Anatomy
The Brain Descending Auditory-Responsive Neuron (B-DARN1) shows a prominent ring-like arborization in the ventral protocerebrum, completely distinct and ventral to the calyx of the mushroom body (Fig. 1). This ring-like arborization is an anatomical feature of several brain auditory interneurons, including the axonal arborisations of the ascending interneurons AN1 and AN2 (Wohlers and Huber 1982), which provide the principal auditory input to the brain, and at least several local sound pattern-recognition brain neurons such as LNs 1–5 (Kostarakos and Hedwig, 2012, Schöneich et al. 2015).
The most anterior and finely branched part of the ring arborization of B-DARN1 (Fig. 1a, d) lies just under the ventral surface of the brain (Fig. 1c, e). Larger antero-medial and postero-medial branches form the enclosed ring-like structure (Fig. 1a, b, d). An axon projects initially dorsally out of the ring-arborization (Fig. 1c, e) before running posteriorly in a dorsal region of the brain towards the circum-oesophageal connectives (Fig. 1b, c). In all recordings the injected dyes faded before, or just after reaching the circum-oesophageal connective and it was not possible to reveal the neuron’s output regions. Within the ring arborisation the smaller branches present a mixture of smooth and beaded processes (Fig. 1d). The heaviest concentration of beaded processes appeared to be in the postero-lateral part of the ring near where the axon arises, whereas the antero-medial arbor was composed predominately of a mass of fine smooth branches. The arborisation pattern of B-DARN1 closely resembles that of the axonal arborisation pattern of the ascending auditory interneuron AN1, except that both stains of B-DARN1 revealed a lateral process that ended in somata in the lateral protocerebrum posterior to the stalk of the optic lobe, in a region occupied by the somata of several other auditory interneurons (Kostarakos and Hedwig, 2012). Both stains of B-DARN1 indicate the presence of two adjacent cell bodies, perhaps indicative of two dye-coupled neurons. In Fig. 1a there may be two primary neurites, but what appears to be a pair of cell bodies may be a single waisted cell. There appear to be two somata in the second stain (Fig. 1b, d) but the more medial soma is diffuse, pallid, and very superficial to the surface of the brain and had no primary neurite. This cell body was not apparent in the image stack taken on the Zeiss Axiophot fluorescence microscope used to reconstruct the neuron (Fig. 1f). The anatomical data presented here were made from a recording in which the electrode was placed well away from the ring-arborization (arrow Fig. 1f).
Response to natural chirp-like song patterns
The response of B-DARN1 was tested to song patterns modelled on natural G. bimaculatus chirps, consisting of four 20 ms pulses at 4.8 kHz separated by 20 ms silent pulse intervals with a chirp interval of 140 ms (chirp rate 3.57 s− 1). B-DARN1 responded vigorously to binaural stimulation by calling song (Fig. 2), with bursts of action potentials (AP) following each sound pulse, which surmounted sustained depolarizations with little repolarization during pulse intervals in one of the recordings (Fig. 2a-d). The spiking response followed the chirp pattern, but not as tightly as in the ascending auditory interneuron AN1 (Wohlers and Huber 1982, Hardt and Watson 1994, Kostarakos and Hedwig 2012), and the pulse intervals were difficult to discern in stimulus histograms or raster plots (Fig. 2b, d). At 75 dB SPL, the first sound pulse led to a burst of nearly evenly spaced AP approximately every 6 ms. The response to the subsequent pulses was more complex and biphasic. It consisted of an initial AP approximately 22 ms after the onset of a sound pulse followed by a gap of about 10 ms in which few AP occurred, but after which several further AP were evoked (Fig. 2b). This biphasic pattern was particularly clear in the less intense response to songs played at 55 dB SPL (Fig. 2c), in which seven distinct peaks occurred in the raster plot and histogram (Fig. 2d). In the second recording of B-DARN1 (Fig. 2e), the membrane potential repolarised after each response to a pulse and the pulse structure could be clearly discerned in the AP response.
