Intersecting the expression of acj6 and unc-4 genes with the Split-GAL4 method faithfully marks hemilineage 23B.

(A-C) Projections of confocal stacks of the adult VNC. Magenta: CadN; green: tdTomato (A) acj6-GAL4 driven nls-tdTomato expression marks Acj6 expressing neurons. (B) unc-4-GAL4 driven nls-tdTomato expression marks Unc-4 expressing neurons. (C) The intersection of acj6 and unc-4 expression (acj6-GAL4AD, unc-4-GAL4DBD> UAS-nls-tdTomato) marks lineage 23B neurons in the SEZ and VNC. (D) A partial confocal projection showing the complete overlap between membranous GFP (green) and Acj6 (magenta) immunostainings in acj6-GAL4AD, unc-4-GAL4DBD-marked 23B neurons in the adult VNC (T1 and T2 segments shown). (E) scRNAseq t-SNE plot shows Acj6 and Unc-4 co-expression in a group of cell clusters.

acj6-GAL4AD, unc-4-GAL4DBD-driven myr-GFP marks 23B neurons throughout development.

(A,B) Acj6 (blue) and Unc-4 (magenta) co-expression shows robust overlap in GFP-marked embryonic progeny of NB7-4, 23B neurons, in a late stage embryo. (C–D) Acj6 (blue) expression marks 23B neurons in an early stage larval VNC (C) and an early stage pupal VNC (D). The only lineages that express Acj6 are 23B, 8B and 9B, and of these only the posterior-dorsal cells, corresponding to hemilineage 23B, co-stained for GFP and Acj6 in the larval and pupal VNC. (E) This driver combination marks a cluster of SEZ neurons (arrowhead) in the adult brain, presumably SEZ 23B neurons in addition to sensory neuron afferents (arrows). (F) Close up of SEZ to highlight the corresponding cell bodies (arrowhead).

Matching the scRNAseq clusters to hemilineages.

(A–C) Confocal stack of larval VNC displaying the overlapping expressions between transcription factors identified from scRNAsec data (Fkh, Kn, and Sp1; green in A, B, and C, respectively) and Hb9 (magenta) in three lineages: 4B, 10B, and 16B (dashed lines). Asterisk in A indicates the Fkh+Hb9 0A lineage neurons. (D) Sox21a-GAL4 driven UAS-GFP (green) marks lineage 2A neurons (E) HmxGFSTF reporter (green) marks lineage 17A neurons. (F, G) Wild-type MARCM clones (green) immunostained for Tj (magenta). The insets show the clone location in the VNC counterstained with CadN (blue) (F) Tj marks subpopulations of neurons in lineage 0A in the T2 segment. These neurons likely belong to cluster 88, the only Tj+ 0A cluster in scRNAsec data. (G) Tj marks nearly all neurons of lineage 21A in the T1 segment. Lineage identification of MARCM clones were performed based on neuronal projections detailed in Truman et al., 2004 (66). scRNAseq clusters with the corresponding lineages shown under each panel. Only one thoracic segment shown. Neuroglian specific antibody BP104 labels axon bundles of all lineages (magenta in D-E).

Overview of cluster annotation, lineage specific marker genes and tested split-GAL4 driver lines.

The VNC expression of select driver lines from the Split-GAL4 library targeting individual hemilineages.

