FIGURE SUMMARY
Title

Presynaptic Inhibition Selectively Gates Auditory Transmission to the Brainstem Startle Circuit

Authors
Tabor, K.M., Smith, T.S., Brown, M., Bergeron, S.A., Briggman, K.L., Burgess, H.A.
Source
Full text @ Curr. Biol.

Large-Scale Calcium Imaging of Neuronal Activity during Prepulse Inhibition

(A) Top: horizontal projection of gsx1-Gal4 expression (orange) with Mauthner cells (y264-Gal4, green, arrows) that drive fast startle and a pan-neuronal counter-label (HuC, blue), registered to the Zebrafish Brain Browser (ZBB). Yellow box: area imaged in (D). Bottom: coronal cross-section in hindbrain (dotted line in top) shows Gsx1 cell distribution in relation to Mauthner cells. Rostral (R), dorsal (D).

(B) Schematic of large-scale calcium imaging during behavioral prepulse inhibition.

(C) Startle responsiveness (short-latency tail-flip responses, SLCs) during acoustic tests. NS, no stimulus; Pre, prepulse-alone trial. n = 18 larvae. Error bars are SEM. WSR test, p = 0.0002 for suppression of responses on PPI trials compared to pulse-alone trials, confirming that behavioral prepulse inhibition was robustly elicited during imaging.

(D) 2-photon image of nls-GCaMP6s expression in Gsx1 cells (gray, left inset) and mean fluorescence change for 12 prepulse inhibition trials (ΔF, color scale, right inset) for two representative neurons. Activity traces are the mean calcium response to pulse-alone (black, n = 9), prepulse-alone (blue, n = 13), prepulse inhibition (red, n = 12), or no stimulus (orange, n = 11) trials. Shading is SEM. Arrows indicate times of prepulse (small) or pulse (large) stimuli.

(E) Raster plot of normalized GCaMP6s fluorescence change (ΔF/F, color scale), for 655 neurons in a representative experiment from a single larva.

(F) Mean normalized GCaMP6s fluorescence change (ΔF/F, color scale) for all segmented neurons after co-registration, for 4 indicated trial types.

See also Figure S1.

Prepulse Inhibition-Active Neurons Cluster near the Mauthner Cell

(A and C) Horizontal view of neurons that respond in the 500-ms windows after a prepulse on prepulse-alone (A) or during prepulse inhibition trials (prior to presentation of the pulse-stimulus) (C). Gray: all Gsx1 neurons. Cyan: Mauthner cells. Scale bar, 50 μm. Histograms represent the relative frequency of prepulse-responsive and PPI-responsive neurons, along the rostro-caudal axis. Neurons are color coded by responsiveness (Resp.), a measure of the response probability of the neuron, normalized by total activity in all neurons on each trial type.

(B and D) Coronal views of neurons that respond to a prepulse (B) or during prepulse inhibition trials (D) in rhombomere 1 (red), rhombomere 3 (green), and rhombomere 4 (blue) and the corresponding distribution of prepulse-responsive neurons along the dorsal-ventral axis. Position of each substack is indicated in corresponding horizontal views.

Optogenetic Activation of Gsx1 Neurons in the PPI-Active Zone Inhibits Startle Responses

(A) Schematic of experimental setup for selective illumination during behavioral prepulse inhibition. Inset: Kaede photoconversion in Gsx1 neurons (gsx1-Gal4, UAS:Kaede) by targeted illumination (cyan diamond). Scale bar, 50 μm.

(B) Prepulse inhibition after LED illumination of the hindbrain (large square, cyan), or R4 alone (small squares, cyan) in larvae with Gsx1 cells expressing chEF (black), and clutchmates lacking chEF (white). Left panel: LED illumination terminating 400 ms before the acoustic startle stimulus (PPI-400, n = 7 chEF+, n = 3 chEF). Main effect of chEF expression using repeated-measures ANOVA on rank-transformed %PPI F1,7 = 10.8, p = 0.013. Error bars are SEM. Right panel: LED illumination immediately before and during an acoustic pulse (PPI-0, n = 5 chEF+, n = 3 chEF). Main effect of chEF expression F1,6 = 3.3, p = 0.12. p < 0.05 Mann-Whitney test. Error bars are SEM.