At 75 dB SPL, the number of APs per pulse was consistent across repeated stimulation, a latency of 20 ms was used as a cut-off to isolate the response from successive pulses (Fig. 2f; see below). 75% of first pulses evoked 5 AP, 60% of second pulses evoked 4 AP, 80% of third pulses evoked 5 AP, and 60% of fourth pulses evoked 5 AP, with all other AP counts being ± 1 of these modal values (N = 20). The decrease in AP count between the first and second pulses was significant (Fisher’s exact test, P = 4 × 10− 4), as was the subsequent increase again in the third pulse (Fisher’s exact test, P = 9 × 10− 5). Even after 58 repeated chirps, the numbers of evoked APs were 5, 4, 4, 4 for each pulse, indicating that B-DARN1 copied sound patterns with high fidelity for extended periods (circles, Fig. 2f).
Decreasing the sound intensity of chirps to 65 dB SPL from 75 dB SPL led to a modest but non-significant decrease in response (Fig. 2g), evoking 16.6 ± 1.8 AP per chirp as compared to 18.5 ± 1.0 (Mann-Whitney test, N = 20, Z = 1.88, P = 0.060). A further reduction to 55 dB SPL (Fig. 2c, d, g) led to a significant decrease in the response, with only 10.1 ± 1.7 AP per chirp (Mann-Whitney test, N = 20, Z = 5.32, P < 1 × 10− 7).
The spiking response was preceded by a Post-Synaptic Potential (PSP), with an amplitude of 3.2 ± 0.3 mV which occurred 20.1 ± 0.6 ms (N = 20, means ± SD) after the onset of sound and showed little temporal variation between stimuli (Fig. 2h). This clearly discernible PSP identifies this neuron as physiologically distinct from AN1, since only axonal AP can be seen in brain recordings of AN1. The latency of B-DARN1 spikes was more variable, with 22.8 ± 0.9 ms after the onset of sound, but the box-plot (Fig. 2h) suggests a more asymmetric distribution, with some AP delayed for a longer period (median time of first AP 22.4 ms, interquartile range 22-23.9 ms). In approximately one in twenty chirps the initial PSP supported an AP.
Response to changes in pulse duration
The effect of pulse duration was investigated by playing songs in which the pulse durations of the four-pulse chirps were systematically varied from 5 to 100 ms, while the pulse interval was kept at 20 ms, and the chirp interval at 140 ms. Trains of 12 chirps were played of each chirp type, separated by 2 s silent intervals. Song patterns were presented in pseudo-random order so that long- and short-pulse duration chirp types alternated. The first response of a new stimulus type was excluded from analysis since this response was typically much greater (up to 30%) following a recovery in the silent period between chirp trains.
The spiking activity in B-DARN1 reflected the pulse duration within chirps (Fig. 3). The strongest responses were to the first pulse within a chirp (black symbols, Fig. 3a), and the number of evoked AP decreased in each successive pulse (grey to white symbols, Fig. 3a). Pulses shorter than 20 ms evoked similar numbers of AP, suggesting a strong phasic initial response. As pulse duration increased, so did the number of evoked AP. Between 20 and 50 ms there was a rising curve, consistent with an ongoing strong phasic response, but for the longest pulse lengths the relationship became approximately linear, suggesting a shift to a tonic response during prolonged stimulation. The data could be fitted with a non-linear regression of the form y0 + axb, with R2s of 0.95–0.96 (Table 1), where y0 is the predicted minimum number of AP that can be evoked and axb describes the phaso-tonic character of the response. The y0 values suggest a minimum burst of 5 AP for the first sound pulse, decreasing to 3 AP for the fourth (Table 1). The constant “a” ranged from 0.01–0.04, while the exponent “b” ranged from 1.58 to 1.25 across the four pulses of a chirp (Table 1). This relationship suggests an initial strong phasic response for pulses up to 50 ms long which gave way to a tonic response in longer pulses which evoke proportionately fewer AP per ms of sound pulse.
Table 1
Estimates and significances of non-linear regressions of the form y = y0 + axb fitted to the number of evoked action potentials (y) in relation to sound pulse length (x). See Fig. 4A.