Projections of confocal stacks showing the expression pattern of Split-GAL4 driven membranous GFP (green) in the larval (A-O) and adult VNC (A’-O’). Only thoracic segments are shown in the larval images (A, A’) Hemilineage 0A, marked by inv-GAL4-DBD, tj-vp16.AD. (B, B’) Hemilineage 1A marked by ets21c- GAL4-DBD, Dr-p65.AD. (C-C’) Hemilineage 2A marked by sox21a GAL4-DBD, VGlut-p65.AD. (D, D’) Hemilineage 4B marked by ap-p65.AD, fkh-GAL4-DBD. (E, E’) Hemilineage 5B marked by vg-p65.AD, toy-GAL4-DBD. (F, F’) Hemilineage 6B marked by sens2-p65.AD, vg-GAL-DBD. (G, G’) Hemilineage 7B marked by mab21-GAL4-DBD, unc-4-p65.AD. (H) Hemilineage 8A marked by ems-GAL4-DBD, ey-p65.AD. (I, I’) Hemilineage 8B marked by lim3-GAL4-DBD, C15-p65.AD. (J, J’) Hemilineage 9A marked by Dr-p65.AD, gad1-GAL4-DBD (K, K’) Hemilineage 9B marked by acj6-p65.AD, VGlut-GAL4-DBD. (L, L’) Hemilineage 10B marked by hb9-p65.AD, knot-GAL4-DBD. (M, M’) Hemilineage 12A marked by TfAP-2-GAL4-DBD, unc-4-p65.AD. (N, N’) Hemilineage 14A marked by Dr-p65.AD, toy-GAL4-DBD. (O, O’) Hemilineage 17A marked by unc-4-p.65AD, hmx-GAL4-DBD. The VNC was counterstained with CadN (magenta). The target lineage is indicated on the left bottom corner of each panel. Z-projections were made of selected regions of the VNC to highlight the cell-body clustering and axonal budling.

The rest of the driver lines from the Split-GAL4 library targeting individual hemilineages.

Projections of confocal stacks showing the expression pattern of Split-GAL4-driven membranous GFP (green) in the larval (A-O) and adult VNC (A’-O’). Only thoracic segments shown in the larval images. (A) Hemilineage 1B marked by HLH4c-GAL4-DBD, H15-p65.AD. (B) Hemilineages 3A, 7B, and 12A are marked by H15-p65.AD, ChAT-GAL4-DBD. (C) Hemilineages 3B and 12B marked by fer3-GAL4-DBD, cg4328-AD. (D) Hemilineage 6A marked by mab21-p65.AD, toy-GAL4-DBD. (E) Hemilineage 11A marked by unc-4-GAL4-DBD, teyVP16.AD. (F) Hemilineage 11B marked by eve-p65.AD, gad1-GAL4-DBD. (G) Hemilineage 12B marked by HGTX-GAL4-DBD, gad1-p65.AD. (H) Hemilineage 13A marked by dbx-GAL4-DBD, dmrt-p65.AD. (I) Hemilineage 13B marked by vg-GAL4-DBD, D-vp16.AD. (J) Hemilineage 15B marked by HGTX-GAL4-DBD, VGlut-p65.AD. (K) Hemilineage 16B marked by hb9-p.65AD, VGlut-GAL4-DBD. (L) Hemilineage 19A marked by dbx-GAL4-DBD, scro-p65.AD. (M) Hemilineage 20/22A marked by bi-GAL4-DBD, shaven-p65.AD. (N) Hemilineage 23B marked by unc-4-p65.AD, acj6-GAL4-DBD. (O) Hemilineage 24B marked by twit-p65.AD, ems-GAL4-DBD.

CRISPR mediated insertion of Trojan Exons.

(A) Construction of CRISPR donor plasmids. For each gene of interest (GOI) a fragment is synthesized into EcoRV restriction site of pU57_gw_OK2 as described before (64). Briefly, this fragment contains a small sequence of the tRNA spacer, the gRNA against the gene of interest (GOI) (turquoise) and the Left HA and Right HA (brown) separated by a spacer containing SacI and KpnI restriction sites (black). A hemidriver cassette (gray, also see B) flanked by SacI and KnpI restriction sites is directionally cloned in between the HAs. (B) Six plasmids containing hemidriver cassettes (gray box) flanked by SacI and KpnI were made in the pBS-KS plasmid backbone. Each plasmid contains either a split-GAL4DBD or p65.AD in phase 0, 1 and 2. Each hemidriver furthermore contains a 5’attP and FRT sequences, followed by a linker, splice acceptor (SA) and T2A proteolytic cleavage site. The linker length varies to keep the hemidriver in phase with the preceding exon (linker length: 24 nucleotides phase 0, 41 nucleotides phase 1 or 40 nucleotides phase2). A hsp70 termination sequence is introduced at the 3’end of the hemidriver followed by a splice donor (SD), FRT, and attP sequence Note that the DBD cassettes do not contain a splice donor to keep them consistent with previously published split-GAL4 Trojan exon donors (28). (C) The HAs promote HDR and the entire hemidriver cassette is inserted at the site of the CRISPR/CAS9 cut, targeted by recognition sequence the gRNA-GOI. The attP sites allow for future cassette exchange with RMCE and genetic crosses.