See also Figure S2 and Table S1.

Intersectional Ablation of Candidate PPI Neurons

(A) Schematic of KillSwitch method: Nitroreductase (epNTR, orange) and RFP expressed in cells with Gal4 and Cre activity (red), while cells with only Gal4 activity express GFP (green). After metronidazole treatment, epNTR-dependent ablation eliminates Gal4+, Cre+ cells.

(B) Projection of Cre lines with differential expression in hindbrain rhombomeres.

(C) Prepulse inhibition in larvae with rhombomere-specific ablation of Gsx1 neurons (red), or metronidazole-treated clutchmates without nitroreductase expression (white). Projections show the full pattern of Cre expression. Schematic indicates rhombomere coverage by each line. Wilcoxon rank-sum (WRS) test, number of fish (control, ablated): hoxa2-Cre p = 0.99 (53, 70), eltC-Cre p = 0.003 (33, 10), y465-Cre p = 0.3 (30, 28), hoxb3a-dr16-Cre p = 0.1 (33, 17), y485-Cre p = 0.5 (21, 11). Error bars are SEM. p < 0.05.

See also Figure S3 and Table S2.

Glutamatergic Gsx1 Neurons in the PPI-Active Zone Project to the Mauthner Lateral Dendrite and Are Required for Prepulse Inhibition

(A–C) Coronal view of glycinergic Gsx1 neurons (A, gsx1:Cre, glyt2:Switch-Gal4, UAS:GFP, red), GABAergic Gsx1 neurons (B, gsx1-Gal4, UAS:Cre-ERT2, gad1b:Switch-GFP, red), and glutamatergic Gsx1 neurons (C, gsx1:Cre, vglut2a:Switch-Gal4, UAS:GFP, red) in R4. Cyan border outlines the PPI-active zone. Green: Mauthner cells. Dorsal (D); all scale bars, 50 μm.

(D) Schematic of the DoubleSwitch method where heat-shock-controlled expression of B3 is used to stochastically remove the stop cassette in a UAS:DoubleSwitch reporter, leading to RFP expression in neurons that also express Cre and Gal4. Inset: reconstruction of an isolated Gsx1 neuron (red) and retrograde labeling of ipsilateral Mauthner cell (black, blue arrowhead). Rostral (R).

(E–G) Reconstructions of morphological subtypes of Gsx1 neuron in the PPI-active region (E, type A, n = 8 ; F, type B, n = 5 ; G, type C, n = 5) with a standard model of Mauthner cells in coronal (top) and horizontal (bottom) views.

(H) Relative positions of type A (n = 6 neurons), B (n = 4 neurons), and C (n = 6 neurons) cell bodies along the medial-lateral axis of the ipsilateral Mauthner cell. Box plots are median and quartiles; whiskers are 10%–90%. WRS test, p < 0.05.

(I–K) Prepulse inhibition (% PPI) and startle responsiveness (% SLC) after ablation of glycinergic (I), GABAergic (J), or glutamatergic (K) neurons in the PPI-active region. White bars, sham ablated. Red bars, ablated. Startle trials used two stimulus intensities corresponding to the prepulse (Pre) and pulse (Pulse) in PPI trials. Number of larvae used was 9/6 (sham/ablated) in (I), 4/4 in (J), and 8/5 in (K). Error bars are SEM. WRS test, p < 0.05.

See also Figures S4 and S5.

Presynaptic Inhibition Mediates Long-Interstimulus Interval Prepulse Inhibition

(A) Startle responsiveness (% SLC) to an acoustic pulse (black) or an electric field pulse (red) at intervals of 100 to 1,000 ms after an acoustic prepulse. n = 9 groups of 20 larvae. Error bars are SEM. Paired t test, p < 0.05.

