Implementation of RVdG[EnvA]-based neural circuit tracing in larval zebrafish

(A) Schematic of the method used to implement RVdG[EnvA]-based circuit tracing from genetically defined cell types in larval zebrafish. (B) Schematic of the three UAS-driven plasmids used to express the helper proteins TVA and G. (C and D) Time-lapse confocal images of the nacre larval hindbrain expressing EGFP and TVA using UGT (C), or EGFP, TVA, and B19G using UGTB (D) in randomly sparse neurons (with expression of UAS constructs driven by the elavl3:GAL4-VP16 plasmid) and infected by SADdG-mCherry[EnvA] (magenta) injected into the hindbrain. Dashed yellow circles, starter neurons (EGFP+/mCherry+); arrows, transneuronally labeled neurons (red) and radial glia (gray) (mCherry+/EGFP−); dashed white lines, hindbrain boundaries. C, caudal; R, rostral. Scale bars, 50 μm. (E) Plots of the numbers of starter neurons, transneuronally traced neurons and glia in each larva expressing one of the three helper plasmids shown in (B) and injected with SADdG-mCherry[EnvA]. Virus injections were carried out at 3–5 dpf; exact ages for individual animals are listed in Figure 1—table supplement 2. Data were collected between 6 and 10 dpi. Numbers in brackets indicate the number of larvae examined. (F and G) Boxplots of the convergence index for the connection of transneuronally traced neurons (F) and glia (G) to starter neurons shown in (E). Center, median; bounds of box, first and third quartiles; whiskers, minimum and maximum values. **P < 0.01, ***P < 0.001 (nonparametric two-tailed Mann-Whitney test).

Optimization and verification of RVdG[EnvA]-based retrograde monosynaptic tracing

(A) Schematic of the method used to trace inputs to PCs with RVdG[EnvA] in larval zebrafish. The cpce:GAL4FF plasmid and one of the three UAS helper plasmids were injected into nacre fish and combined with subsequent injection of SADdG-mCherry[EnvA] or CVSdG-tdTomato[EnvA]. The virus-injected larvae were raised at either 28 or 36 °C, and the resulting labeling was monitored from 2 to 16 dpi. (B) Schematic of the zebrafish cerebellar circuit related to inputs to PCs. (C) Top, examples of in situ complementation in PCs under six tracing conditions. Confocal images and corresponding schematic illustrations of one cerebellar hemisphere are shown. Bottom,plots of the numbers of starter neurons, transneuronally traced neurons and glia in each larva under one of the six tracing conditions at 6–10 dpi. Virus injections were carried out at 4–6 dpf; exact ages for individual animals are listed in Figure 1,2—table supplement 2. Dashed lines, the cerebellar boundaries. CI, convergence index of traced neurons. Numbers in brackets indicate the number of larvae examined. (D–F) CI for the connection of transneuronally traced neurons to starter PCs (D), the ratio of the number of traced neurons to that of glia (E), and the proportion of infected fish (F, 3 to 5 independent experiments) under the six tracing conditions shown in (C, bottom). (G) Confocal image of a larva infected with CVSdG-tdTomato[EnvA] that was trans-complemented with CVS N2cG at 36 °C. Dashed yellow circle, starter PC; dashed white line, the cerebellar boundary. (H and I) Enlarged view of the boxed regions in (G), showing transneuronally traced parallel fibers of GCs (H) and two IOCs (I). Blue arrows, somata of IOCs. (J) Schematic of the experiment for verifying the functional connection between traced GCs and the starter PC in nacre larvae injected with cpce:GAL4FF and advanced UG6sNT-A plasmids and infected with CVSdG-tdTomato[EnvA] at 36 °C. Ca2+ imaging was conducted on the starter PC in response to electrical stimulation (ES) at the GC position before and after two-photon laser ablation of the GC .(K) Representative image (top) showing both the stimulated GC and the simultaneously recorded PC. Bottom panels display the stimulated GC before (left) and after (right) two-photon laser ablation. Dashed yellow circle, recorded PC (GCaMP6s+/tdTomato+); dashed white circles, stimulated GC (GCaMP6s−/tdTomato+). (L) Average Ca2+ response (5 trials) of the same starter PC before (red) and after (black) GC ablation. The shadowed area represents SEM. (M) Quantification of the amplitude of PC Ca2+ responses before and after GC ablation. The number in bracket indicates the number of GC-PC pairs examined. Scale bars, 50 μm (C, G–I), 10 μm (K, top), 5 μm (K, bottom). SAD/CVS|B19G/N2cG/A-N2cG|28/36 °C, the tracing conditions written as the G-deleted RV strain that was injected | the helper glycoprotein used for trans-complementation | fish rearing temperature after virus injection. C, caudal; L, lateral; R, rostral. Boxplots in (D and M): center, median; bounds of box, first and third quartiles; whiskers, minimum and maximum values. n.s., not-significant; *P < 0.05, **P < 0.01, ***P < 0.001 (nonparametric two-tailed Mann-Whitney test in (D–F); two-tailed unpaired Student’s t test in (L)). Error bars denote SEM.

