Endogenous fluorescence labeling of synaptic vesicle transporters

(Left) FLP-on and (Right) split GFP strategies to endogenously label synaptic vesicle transporters. (A) Cartoon representation of the CRISPR-engineered worm in the endogenous locus of the target synaptic vesicle gene. (B) Cell-specific driver that expresses (Left) Flippase or (Right) GFP1-10. (C) Resulting tagged synaptic vesicle proteins with a (Left) full-length GFP or mRuby3 or (Right) by reconstituting GFP in the cell of interest. (D) Schematic of the resulting toolkit to label and eliminate the endogenous machinery that packages or synthesizes glutamate, GABA, acetylcholine and monoamines. This panel was created using BioRender.com.

Cellular, genetic and molecular tools to probe neurotransmission in single cells.

Probing glutamatergic transmission in C. elegans

(A) Predicted protein structure (using AlphaFold Protein Structure Database) of EAT-4 (cyan) tagged with GFP (green) (Abramson et al., 2024). The last amino acid in the C-terminus end (W576) corresponds to the tagged residue. (B) Schematic of the Vesicular Glutamate Transporter (VGLUT/eat-4) gene structure, loss of function allele (ky5) and endogenously tagged versions built for this study and by others. eat-4 (ky5) mutants lack the first three exons. For cell-specific knock-in (KI) tools, GFP (syb8568) and mRuby (syb9193) FLP-ON cassettes (Schwartz & Jorgensen, 2016) were inserted before the STOP codon. To cell-specifically knock out (KO) eat-4, two FRT sites flank the coding sequence of the eat-4 gene (kySi76 kySi77) (Lopez-Cruz et al., 2019). When recombination takes place, the eat-4 coding sequence is removed and cytosolic mCherry is inserted in-frame to be expressed as a proxy for eat-4 sequence removal. (C) Mutations in the che-1 (prevents ASE development (Uchida et al., 2003)) (17.42 ± 7.7 mm) or eat-4 gene (13.15 ± 4 mm) results in disrupted migration across the salt gradient. Wild-type animals (7.76 ± 5.2 mm) migrate across the salt gradient similarly to EAT-4::GFP FLP-on animals that express (7.6 ± 3.6 mm) or not (4.9 ± 4.3 mm) flippase pan-cellularly. Animals that express flippase in all cells (5.61 ± 3.9 mm) migrate across the salt gradient as a wild-type animals. Results represent the mean distance of each worm from the salt peak, averaged across the final minute of the assay, with each dot representing a single animal. Plots are overlaid with Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; and NS means “not significant”. (D) (Top) Fluorescent image of an adult worm expressing endogenously labeled EAT-4 with GFP in all cells (Peft-3::Flippase). Scale Bar = 50μm. (Bottom) Zoom-in area of the head shows EAT-4 expressed predominantly in the nerve ring. Scale Bar = 10μm. (E) (Left) Electron microscopy rendering of ASE synapses in an L4 wild-type animal (White et al., 1986) (image generated with NeuroSC (Koonce et al., 2025). (Right) Fluorescent image of endogenously tagged EAT-4 protein cell-specifically in the ASE axons (Pflp-6::Flippase). Scale Bar = 10μm.

Probing GABAergic transmission in C. elegans

(A) UNC-47 predicted protein structure from AlphaFold Protein Structure Database. GFP11 tag (red) was added between amino acids E145 and N146. Complementary GFP1-10 (green) was modeled as bound to GFP11. (B) Schematics of the Vesicular GABA Transporter (VGAT/unc-47) loss of function allele and endogenously tagged versions built for this study. unc-47(e307) mutant animals have a single base pair substitution (G to A in the first nucleotide of exon 6) that results in a splicing acceptor mutant. Fluorescent tags GFP (syb6990) and mKate2 (syb7358) were inserted between amino acids E145 and N146. For cell-specific endogenous labeling, UNC-47 was tagged with one (syb7313) or three copies (syb7849) of GFP11. To silence GABA transmission, we flanked the Glutamic Acid Decarboxylase/unc-25 coding sequence with two FRT sites (syb5949 syb6275). Upon recombination, the unc-25 coding sequence is removed and nuclear (black) mCherry (purple) is designed to be in-frame and expressed as a proxy for unc-25 sequence removal. (C) (Top) Fluorescent image of an adult worm expressing endogenously labeled UNC-47 with reconstituted split-GFP in all cells (Peft-3::GFP1-10). Scale Bar = 50μm. (Bottom) Zoom-in area of the head shows UNC-47 expressed in the nerve ring and nerve cords. Scale Bar = 10μm. (D) (Left) Electron microscopy renderings of RIB synapses in an L4 wild-type animal (White et al., 1986) (image generated with NeuroSC (Koonce et al., 2025)). (Right) RIB-specific UNC-47 puncta (green) in-vivo using the UNC-47:GFP11 with reconstituted split-GFP in RIB (Psto-3b::GFP1-10) Scale Bar = 10μm. (E) unc-47(e307) mutant animals thrash significantly less per minute (30.8 ± 17) than wild-type animals (94.5 ± 10). Animals with non-reconstituted UNC-47::GFP11X3 (82.2 ± 9) thrash slightly less than wild-type animals (94.5 ± 10), while animals that express pan-cellular GFP1-10 (100.5 ± 9) are no different than wild-type animals. UNC-47::GFP11×3 animals that express Pan-cellular::Flippase (with reconstituted GFP) (88.6 ± 7) thrash similarly to wild-type animals. FRT-flanked unc-25 animals (syb5949 syb6275) that do not express Flippase (95.8 ± 12) thrash similarly to wild-type animals (91 ± 8). Cell-specific unc-25 knockout animals (syb5949 syb6275) that express Flippase in every cell (Peft-3::Flippase) (49.36 ± 19) thrash to the same extent as unc-25 (e156) mutant animals (40.2 ± 19). Plots are overlaid with Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; *** represents p<0.001; and NS means “not significant”.