(B) Projection showing the auditory ganglia (y256-Gal4, red) and Mauthner cells (y264-Gal4, black). Right: auditory nerve expressing iGluSnFR (y256-Gal4, UAS:iGluSnFR, red) and the Mauthner lateral dendrite (black).

(C) Mean iGLuSnFR signals of VIIIth nerve clusters (color-coded) in response to pulse alone (left traces) and PPI trials (right traces). Arrowheads: time of a prepulse (small) and pulse (large). Black outline: Mauthner lateral dendrite. Scale bar, 50 μm.

(D) PPI of iGluSnFR signal in VIIIth nerve segments adjacent to the Mauthner dendrite (Apposed), or not (Distant). n = 10 larvae. WSR test, p = 0.002. Red in (D)–(F) is mean ± SEM.

(E) iGluSnFR PPI for apposed segments at 100 and 500 ms interstimulus interval (ISI) PPI tests. n = 4 larvae. WSR test, p = 0.02.

(F) iGluSnFR PPI for apposed segments when Mauthner action potential was suppressed (No spike) or present (Mc spike). n = 9 larvae. WSR test, p = 0.04.

See also Figure S6.

Large-scale calcium imaging of neuronal activity during prepulse inhibition. Related to Figure 1

A. Single horizontal confocal section showing Gsx1 cells expressing both nls-GCaMP6s (pink) and nls-DsRed2 (blue) in the hindbrain of a gsx1-Gal4, UAS:nls-GCaMP6s-2a-nls-DsRed2 larva. Insets are separate color channels. Orange arrows indicate the same cell. R, Rostral. All scale bars 50 μm.

B. Automated segmentation of 2-photon images (binarized nls-DsRed2, gray) by seeding at local maximum and temporal correlation showing hits (red), false positives (green) and false negatives (blue) compared with manual segmentation. Inset shows nls-DsRed signal.

C. F1 scores, precision and recall measures of automated compared with manual segmentation. n=6 experiments, means (±s.e.m) are red.

D. Percent of Gsx1 neurons active during acoustic tests. n=34050 total Gsx1 neurons. No stimulus trial (NS), prepulse alone trial (Pre), pulse alone trial (Pulse).

Optogenetic excitation of Gsx1 neurons and startle command Mauthner cells. Related to Figure 3

A. 2-photon image of chEF-expressing Gsx1 cells (red) and pan-neuronal GCaMP6s signal (black) in the hindbrain of a gsx1-Gal4, UAS:chEF-RFP, elavl3:H2B-GCaMP6s larva. Manually identified neurons (cyan markers) used to analyze optogenetically-evoked calcium signals. Rostral (R). Scale bar 50 μm.

B. Normalized GCamP6s fluorescence change (ΔF/F, color scale), for control neurons (left, 431 neurons, 2 elavl3:H2B-GCaMP6s larvae) and chEF-expressing Gsx1 neurons (right, 634 neurons, 3 gsx1-Gal4, UAS:chEF-RFP, elavl3:H2B-GCaMP6s larvae). Neurons are sorted by ΔF/F immediately after LED illumination of the hindbrain. Arrows indicate time of LED stimuli. Scale bar 1 s.

C. Mean (±s.e.m) calcium signal immediately after LED illumination for chEF-expressing Gsx1 neurons (chEF) and neurons without chEF (Control). n=3 chEF-expressing larvae and n=2 control larvae. WRS test, * P=0.02. Note, that when analyzed using the same criteria as for Figure 2 to identify active neurons, 8±4% of gsx1 neurons without chEF responded on at least one trial, and their mean response frequency was 33 ± 4% (i.e. most responded on only a single trial), whereas 36 ± 2% neurons expressing chEF responded at least once, with a response frequency of 49 ± 2%. So a small number of Gsx1 neurons are activated by the light flash, but the fraction and response rate is greatly increased after expression of chEF.