Neuronal health and physiology appeared normal for at least 10 days after RV infection

(A) Survival rate of larvae injected with SADdG-mCherry[EnvA] or CVSdG-tdTomato[EnvA]. Five independent experiments were repeated. (B) Time-lapse confocal images showing that a starter cell (unfilled white arrowhead) present at 2 dpi disappears by 8 dpi, and two starter cells (white arrowheads) remain present from 2 to 8 dpi. Boxed regions are enlarged on the right. Dashed lines indicate the cerebellar boundaries. (C) Cumulative proportion of starter cells that have appeared at 2, 4, 6, and 10 dpi to the total number of observed starter cells under the three tracing conditions, showing the continuous appearing of starter cells at 2–6 dpi. Numbers in brackets indicate the number of larvae examined. The shadowed area represents SEM. (D) Summary of the lifetime of starter cells appeared at 2 dpi under the three tracing conditions. The color scheme follows the same conventions shown in (C). Numbers in brackets indicate the number of starter cells examined. (E) Schematic illustration of in vivo whole-cell recording on a starter PC at 11 dpi. (F) Example traces of spontaneous EPSCs (left) and EPSPs (right) from starter PCs. (G) Example trace of spikes in response to current-step injection into a starter PC. (H) Confocal image (top) and Ca2+ responses (bottom) of a starter PC (dashed orange circle) and non-starter PC (dashed blue circle) to visual stimuli at 10 dpi. (I–L) Boxplots (I and J) and distributions (K and L) of Ca2+ activity event number (I and K) and area under curve (AUC) of Ca2+ activity events (J and L) of starter PCs (n = 88) and non-starter PCs (n = 127) in larvae (N = 8) at 6–10 dpi. All tracing experimental procedures were identical to those described in Figure 2. In panels H–L, the helper plasmid UG6sNT-A was used to express GCaMP6s for calcium imaging of PCs. Scale bars, 5 μm (H), 10 μm (B), 20 μm (E). SAD/CVS|B19G/N2cG/A-N2cG|28/36 °C, the tracing conditions written as the G-deleted RV strain that was injected | the helper glycoprotein used for trans-complementation | fish rearing temperature after virus injection. L, lateral; R, rostral. Boxplots in (D, I, and J): center, median; cross symbol, mean; bounds of box, first and third quartiles; whiskers, minimum and maximum values. n.s., not-significant; *P < 0.05; ***P < 0.001 (two-tailed unpaired Student’s t test). Error bars denote SEM.

