magnitudes of external Ca'+ modulation.  4. The probability of a Nn-AChR channel being open is increased in elevated Ca". Single-channel currents are shown from an outside/out patch of membrane excised from an ndrenal chromaftin cell and exposed to IO FM DMPP in a solution containing either I mM or IO mM Ca". The patch was moved alternately between the two solutions and was held in each solution for 847 sec. Each record shows the first 5 set of single-channel activity for the first IO solution changes. The holding potential was -70 mV. Consistent with other experiments that are not shown, the single-channel activity runs down with time in both solutions, but more rapidly in I mM Ca". The amplitude of the single-channel events is larger in I mM C$ , but the probability of being open is greater in IO mM Ca". Integration of the current over the full length of the recording revealed that NnAChRs are 6 times more likely to be open in the IO IllM CaZA solution.
1 Ca 10 Ca damine, was used to determine whether Ca?' modulates already active NnAChRs or converts an unresponsive population of channels. Chlorisondamine enters NnAChR channels after they open and blocks them like a cork in a bottle. Also, chlorisondamine becomes trapped in the pore after the nAChR closes and continues to block until the channel is reopened and chlorisondamine has time to diffuse away (shown with muscle nAChRs by Neely and Lingle, 1986).
Our strategy for testing whether or not Ca" enhances currents by opening a new population of NnAChRs was as follows. Use chlorisondamine to block all the NnAChR channels that open in a solution of I mM Ca'-and, then, see whether large nicotinic currents can still be seen in a solution of 5 ITIM Ca". If large nicotinic currents are seen in 5 mM Ca", then that current is passing through a new population of unblocked NnAChRs that did not open in the low Ca" solution. In separate experiments. chlorisondamine's mechanism of action was examined with NnAChRs to verify the work of Neely and Lingle ( 1986) on muscle nAChRs. In Figure 2, a SCG sympathetic neuron is stimulated by 56 FM nicotine producing four nicotinic currents each separated by I min. The solid bars represent I min of exposure to 5 pM chlorisondamine that is terminated 200 msec before the next agonist application. Figure 2 shows that chlorisondamine does not inhibit NnAChRs that have not opened. Therefore, chlorisondamine can be used to identify NnAChRs that have opened by blocking them and becoming trapped within the pore.
In Figure 3A application of 32 FM DMPP to a SCG sympathetic neuron induces a current that is thrqe times larger in a solution containing 5 mM Ca?+ as compared with the current in I ITIM Ca". Then, repeated applications of agonist in the presence of chlorisondamine were made until all the channels that opened in I mM Ca?+ became blocked (Fig. 3B). Next, chlorisondamine was washed away, and the modulation by Ca" was retested (Fig. 3C). After I5 set of wash, the first application of agonist activates only a small current in I mM Cal+. The enhancement of that current by 5 mM Ca" is threefold as before, but the size of the current is small. After IO exposures to agonist and 25 min of recovery from chlorison-----kpA IS damine, the currents are larger, but the modulation by Ca" remains the same (threefold). Since only the channels that open in I mM Ca" were blocked by chlorisondamine.
it is clear that elevated Ca" does not cause the activation of another population of previously unresponsive channels. If another population of channels had become active, chlorisondamine would not have decreased the extra current originally seen in 5 mM Ca" (Fig. 3A). Similar results giving identical conclusions were obtained in four other cells using 5 FM chlorisondamine with either 32 pM DMPP or 56 FM nicotine. The recovery from blockade by chlorisondamine continued during the course of these experiments, but the recovery was slow because the neurons where always held at -70 mV in these experiments.
Recovery was faster and more complete when the holding potential was depolarized to drive chlorisondamine out of the channel.