D. Horizontal projection of gsx1-Gal4 labeled Gsx1 cells (pink). Tail-flips (% SLC) during LED illumination of the hindbrain (left), and to an acoustic pulse (right) in larvae with Gsx1 cells with chEF (gsx1-Gal4, UAS:chEF-RFP, black), and clutchmates lacking chEF (white). WRS test; LED illumination: P=0.5, n=7 chEF-expressing larvae, n=3 controls; acoustic pulse: P=0.99, n=7 chEF-expressing larvae, n=3 controls.

E. Horizontal projection of y264-Gal4 (pink).

F. Short-latency tail-flips (% SLC) during LED illumination of the hindbrain, in larvae with Mauthner cells with chEF (y264-Gal4, UAS:chEF-RFP, black), and clutchmates lacking chEF (white). n=4 chEF-expressing larvae, n=3 controls. WSR test against 0, *P<0.05.

G. SLC to an acoustic pulse, and during bilateral (two cyan squares), or unilateral (one cyan square) LED illumination immediately before and during an acoustic pulse, in larvae with Mauthner cells with chEF (black), and clutchmates lacking chEF (white). n=4 chEF-expressing larvae, n=3 controls. Paired t-test, *P<0.05. All error bars are s.e.m.

H. Rightward tail-flips to an acoustic pulse, and during LED illumination of the left Mauthner cell, or both Mauthner cells before and during an acoustic pulse, in larvae with Mauthner cells with chEF (black) and clutchmates lacking chEF (white). n=4 chEF-expressing larvae, n=3 controls. Paired t-test, *P<0.05.

Intersectional genetic ablations and developmentally restricted ablations parse the Gsx1 population to reveal subsets that regulate acoustic startle. Related to Figure 4.

A. Horizontal projection of the genetic intersect pattern (red) of Gsx1 (gsx1-Gal4) and a Cre line registered to ZBB (HuC, blue). Border indicates intersection ablation that decreased prepulse inhibition of startle.

B. Prepulse inhibition (% PPI) of acoustic startle in larvae with rhombomere-specific ablation of Gsx1 neurons (black), or metronidazole-treated clutchmates without nitroreductase expression (white). WRS test, number of fish (control, ablated): y380-Cre P=0.03 (17, 14), hoxb1a-Cre P=0.0002 (23,25), y445-Cre P=0.002 (40,35), y523-Cre P=0.5 (34,19), y371-Cre P= 0.96 (23,18).

C. Startle responsiveness (% SLC) to a prepulse alone in larvae with rhombomere-specific ablation of Gsx1 neurons (black), or metronidazole-treated clutchmates without nitroreductase expression (white). Prepulse inhibition (% PPI) of startle in the same transgenic lines in Figure 4C. Unpaired t-test, number of larvae (control, ablated): hoxa2-Cre P=0.06 (133, 149), eltC-Cre P=3x10-14 (77, 34), y465-Cre P=0.3 (69, 56), hoxb3a-dr16-Cre P=0.004 (102, 102), y485-Cre P=0.001 (35, 43). C,D: Projections show the full pattern of Cre expression with a schematic indicating rhombomere expression in each line.

D. SLC to a prepulse alone for larvae with rhombomere-specific ablation of Gsx1 neurons (black), or metronidazole-treated clutchmates without nitroreductase expression (white). Unpaired t-test, number of fish (control, ablated): y380-Cre P=0.01 (33,28), hoxb1a-Cre P=0.00002 (51,51), y445-Cre P=0.04 (64,71), y523-Cre P=0.6 (54,35), y371-Cre P=0.4 (34,34). *P<0.05. Increased responsiveness was only seen when R4 Gsx1 neurons were ablated, indicating that these neurons negatively regulate startle responsiveness. Conversely, ablation using 3 different Cre lines (hoxb3a-dr16-Cre, y485-Cre and y445-Cre) that comprise R6 Gsx1 neurons reduced startle responsiveness, uncovering an unexpected role for these neurons in maintaining normal acoustic startle sensitivity that is masked by ablation of all Gsx1 neurons.