Cell-type-specific input tracing and circuit reconstruction in the cerebellum

(A) A single Cre-Switch (Saunders et al., 2012) transgene (elavl3:Tetoff-DO_DIO-Hsa.H2B-mTagBFP2_tdTomato-CAAX; abbr., elavl3DoDioBR) was generated to achieve a reference transgene (elavl3:H2B-mTagBFP2) for image registration and the Cre-dependent reporter expression (tdTomato-CAAX) specifically in neurons (elavl3+). Two ORFs are flanked by two pairs of LOX sites in opposite orientations, with activation of the forward-oriented H2B-mTagBFP2 ORF under the elavl3 promoter in global neurons. After Cre-expressing virus infection, Cre recombines either pair of LOX sites (LOXP (gray triangles) or LOXP2272 (black triangles)), resulting in stable inversion of the two ORFs and activation of the tdTomato-CAAX ORF under the elavl3 promoter in initially infected starter neurons and transneuronally infected input neurons. (B) Schematic of building PC input atlas in larval zebrafish. The elavl3DoDioBR reporter fish (pan-neuronal H2B-mTagBFP2+) expressing helper proteins in PCs (sfGFP+/mTagBFP2+) were injected with Cre-expressing CVSdG[EnvA], which were trans-complemented in starter PCs and spread to direct inputs, resulting in Cre-dependent expression of the reporter protein (tdTomato-CAAX) in elavl3+ starter PCs (sfGFP+/tdTomato+/mTagBFP2−) and input GCs (sfGFP−/tdTomato+/mTagBFP2−). The pan-neuronally-expressing H2B-mTagBFP2 in elavl3DoDioBR serves as the reference and bridge for the registration of reconstructed circuits. BCs, Bergmann glia; GCs, granule cells; PCs, Purkinje cells. (C) Example of in situ complementation (11 dpi) in nacre larvae (left) and elavl3DoDioBR reporter larvae (right). Both were co-injected with cpce-GAL4FF and UGNT plasmids, and infected with CVSdG-tdTomato[EnvA] and CVSdG-Cre[EnvA], respectively. All larvae were maintained at 36 °C. In the reporter larvae, Cre-dependent tdTomato expression was confined to neurons, driven by neuron-specific elavl3 promoter. Confocal images (top) and corresponding schematic illustrations (bottom) are shown. Dashed yellow circles, starter PCs; dashed lines, the cerebellar boundaries. Black arrows indicate the reconstructed neurons in (D). (D) Dorsal 3D view of the reconstructed starter PC and its two input GCs. Dashed line indicates the cerebellar boundary. The color scheme follows the same conventions as in (F). (E) Left, single horizontal section view of Tg(2×en.cpce-E1B:tdTomato-CAAX) (red), Tg(cbln12:GAL4FF);Tg(5×UAS:EGFP) (green), and elavl3DoDioBR (blue) templates aligned in a common coordinate space. Right, volume rending of the cerebellar area (dashed white rectangle on the left) of the templates labeling PCs (red) and GCs (green) in cerebellum region outline (gray). (F) Dorsal (left) and lateral (right) 3D view of reconstructed starter PCs (n = 24) and input GCs (n = 58) in the cerebellum (gray). Different subcellular parts of PCs and GCs are color-coded. Data were collected from 13 larvae at 10–16 dpi (15–21 dpf). (G and H) Ipsilateral vs. contralateral proportions of GC inputs to PCs in the left and right cerebellum. Analysis was performed using the reconstructed data in (E) and the viral tracing data shown in Figure 2C,D for (G) and (H), respectively. Numbers in brackets indicate the number of larvae examined which only had starter PCs on one cerebellar hemisphere. (I) Subtypes of reconstructed PCs and GCs. Two typical cells are shown for each subtype. (J) Example showing three GCs traced from a single PC. (K) Summed input proportion of two subtypes of GCs to two subtypes of PCs. Cases with only one subtype of PC and clearly identifiable GCs were included. The cell numbers are as follows: PC1, 12; PC2, 6; GC1, 19; GC2, 18. (L) Schematic representation of subtype-specific connectivity patterns from GCs to PCs. High saturation of color indicates a higher input strength. CCe, corpus cerebelli; EG, eminentia granularis; LCa, lobus caudalis cerebelli; Va, valvula cerebelli. Scale bars, 50 μm. A, anterior; C, caudal; L, lateral; R, rostral; V, ventral. Dashed vertical lines indicate the midline of the cerebellum.

No infectiousness of RVdG[EnvA] in the absence of TVA

(A) Injection of SADdG-mCherry[EnvA] into the brain of nacre larvae. Left, schematic; Right, time-lapse confocal images of the optic tectum of a WT larva injected with the virus. (B) Time-lapse confocal images of the optic tectum of a Tg(apoeb:LY-EGFP) larva injected with SADdG-mCherry[EnvA], showing the location of mCherry-positive puncta in EGFP-positive microglia. Scale bars, 50 μm. Arrowheads, fluorescent-positive punctate structure with no discernible cellular morphology under high laser excitation; dashed lines, boundaries of the cell body layer of the optic tectum. C, caudal; R, rostral.