Direct tagging with CRISPR.

Schematic representation of the direct tagging method that establishes split-GAL4DBD lines without any cloning.

The gRNA against the gene of interest (GOI) cuts in the direct vicinity of the stop codon (+/- 20 nt). The left HA 3’ end reaches up to, but does not include the stop codon, and the right HA 5’ end starts at the first nucleotide of the 3’ UTR. This ensures that the T2A-DBD fragment will be inserted at the 3’ end of the gene and is translated in frame with the GOI. (A) Construction of the CRISPR donor for direct tagging. A fragment that contains a small portion of the tRNA spacer, the gRNA-GOI, and the LHA, T2A-DBD and RHA sequence is directly synthesized into the EcoRV site of pU57_gw_OK2. (B) Upon embryo injection, expression of gRNA1 linearizes the donor constructs and the LHA-T2A-DBD-RHA fragment is used for CRISPR/Cas9 guided HDR. As a result, the T2A-DBD is inserted in frame at the 3’ end of the gene, and endogenous 3’ UTR posttranslational regulation mechanisms remain intact.

Neurons of hemilineage 4B show profound morphological changes during development.

Projection of confocal stacks showing the morphology of 4B neurons (green) marked with the ap-GAL4AD and fkh-GAL4DBD driver combination across different developmental time points during metamorphosis: 0, 3, 12, 24 and 48 hours after puparium formation (APF). The VNC is counterstained with CadN (magenta). Cell bodies of 4B neurons are marked with asterisks. A-F show the complete projections in T2-T3 segments. Anterior (A) up; posterior (P) down. A’-F’ show transverse views of the entire T3 segments across the dorso-ventral (D–V) axis; Dorsal is up. Arrowheads in B’ mark growth cones. Arrowheads in C’ mark three new branches towards the medial (m), lateral (l) and dorsal (d) part of the leg neuropil.

Acj6-positive neurons in the VNC are glutamatergic or cholinergic.

(A–C) Split-GAL4 line reporting Acj6 expression intersected with a cognate split-GAL4 line reporting the expression of Gad1, ChAT or VGlut to visualize GABAergic, cholinergic, and glutamatergic populations of Acj6-positive neurons, respectively. The VNC is counterstained with CadN (magenta). (A) Split-GAL4 combination acj6-p65.AD, gad1-GAL4-DBD>UAS-GFP driven UAS-GFP shows that the optic lobes contain cholinergic Acj6-positive neurons in addition to a few clusters of neurons with prominent long projections. In the VNC, two cholinergic clusters per hemisegment corresponding to 8B (arrowheads) and 23B (arrows) hemilineages are labeled in addition to some sensory neurons (asterisks). (B) Split-GAL4 combination acj6-p65.AD, VGlut-GAL4-DBD> UAS-GFP marks a single glutamatergic lineage in the dorsal part of the brain and one 9A glutamatergic cluster in the VNC. (C) Split-GAL4 combination acj6-p65.AD, gad1-GAL4-DBD>UAS-GFP marks two GABAergic lineages in the brain and nothing in the VNC.

Overview of behavioral phenotypes upon optogenetic activation of specific hemilineages.

Behavioral analysis with targeted lineage manipulation.