Pulse | Y0 | a | b | R2 | F2,162 | P |
1 | 5.0 ± 0.19 | 0.01 ± 0.003 | 1.58 ± 0.07 | 0.96 | 1970.1 | 8.2 × 10− 116 |
2 | 3.7 ± 0.18 | 0.02 ± 0.006 | 1.40 ± 0.06 | 0.96 | 1818.8 | 4.4 × 10− 113 |
3 | 3.2 ± 0.19 | 0.04 ± 0.010 | 1.25 ± 0.06 | 0.95 | 1668.1 | 3.8 × 10− 110 |
4 | 2.99 ± 0.17 | 0.03 ± 0.077 | 1.31 ± 0.06 | 0.95 | 1730.5 | 2.2 × 10− 111 |
The phaso-tonic characteristics of B-DARN1 were revealed in its response to sound stimuli consisting of 500 ms long sound pulses separated by 140 ms silent intervals (Fig. 3b; mean ± SEM firing rate across the stimulus in Fig. 3c). The peak spike rate occurred between 28–34 ms after the onset of the sound pulse when B-DARN1 fired at 322 ± 16 AP s− 1, but activity halved over the subsequent 60 ms, and then decreased more slowly as the stimulus persisted, reaching a tonic firing rate after approximately 225 ms, which by the end of the stimulus was 106 ± 8 AP s− 1.
Response to changes in pulse interval
The effect of changes in pulse interval was tested with different trains of chirps consisting of four 20 ms pulses separated by pulse intervals that were varied in 5 ms steps from 5 to 60 ms. The chirp interval was kept at 140 ms. Each chirp pattern was presented twelve times sequentially and different chirp types were presented in pseudo-random order so that sequences with short pulse intervals were followed by chirps with long pulse intervals.
When the pulse interval was 5 ms the neuronal response to individual sound pulses merged together into a single burst of AP surmounting a single sustained depolarisation (5 ms, Fig. 4a). With a pulse interval of 60 ms, the neuron almost completely repolarised between sound pulses and each response to a pulse was sharply defined by well separated bursts of AP surmounting a sustained depolarisation (Fig. 4a). Songs with intermediate pulse intervals were characterised by a partial repolarisation between bursts (e.g., 15 and 25 ms, Fig. 4a, and Fig. 2), with a spiking response to each individual sound pulse occurring.
Pulse interval affected the total number of AP evoked by a chirp even though the overall duration of sound was a constant 80 ms in each chirp type, likely due to a recovery from adaptation occurring over the time course of a chirp (graph Fig. 4a; Kruskal-Wallis test, Chi-squared11 = 107.1, P = 3.2 × 10− 18). The number of AP per chirp was 18.3 ± 1.6 AP when the pulse interval was 5 ms, 32.5% lower than the 27.1 ± 1.2 AP evoked when the pulse interval was 55 ms. Numbers of AP were similar for pulse intervals from 10 to 20 ms and then increased near linearly for pulse intervals between 25 to 45 ms, before reaching a saturated response for pulse intervals greater than 45 ms (Fig. 4a).
Adaptation over the course of a chirp was analysed by comparing the peak instantaneous AP frequency evoked after the first and fourth pulses for different pulse intervals (Fig. 4b). The peak AP frequency of the first pulse did not show any significant variation, indicating a constant level of activity following the consistent inter-chirp interval of 140 ms (Fig. 4b grey bars; average 293.3 ± 31.9 AP s− 1 ; ANOVA, F11, 120 = 0.652, P = 0.781). The peak AP frequency in response to the fourth pulse however, increased significantly with the pulse interval (Fig. 4b black bars; ANOVA, F11, 120 = 20.677, P = 1.3 × 10− 22). Post-hoc paired t-tests indicated that the peak AP frequencies of the fourth pulses were all significantly less than those of first pulses for pulse intervals < 40ms, but for larger pulse intervals there were no significant differences (Table 2). In response to the fourth sound pulse, the minimum peak instantaneous AP frequencies occurred for the 5 ms pulse interval chirps (200.3 ± 24.7 AP s− 1); as the interval increased the peak firing rate progressively increased until it approached about 290 AP s− 1, which was similar to the firing rate in response to the first pulse in a chirp.
Table 2
Results of paired t-tests comparing the highest instantaneous firing rate of B-DARN1 during the first and fourth pulse of chirps in which the pulse duration was held at a constant 20 ms, but the pulse interval was varied from 5 to 60 ms.