Location of transneuronally labeled glial cells

(A and B) Confocal images of the right tectum (A) and posterior hindbrain (B) from different WT larvae, showing randomly sparse vglut2a+ neurons expressing EGFP and helper proteins via UGNT, following infection with CVSdG-tdTomato[EnvA] (magenta) injected into the anterior hindbrain (see Figure 2 for the helper plasmid and virus vector). Scale bars, 20 μm. Dashed yellow circles, starter neurons (EGFP+/tdTomato+); gray arrows, transneuronally labeled radial glia (tdTomato+/EGFP−); dashed white lines, tectal or hindbrain boundaries. C, caudal; R, rostral.

Viral tracing initiated from glia

(A) Confocal images of the tectum of a larva showing randomly sparse gfap+ glial cells expressing EGFP and helper proteins via UGBT, following infection with SADdG-mCherry[EnvA] (magenta) injected into the anterior hindbrain at 28 °C. (B) Confocal images of the posterior hindbrain of a larva showing randomly sparse gfap+ glial cells expressing EGFP and helper proteins via UGNT, following infection with CVSdG-tdTomato[EnvA] (magenta) injected into the anterior hindbrain at 28 °C. Scale bars, 20 μm. Dashed yellow circles, starter glial cells (EGFP+/mCherry+ or EGFP+/tdTomato+); gray arrows, secondarily labeled glial cells (tdTomato+/EGFP−); dashed white lines, tectal or hindbrain boundaries. C, caudal; R, rostral.

Time-lapse images of trans-complemented tracing from Purkinje cells in the cerebellum under different tracing conditions

(A, B, and D–F) Time-lapse (2, 4, and 6 dpi) confocal images of in situ complementation in Purkinje cells (PCs) under tracing conditions of SAD|B19G|28 °C (A), CVS|B19G|28 °C (B), SAD|B19G|36 °C (D), CVS|N2cG|36 °C (E), and CVS|A-N2cG|36 °C (F). The schematic drawing of the advanced helper plasmid (UGNT-A) is shown in (F). Dashed yellow circles, starter PCs; dashed white lines, the cerebellar boundaries. L, lateral; R, rostral. Scale bars, 20 μm. (C and G) Summary of CI for the connection of traced neurons, glia, or all cells to starter PCs at 2, 4, 6, and10 dpi under two tracing conditions at 28 °C (C) and three tracing conditions at 36 °C (G). Significant intergroup differences exist in CI for traced neurons, glia, or all cells (see the P values). There are no statistical differences (n.s.) between 6 and 10 dpi in any of the groups. Two-way ANOVA and nonparametric two-tailed Mann-Whitney test were used for intergroup and intragroup statistics, respectively. Error bars denote SEM. Numbers in brackets indicate the number of larvae examined. (H and I) Boxplots of CI for the connection of traced glia (H) or all cells (I) to starter PCs under six tracing conditions at 6–10 dpi. See also Figures 2C–E. Center, median; bounds of box, first and third quartiles; whiskers, minimum and maximum values. n.s., not-significant; *P < 0.05; **P < 0.01; ***P < 0.001 (nonparametric two-tailed Mann-Whitney test).

Calculation of the labeling rate index across two time intervals

The labeling rate for neurons and glial cells under each experimental condition was calculated as the daily increase in the mean CI across defined time intervals, based on data from Figure 2—figure supplement 1C and G. The mean CI was quantified separately for neurons and glia at each time point (dpi) by averaging CI values from individual fish within each group. Changes in mean CI across two consecutive time intervals (2–6 and 6–10 dpi) were normalized by the interval duration (4 days) to obtain daily labeling rates for each cell type. The labeling rate index (Rglia − Rneuron) was defined as the difference between glial (Rglia) and neuronal (Rneuron) labeling rates. *P < 0.05 (nonparametric two-tailed Mann-Whitney test).

Fraction of glial cells retaining fluorescence

(A and B) Boxplots of the proportion of glial cells labeled at 2 dpi (A) or 4 dpi (B) that retained detectable fluorescence at each subsequent time point. Data are from the CVS|N2cG|36 °C group shown in Figure 2—figure supplement 1G. Box center, median; bounds of box, first and third quartiles; whiskers, minimum and maximum values. n.s., not-significant; *P < 0.05, **P < 0.01 (nonparametric two-tailed Mann-Whitney test).

The efficiency of viral transfer under different tracing conditions in larval zebrafish.

The convergence index (CI) was calculated as the number of traced cells divided by the number of starter cells.

Injection age and corresponding fish numbers for each tracing condition.

The proportion of fish at each injection age is shown in parentheses.

Properties of currently available viral circuit tracing systems for zebrafish.