(A–D) Optogenetic activation of hemilineage 8A in the VNC triggers jump behavior. lim3-GAL4DBD; c15-GAL4AD driven CsChrimson::mVenus (green) targets 8B neurons in the VNC but also shows an unwanted broad brain expression (A), which can be suppressed via an additional layer of intersection using teashirt (tsh)-lexA driven FLP strategy (B). (C, D) Overlay of video frames to capture the jump sequence induced by optogenetic activation of lineage 8B in the VNC. Intact flies (C) and decapitated flies (D) jump without raising their wings upon optogenetic activation, but decapitated flies were slower to initiate the jump. (E) Optogenetic activation of hemilineage 9A induces forward walking in decapitated flies. (F, G) Clonal stimulation of hemilineage 12A in the VNC in decapitated flies induces bilateral wing opening and single-step behavior. (F) Confocal stack displaying the lineage 12A clone that extends from T2 into T1 and T3. (G) Overlay of movie frames. The fly folds both wings outward and swings its right front leg forward upon optogenetic activation. (H, L) Optogenetic activation of hemilineage 21A in the VNC on a tethered, intact fly triggers flexion of the tibia-femur joint. (H) Without stimulus, all the legs move erratically in response to being tethered. (I) Upon optogenetic activation, all legs are pulled toward the body, the tibia-femur joints are flexed, and animals stay in this position until the end of stimulus. (J) Overlay of the movie shown in panel H and I, zoomed in on the left T1 leg. Note how the leg is pulled towards the body upon activation (520ms) compared to its more lateral position without activation (315 ms). (K, L) Elimination of 21A neurons makes hind leg femur-tibia joints protrude laterally (L) compared to control animals (K). For all overlays of movies, green display frames without optogenetic activation, magenta with optogenetic activation.

Giant Fiber (GF) Connectome.

Synaptic connectivity of the GF neuron extracted from the data generated by Marin et al., (1). (A–C) Analysis of GF input connections. (D–F) Analysis of GF output connections. (A) Count of neurons per hemilineage that form synapses with GF dendrites. A total of ten hemilineages form synapses with GF dendrites. Five neurons originate from hemilineage 8B, six from hemilineage 7B, five from lineage 5B and three from lineage 21A. (B) Combined connectivity per hemilineage, cumulative count of synapses between GF dendrites and hemilineage neurons. The connectivity between hemilineage 8B and the GF is significant, spanning 339 synapses. Hemilineage 7B, 5B and 21A forms 45, 205 and 108 connections, respectively. (C) Weighted connectivity per hemilineage, calculated as the cumulative count of synapses between GF dendrites and hemilineage neurons, divided by the total number of GF output connections observed at a threshold of five synapses per neuron. Hemilineage 8B contributes heavily, making up 25% of GF input, followed by 15% from lineage 5B. Lineage 7B contributes 3.3% and lineage 21A 8%. (D) Count of neurons per hemilineage that form synapses with GF axons. A total of 13 hemilineages are downstream synaptic partners of the GF. Of those, the synapses formed with lineage 8B are most divergent and span 12 neurons. (E) Combined connectivity per hemilineage, cumulative count of synapses between GF axons and hemilineage neurons. Hemilineage 8B makes 208 synaptic contacts. Hemilineage 18B and 6B also form strong connections, 206 and 121 connections, albeit with fewer neurons (5 and 6, respectively). (F) Weighted connectivity per hemilineage, calculated as the cumulative count of synapses between GF axons and hemilineage neurons, divided by the total number of GF output connections observed at a threshold of five synapses per neuron. 12.5% of output GF synaptic contacts are made with hemilineage 8B, followed by 12.4 % with lineage 18B and 7.3% with lineage 6B.

Key Resources Table

Detailed description of the expression patterns of the driver lines used in Figure 3 and Figure 3 Supplemental Figure 1.

synaptic inputs of the Giant Fiber neuron related to Figure 6.

synaptic outputs of the Giant Fiber neuron related to Figure 6.

Additional information on CRISPR genomic edits.

Genotypes of animals used for each figure and video