Pulse Interval (ms) | t10 | P |
5 | 4.87 | 6.5 × 10− 4 |
10 | 11.98 | 3.0 × 10− 7 |
15 | 6.88 | 4 × 10− 5 |
20 | 12.19 | 2.5 × 10− 7 |
25 | 3.0 | 0.013 |
30 | 2.5 | 0.032 |
35 | 2.57 | 0.028 |
40 | 0.79 | 0.448 |
45 | 0.05 | 0.980 |
50 | 0.48 | 0.640 |
55 | 2.03 | 0.070 |
60 | 0.18 | 0.860 |
Response to attractive and unattractive chirp patterns
Although natural songs show little variation from the 40 ms pulse-period pattern, female G. bimaculatus will show clear phonotaxis to some artificial song-patterns, particularly if the first pulse is unnaturally short, whereas the length of the final pulse in a chirp is less critical (Hedwig and Sarmiento-Ponce, 2017). Conversely, reversing these artificial patterns so that chirps lead with a long pulse and end with a short pulse, generates unattractive songs with significantly reduced phonotaxis. The response of B-DARN1 to attractive chirps consisting of 5, 20 and 50 ms pulses (Fig. 5a,b) was compared to its response to a unattractive pattern with 50, 20 and 5 ms pulses (Fig. 5c, d). The pulse intervals were 20 ms and the chirp interval was 185 ms (chirp rate 3.33 s− 1).
As expected from the phaso-tonic responses to songs with different pulse durations (Fig. 3), the longer pulses within these attractive and unattractive chirps evoked larger numbers of AP (Fig. 5e), but the short pulses were over-represented in the spiking activity. The average duration of spiking activity lasted for 14.3 ± 4.4 ms for the 5 ms pulses (2.86× longer than the duration of the stimulus), 27.6 ± 4.6 ms for the 20 ms pulses (1.38× longer than stimulation duration) and 52.2 ± 3.7 ms for the 50 ms pulses (1.04× longer than stimulus duration). The mean number of AP increased from 3.0 ± 1.2 for 5 ms stimuli, through to 6.4 ± 0.8 for 20 ms stimuli and 13.5 ± 2.1 for 50 ms stimuli, so while the pulse duration increased tenfold, the number of AP only increased by a factor of 4.5 (Fig. 5e).
The order of presentation also affected the number and frequency of AP elicited by each pulse, with evidence of adaptation occurring over the course of a chirp (Fig. 5e, f). In the attractive chirps, the leading 5 ms pulse evoked 3.9 ± 1.1 AP compared to only 2.2 ± 0.4 AP evoked by the 5 ms in the unattractive chirp when it was in the last position (Mann-Whitney U test, Z30 = 4.66, P = 3 × 10− 6). Similarly, the 50 ms pulse evoked more AP when it was in the leading position in the unattractive chirp (15.1 ± 1.2) compared to when it was last in the attractive chirp (11.4 ± 0.7 AP; Mann-Whitney, Z30 = 4.92, P = 1 × 10− 6). Also, when the central 20 ms pulse was preceded by the long 50 ms pulse in the unattractive pulse there was a significant decrease in the number of evoked AP (6.0 ± 0.5) compared to the number of AP in the attractive pulse (6.9 ± 0.7 AP; Mann-Whitney, Z30 = 3.31, P = 9.3 × 10− 4). This effect led to a remarkable difference in average AP firing rate across the different chirp types (Fig. 3f): in the attractive chirps the mean firing rate was very similar across all three pulses (means 231 ± 41, 225 ± 25 and 231 ± 19 AP s− 1; Kruskal Wallis test, Chi-squared2 = 0.23, P = 0.891), whereas for the unattractive chirp the firing rate over the course of the chirp decreased progressively with each sound pulse by a total of 31% (275 ± 18, 247 ± 23 and 189 ± 42 AP s− 1; Kruskal Wallis test, Chi-squared2 = 30.4, P = 1.3 × 10− 7).
