Neuromodulation of spike-timing precision in sensory neurons

Abstract

The neuropeptide allatostatin decreases the spike rate in response to time-varying stretches of two different crustacean mechanoreceptors, the gastropyloric receptor 2 in the crab Cancer borealis and the coxobasal chordotonal organ (CBCTO) in the crab Carcinus maenas. In each system, the decrease in firing rate is accompanied by an increase in the timing precision of spikes triggered by discrete temporal features in the stimulus. This was quantified by calculating the standard deviation or "jitter" in the times of individual identified spikes elicited in response to repeated presentations of the stimulus. Conversely, serotonin increases the firing rate but decreases the timing precision of the CBCTO response. Intracellular recordings from the afferents of this receptor demonstrate that allatostatin increases the conductance of the neurons, consistent with its inhibitory action on spike rate, whereas serotonin decreases the overall membrane conductance. We conclude that spike-timing precision of mechanoreceptor afferents in response to dynamic stimulation can be altered by neuromodulators acting directly on the afferent neurons.

Figure 1.

Schematic diagram of the two experimental preparations. A, GPR2 preparation from the crab C. borealis. The cpv3a muscle is attached to and stretched by the movement of a loudspeaker while spikes from the GPR2 axon are recorded by a suction electrode on the gastropyloric nerve (gpn). B, Isolated preparation of the CBCTO from the fifth leg of the common green shore crab C. maenas. The elastic strand of the CBCTO is connected to an electromechanical puller monitored by an optical position sensor (opto) and controlled by a PID controller operating in an active length feedback mode. Intracellular recordings are made from a single chordotonal afferent axon in the chordotonal nerve distal to the thoracic ganglion, which is partially outlined at the bottom right of the figure.

Figure 2.

Allatostatin causes a decrease in spike rate and a decrease in spike-timing jitter in response to repetitions of a constantly varying stimulus to the GPR2 neuron. A, Extracellular recordings from the gpn during 20 Hz low-pass-filtered white noise stimulus, spike time raster, and PSTH with 50 repetitions of the stimulus in control saline (left) and in 10⁻⁶ m AST (right). The dashed line denotes the threshold for identification of potential events; actual events are labeled by “+.” B, Histogram of deviations of individual spike times from the average spike time for precisely timed events in nine experiments (50 trials each). The distribution of spikes in AST is narrower and taller indicating lower jitter than in control. Gray color indicates overlap between control and AST distributions. Over the nine experiments, a total of 2730 event spikes were measured under control conditions whereas 3379 were measured in AST. Inset shows spike time differences for the single event (#) in the experiment shown in A. The black and gray curves are Gaussian fits to the control and AST data, respectively. C, Effect of 10⁻⁶ m AST on spike-timing jitter (p = 0.009; N = 9). D, Effect of 10⁻⁶ m AST on spike rate (p = 0.013; N = 9). E, Jitter plotted against reliability for all events measured in the 9 experiments (96 control events, 113 AST events). There is no significant correlation between jitter and reliability when considering all events (n = 209; CC = −0.0878; p = 0.205). F, Mean jitter plotted against mean spike rate for each of the nine GPR2 experiments under control conditions (filled circles) and in AST (open circles). There is a significant positive correlation between jitter and mean rate (CC = 0.838; p < 0.001).

Figure 3.

In the CBCTO afferent of C. maenas, allatostatin causes a decrease in spike rate and a decrease in spike-timing jitter in response to a constantly varying stimulus. A, A 100 ms segment of the intracellular recording, stimulus, spike time raster, and PSTH for 60 repetitions of a 5 s duration, 140 Hz low-pass-filtered white noise stretch in control saline (left) and in 10⁻⁶ m AST (right). The dashed line denotes event threshold. B, Histogram of deviations of individual spike times from the average spike time for all matched events (n = 118) in this experiment under both control and AST modulated conditions. Overlap between the two distributions is shown in gray. In this experiment, a total of 10,132 event spikes were measured under control conditions, whereas 9603 were measured in AST. Inset, Histogram for a single event marked # in A. The black and gray curves are Gaussian fits to the control and AST data, respectively. C, Effect of 10⁻⁶ m AST on the spike-timing jitter in the experiment shown in A (p < 0.001; n = 118). D, Effect of 10⁻⁶ m AST on CBCTO spike-timing jitter in 28 chordotonal afferents (p < 0.001). E, Effect of 10⁻⁶ m AST on spike rate (p = 0.024; N = 28). F, Jitter plotted against reliability for 500 control and 500 AST events randomly selected from all of the events (2003 control; 1874 AST) measured in 28 experiments. Plotting all events would have resulted in overlapping points that would have obscured the structure of the distribution of data. There is no significant correlation between jitter and reliability when considering all events in all experiments (n = 3877; CC = −0.0222; p = 0.166).

Figure 4.

Serotonin causes an increase in spike jitter and mean spike rate in response to a constantly varying stimulus in the CBCTO. A, A 200 ms segment of the intracellular recordings, stimulus, spike time raster, and PSTH for 70 repetitions of a 5 s duration, 140 Hz low-pass-filtered white noise stimulus in control saline (left) and in 10⁻⁶ m serotonin (right). Dashed line denotes event threshold. B, Histogram of deviations of individual spike times from the average spike time for all matched events (n = 87) in this experiment under both control and serotonin-modulated conditions. Overlap between the two distributions is shown in gray. In this experiment, a total of 7327 event spikes were measured under control conditions, whereas 8728 were measured in serotonin. Inset, Histogram for a single event marked # in A. The black and gray curves are Gaussian fits to the control and serotonin data, respectively. C, Effect of 10⁻⁶ m serotonin on the spike-timing jitter in the experiment shown in A (p < 0.001; n = 87). D, Effect of 10⁻⁶ m serotonin on CBCTO spike-timing jitter in nine chordotonal afferents (p < 0.001). E, Effect of 10⁻⁶ m serotonin on spike rate (p = 0.0016; N = 9). F, Jitter plotted against reliability for 500 control and 500 serotonin events randomly selected from all of the events (1322 control; 1392 serotonin) measured in nine experiments. A small but statistically significant negative correlation between reliability and jitter is obtained when considering all events in all experiments (n = 2714; CC = −0.0392; p = 0.031).

Figure 5.

The lack of positive correlations between jitter and mean firing rate for the CBCTO afferents indicates that effects on jitter are not mediated by effects on rate. A, Mean jitter plotted against mean spike rate for each CBCTO experiment in control (filled circles) and AST (open circles). There is a significant negative correlation between jitter and mean rate (CC = −0.369; p = 0.005). B, Mean jitter plotted against mean spike rate for each CBCTO experiment in control (filled circles) and serotonin (5-HT; open circles). There is no significant correlation between jitter and rate (CC = −0.201; p = 0.423).

Figure 6.

Allatostatin induces a conductance increase and serotonin induces a conductance decrease in the chordotonal afferents. A, Current-clamp recording of a CBCTO afferent voltage (V_m) in response to injected current (I_e) in the presence of 2 × 10⁻⁷ m TTX (left) and in TTX and 10⁻⁶ m AST (right). B, Steady state I–V curve showing decrease in membrane resistance (slope) in the presence of AST. C, Current-clamp recording of a CBCTO afferent voltage in response to injected current in the presence of 2 × 10⁻⁷ m TTX (left) and in TTX and 10⁻⁶ m serotonin (5-HT; right). D, Steady-state I–V curve showing an increase in membrane resistance (slope) in the presence of serotonin. Each I--V curve was measured three times under each condition; the error bars correspond to SDs in the voltage measured for each current injection.