E-F. Earlier-generated Gsx1 neurons are required for prepulse inhibition but not startle threshold regulation. Prepulse inhibition (% PPI, E) and startle responsiveness (% SLC, F) after a 1-day metronidazole treatment starting on the indicated day to ablate Gsx1 cells in nitroreductaseexpressing y252-Gal4, UAS:epNTR larvae (black) and non-transgenic clutchmates (white). Twosided unpaired t-test, number of fish (control, ablated): day 2 treatment, PPI: P=0.03 (28, 33), SLC: P=0.2 (42, 37), day 4 treatment, PPI: P=3x10-12 (51, 31), SLC: P=7x10-26 (72, 72). *P<0.05. All error bars are s.e.m. Thus, ablation of all Gsx1 neurons generated before 5 dpf both disrupted prepulse inhibition and reduced acoustic startle thresholds. In contrast, prepulse inhibition was selectively impaired by ablation of Gsx1 neurons generated before, but not after, 3 dpf. Distinct Gsx1 neurons thus regulate prepulse inhibition and startle thresholds, despite both of these groups being present in R4.

DoubleSwitch method to stochastically label and reconstruct Gsx1 R4 neurons.Related to Figure 5.

A. Top: tails of larvae (6 dpf) with either a Cre reporter (left, βactin:Switch) or B3 reporter (right, HuC:Gal4, UAS:bloSwitch) after injection of 1 pg of recombinase RNA into the singlecell embryo. In both recombinase systems, the intact switch reporter expressed GFP (gray), and after recombinase-dependent excision, expressed RFP (red). Bottom: Western blot analyses of recombinase-dependent RFP fluorescence in larvae (6 dpf) with either a Cre reporter (Switch) or B3 reporter (bloSwitch) after injection of recombinase RNA into the single-cell embryo.

B. Lethality (% Death) of recombinase RNA injection into 1-cell stage embryos (Cre red, B3 blue). n=2 clutches for each recombinase. Error bars are s.e.m.

C. Reconstructions of Gsx1 neurons in the PPI-active region with a standard model of Mauthner cells in the coronal (left) and horizontal (right) views. Type A neurons projected bilaterally beyond R4 to rostral and caudal brain regions. Type B neurons, located in a lateral zone, confined their bilateral processes to rhombomeres 4-5, where they projected ventrally and ramified below the Mauthner lateral dendrite. Type C neurons projected ipsilaterally. D, dorsal. R, rostral. Scale bar 50 μm.

D. Projections from type A Gsx1 neurons bilaterally appose the Mauthner cells at positions on the soma, with a strong correlation between the medio-lateral location of the apposition on the ipsi and contra-lateral Mauthner somas. n=5 neurons. Pearson correlation coefficient R2 = 0.9, P=0.04.

Ablation of neurotransmitter-specified Gsx1 neurons dorsal to Mauthner cell. Related to Figure 5

A-C. Glutamatergic (A: gsx1-Gal4, UAS:Cre-ERT2, vglut2a:Switch-GFP, red), GABAergic (B: gsx1-Gal4, UAS:Cre-ERT2, gad1b:Switch-GFP, red), and glycinergic (C: gsx1:Cre, glyt2:Switch-Gal4,UAS:GFP, red) Gsx1 neurons in the dorsal vicinity of the Mauthner cell (retrogradely labeled, red) before (top) and after (bottom) laser ablation. Scale bar 50 μm applies to all panels.

Acknowledgments
This image is the copyrighted work of the attributed author or publisher, and ZFIN has permission only to display this image to its users. Additional permissions should be obtained from the applicable author or publisher of the image. Full text @ Curr. Biol.