Effect of sound direction
In the previous tests, B-DARN1 had been stimulated binaurally by two simultaneously active speakers positioned 36° relative to the longitudinal axis of the cricket but activating each speaker alternately revealed strong direction-dependent responses to natural chirp patterns (Fig. 6). The left speaker was ipsilateral to the recording electrode (see Fig. 1) while the active speaker changed every 10 chirps (forming a series of alternating blocks, Fig. 6b, c). Sound from the ipsilateral side caused lower AP firing rates in B-DARN1, as shown by the instantaneous firing rate (Fig. 6a lowest trace) and the average firing rate to each sound pulse of a chirp (Fig. 6f). On average, one more AP per sound pulse was evoked when presented by the right/contralateral speaker (5.5 ± 0.8) than from the ipsilateral/left speaker (4.5 ± 0.6; t158 = 8.67, P = 5 × 10− 15), and there was also a difference in the duration of the burst of AP (21.8 ± 3.0 ms when stimulated contralateral and 25.4 ± 3.7 ms when ipsilateral; t158 = 1.41, P = 5 × 10− 10; Fig. 6a; see Table 3 for a pulse-by-pulse comparison). The B-DARN1 activity is demonstrated in the collated raster plots and histograms for sound coming from either side (Fig. 6b, c). They reveal a 29% difference in the mean firing rate depending on the sound direction (Fig. 6f; 254.3 ± 31.3 AP s− 1 when contralateral; 179.6 ± 20.7 AP s− 1 when ipsilateral; ANOVA, effect of stimulus side, F1, 152 = 379.51, P = 5.2 × 10− 44; effect of pulse number, F3, 152 = 10.83, P = 2.0 × 10− 6). A post-hoc comparison suggested that the first pulse in a chirp evoked a significantly higher AP frequency than subsequent pulses.
Table 3
Differences in the number of evoked AP and burst duration when sound was presented from either the ipsilateral or contralateral direction. Data are means ± SD.
Pulse | Number of AP | Mann-Whitney | Burst Duration (ms) | Mann-Whitney |
| Ipsi | Contra | Z, P | Ipsi | Contra | Z, P |
1 | 4.95 ± 0.4 | 6.0 ± 1.1 | 3.77, 1.7 × 10− 4 | 26.1 ± 2.4 | 21.7 ± 4.2 | 3.48, 5.1 × 10− 4 |
2 | 4.9 ± 0.6 | 5.4 ± 0.5 | 2.75, 6.0 × 10− 3 | 28.1 ± 4.1 | 21.8 ± 2.3 | 4.34, 1.4 × 10− 5 |
3 | 3.95 ± 0.4 | 5.4 ± 0.6 | 5.57, < 1 ×10− 7 | 23.6 ± 3.4 | 22.2 ± 2.4 | 1.64, 0.102 |
4 | 4.3 ± 0.5 | 5.2 ± 0.5 | 4.33, 1.6 × 10− 5 | 23.9 ± 2.9 | 21.2 ± 2.6 | 2.94, 0.003 |
The latency to the first PSP in the response also differed significantly depending on the sound direction (Fig. 6d, e). The latency to ipsilateral presented pulses (17.0 ± 0.3 ms) was 0.8 ms shorter than the response to contralateral stimuli (17.8 ± 0.3 ms; t38 = 7.0, P = 4.1 × 10− 13). The latency to the first AP when stimulated on the contralateral side occurred slightly earlier (21.6 ± 0.7 ms), than the latency to the first PSP, when stimulated on the ipsilateral side (22.0 ± 0.5 ms; t38 = 2.15, P = 0.038; Fig. 6d, e,). The PSP latencies to unidirectional sound were both considerably shorter than when both loudspeakers were simultaneously active (20.1 ± 0.6 ms). As with the latency to PSP the latency to spiking was longer when both loudspeakers were active (22.8 ± 0.9 ms).
Similar response properties to monaural stimulation were also seen in another recording from a different cricket (Fig. 6g, h). Stimulation from either side evoked a strong spiking response, with similar numbers of AP per pulse (median number of AP per pulse shown in Fig. 6h; ANOVA, F1, 156 = 3.02, P = 0.084). The latencies to response did not differ between sides (ipsilateral, 17.1 ± 1.8 ms; contralateral 17.0 ± 0.8 ms; t27.19 = 0.106, P = 0.912). There was, however, an 8.5% difference in mean AP firing rate between sides, with the higher frequencies evoked when stimulation came from the contralateral side (239.1 ± 27.0 AP s− 1, compared to 218.8 ± 37.2 AP s− 1 on the ipsilateral side; Fig. 6h; ANOVA, F1, 156 = 19.77, P = 1.6 × 10− 5; effect of pulse number F3, 156 = 13.05, P = 1.2 × 10− 7).