Although Ca?' does not increase the population of responsive NnAChRs, single-channel recordings in excised patches of membrane indicated that Ca" does increase the probability of responsive NnAChRs being open. Figure 4 shows singlechannel currents induced in a single excised patch of membrane from a chromaffin cell. The agonist concentration. IO p.~ DMPP, remained constant as the patch was moved back and forth between solutions containing I or IO IIIM Ca2 + Each record represents the first 5 set after a solution change. The fraction of time that a NnAChR channel is open is greater in IO mM Ca". Over the course of the experiment, integration ot the single-channel currents for the same length of time indicated that a channel is open six times more often in IO mM Ca'+. The frequency of channel openings (bursts) increased 4.0 + I .2 fold, and the burst length increased slightly. I .7 & 0.3 fold. In a total of four patches, integration of the currents gave a 3.4 ? I.2 fold increase in the probability of a channel being open in IO mM versus I mM Ca'-. This increased probability of opening quantitatively accounts for the Ca' ' modulation we see with macroscopic currents. The amplitude of the singlechannel currents is larger in low Ca'., as has been explained previously (Decker and Dani, 1990: Vernino et al., 1992. Calcium modulation in chromaffin cells appears the same as for SCG sympathetic neurons (Vernino et al., 1992) but much effort was made to obtain comparable single-channel results with membrane patches excised from SCG neurons. Stable single-channel records were impossible with the SCG neurons, however, because the patches had a low density of channels that "run down" very rapidly. In much less than a minute no channel events could be seen. Therefore, we could not obtain single-channel numbers as described for the chromaffin cells, but channel openings were more frequent and could be followed longer in high calcium with the SCG neurons. It also was not possible to characterize the run down process is SCG neurons because the process was so rapid, but we have characterized the effect in habenula neurons (Lester and Dani, 1994). Mulle et al. (1992b) had similar problems with run down in habenula neurons, where they compared 5-10 set of NnAChR activity in 0 versus 4 IIIM Cal+. Similar to our findings with chromaftin cells, they found a threefold increase in the opening frequency of the NnAChRs in higher CaL+ explained the modulatory effect.
Externcd Cu. ' alters cholinrrgic-synaptic trunsmission consistrrlt bt'ith the postsynuptic modulation of NrzAChRs Since external Ca'+ modulates NnAChRs, changes in external Ca'+ could alter cholinergic synaptic transmission. Figure SA shows cholinergic, nicotinic spontaneous synaptic currents (SSCs) between SCG sympathetic neurons that were grown in culture to form cholinergic/cholinoceptive synapses (Hawrot and Patterson, 1979;Furshpan et al., 1986). In the presence of the NnAChR antagonist, hexamethonium, no large SSCs were ever seen. Propagated excitation was inhibited by blocking voltage-dependent Na' channels with I pM TTX. The SSCs were obtained at a holding potential of ~70 mV while the external solution was alternated 18 times between agonist-free solutions containing either 2 or 5 mM Ca' +. The amplitudes of the SSCs were used to build histograms with bin widths of 3 pA. The most populated bin for each of the two histograms represents the most commonly observe SSC amplitude: 7 pA in 2 mM CaL' and IO pA in 5 mM Ca" (Fig. SB). When all the current amplitudes were averaged together, the bin containing the average-size current was at I7 pA in 2 ITIM Ca" and at 24 pA in 5 mM Ca" (Fig. 5C). After hundreds of SSCs were obtained, 32 IJ,M nicotine was exogenously applied to the voltage-clamped neuron in 2 mM CaL+ and then in 5 mM CaL' (Fig.  5D). As described earlier, external Ca" enhanced the wholecell current induced by nicotine: the current was 700 pA in 2 mM CaL' and 1070 pA in 5 mM Ca". The ratio 1070/700 can be defined as a modulation factor (,f') that accounts for changes in the amplitudes of the SSCs caused by external Cal' modulating NnAChRs. If the amplitude of the most common SSC in 2 nlM Ca' is multiplied by ,f; the product equals the amplitude of the most common SSC in 5 IIlM Ca' ' : 7 pA X 1070/ 700 = 10.7 = 10 pA. The same relationship holds for the average-size SSC: 17 pA X 1070/700 = 26.0 = 24 pA.
The two SSC amplitude distributions are shown in Figure  6A. The number of SSCs in the peak bin is normalized to 100 (ordinate). The distribution in 2 mM Ca" lies at smaller amplitudes than the distribution in 5 mM CaL ' . Then, all the amplitudes of the distribution obtained in 2 mM Ca" were multiplied by ,f; and the distributions were plotted again in Figure   6B. Such skewed and wide SSC distributions are not seen for mature neuromuscular junctions, but they are common in the central nervous system. Bekkers et al. (1990) investigated the source of the variability for glutamatergic synapses in cultured hippocampal neurons. They reported that the variability of quanta1 size is the major determinant. Hume and Honig (1991) proposed similar mechanisms to explain the extremely broad amplitude histograms obtained from preganglionic and sympathetic neurons. Those reports also discuss other potential sources of the variability.