In two other recordings, a different response to monaural stimulation was seen (Fig. 7), even though the neuron was anatomically similar in each case. In these recordings, the response to ipsilateral auditory stimulation was clearly stronger than when stimulated on the contralateral side. In the recording shown in Fig. 7, there were 77.8% more AP per pulse when stimulated ipsilaterally, compared to contralaterally (ANOVA, F1, 156 = 555.97, P = 2.7 × 10− 53; effect of pulse number F3, 156 = 25.29, P = 2.2 × 10− 13; Median number of AP per pulse, 5, 5, 5, 4 for the ipsilateral side and 3, 3, 2, 2 for contralateral side). The average firing rate of the neuron was 54.6% higher when stimulated on the ipsilateral side (excluding the 10% of contralateral responses where only a single AP was evoked; (ANOVA, F1, 144 = 379.75, P = 3.3 × 10− 42; effect of pulse number F3, 144 = 12.15, P = 3.9 × 10− 7). There was also a briefer latency of response when stimulated on the ipsilateral side (21.7 ± 0.9 ms compared to 25.6 ± 2.3 ms on the contralateral side; t23.32 =6.525, P = 1.1 × 10− 6).
Effect of high frequency sound stimulation
When the cricket was stimulated at 500 ms intervals by brief 20 ms pulses of 13.8 kHz and 75 dB SPL, long-lasting (109.0 ± 14.6 ms) compound PSPs were evoked after a latency of 14.6 ± 0.6 ms (Fig. 8a, b). The PSPs were surmounted by a brief burst of 3.8 ± 0.6 AP (Fig. 8c), with an average frequency of 175.1 ± 21.2 AP s− 1. Intercalating such high frequencies pulses within a train of ‘normal’ chirps at 4.8 kHz at an incidence of one high frequency pulse per ten calling-song chirps, abolished the spiking response to the high frequency sounds, whilst leaving the response to the chirps unaltered (Fig. 8d). Abolishing the spiking response revealed the barrage of individual PSPs producing the compound response (indicated by arrows in Fig. 8d). The high frequency of inputs and short latency of the graded response suggests that they come from AN2, an ascending interneuron that is more strongly tuned to higher frequencies (Wohlers and Huber, 1982). Playing chirp-like patterns (chirps with four pulses of 20 ms with 20 ms pulse intervals and 140 ms chirp intervals) at 13.8 kHz also did not evoke AP in B-DARN1, but instead compound PSPs which followed the pulse pattern of the chirps (Fig. 8e). Again, the structure of individual PSPs forming this compound response could be discerned. In another recording of B-DARN1, the neuron was presented with a sequence of three 50 ms pulses with a sound intensity of 75 dB SPL and escalating frequency, 4.8, 10 and 20 kHz at 110 ms intervals (Fig. 8f). The number of evoked APs decreased with increasing carrier frequency, from a peak instantaneous frequency of 448 AP s− 1 at 4.8 kHz, to 250 AP s− 1 at 10 kHz and only 74 AP s− 1 at 20 kHz, with latencies of 20 ms in this recording.
Courtship song in G. bimaculatus consists of 12–15 ms long sound pulses repeated at ~ 3.75 Hz and is dominated by frequencies of 12–16 kHz (Libersat et al., 1994). B-DARN1 responded to courtship song with an elevated rate of ‘background’ AP firing (approximately 10–20 AP s− 1) not obviously tied to the pulses of the song (Fig. 9a). Each sound pulse evoked a high amplitude long-lasting compound PSP (39.2 ± 9.1 ms; Fig. 9b) surmounted by 2.1 ± 0.4 AP reaching frequencies of 143 ± 46 AP s− 1, which did not show strong alignment with the onset of the courtship song pulse (Fig. 9c). The rising phase of the compound PSP was composed of at least two individual PSPs before the spiking threshold was reached (Fig. 9).