Probing cholinergic transmission in C. elegans

(A) Predicted UNC-17 protein structure (magenta) tagged with GFP (green) on the last amino acid in the C-terminus end (W532). (B) Schematics of the Vesicular Acetylcholine Transporter (VACHT/unc-17) loss-of-function allele and endogenously tagged versions tested in this study (and built by others). unc-17(e245) mutant animals have a single base pair substitution in the third exon which leads to an amino acid change (arrowhead). Full-body knock-in animal labels UNC-17 with mKate (ot907). For cell-specific labeling, the N-terminus GFP FLP-ON cassette (syb7251) was inserted between amino acids P6 and V7 (See Methods). Similarly, the C-terminus GFP FLP-ON cassette (ola503) and mRuby FLP-on cassette (syb7882) were inserted before the STOP codon. (C) unc-17(e245) mutant animals (1.9 ± 2) thrash significantly less than wild-type animals (93 ± 10). UNC-17::GFP FLP-on animals that express pan-cellular::Flippase (88 ± 1) or animals that only express pan-cellular::Flippase (94.8 ± 12) are indistinguishable from wild-type animals in their thrashing behavior. UNC-17::GFP FLP-on animals (98.7 ± 8) thrash significantly more than wild-type animals. Mean ± Standard Deviation. Brown-Forsythe ANOVA test with Dunnett’s T3 multiple comparisons post hoc test. **** represents p<0.0001; * represents p<0.05; and NS means “not significant”. (D) (Top) Fluorescent image of an adult worm expressing endogenously labeled UNC-17::GFP in all cells (Peft-3::Flippase). Scale Bar = 50μm. (Bottom) Zoom-in area of the head shows UNC-17 expressed in the nerve ring, nerve cords and sub-lateral cords. DNC = Dorsal Nerve Cord, VNC = Ventral Nerve Cord. Scale Bar = 10μm. (E) (Left) Electron microscopy rendering of ADF synapses in an L4 wild-type animal (White et al., 1986) (image generated with NeuroSC (Koonce et al., 2025). (Right) Fluorescence image of endogenously tagged UNC-17::GFP protein specifically in the ADF neuron. . Scale Bar = 10μm.

Probing Monoaminergic transmission in C. elegans

(A) Predicted CAT-1 protein structure (blue) labeled with GFP11 (red) at C-terminus end (E145). Complementary GFP1-10 (green) was modeled with bound GFP11. (B) Schematic of the Vesicular Monoamine Transporter (VMAT/cat-1) loss-of-function allele and endogenously tagged versions used to monitor monoaminergic transmission. cat-1(ok411) mutant animals have a deletion that spans exons 7 and 8 (black line) of the cat-1 gene. For cell-specific labeling, three copies of GFP11 (syb7239) were inserted before the STOP codon. For cell-specific silencing of cat-1 activity, two FRT sites (ky1101 ky1118) flank the coding sequence of the cat-1 gene. Upon recombination, expression of cytosolic mCherry (magenta) is used as a proxy for deletion of the cat-1 coding sequence. (C) (Top) Fluorescence image of reconstituted CAT-1::GFP11×3 with expression of complementary GFP1-10 in the whole animal (Peft-3::GFP1-10). Scale Bar = 50μm. (Bottom) Zoom-in of the (Left) head and (Right) vulva area of an adult worm shows CAT-1 expressed in the nerve ring and pharyngeal neurons as well as in the vulva. Scale Bar = 10μm. (D) (Top) A well-fed day-1 adult animal is placed on NGM plates covered with a thin layer of bacteria, allowed to roam for 16 hours and the number of squares traveled was recorded. (Bottom) Wild-type animals (43.6 ± 10) roam less than cat-1(ok411) mutants (87.5 ± 12). CAT-1::GFP11×3 animals with GFP reconstituted (43.3 ± 20) (or not (56.8 ± 14)) roam similar to wild-type animals. Animals with pan-cellular expression of GFP1-10 (53.3 ± 20) also roam like wild-type animals. (E) (Top) A well-fed day-1 adult animal is placed on NGM plates covered with a thin layer of E. coli, allowed to roam for 20 hours and the aversion ratio was calculated. (Bottom) Wild-type animals (0.12 ± 0.07) display less aversion to E. coli lawns than ADF-specific cat-1 conditional KO animals (0.4 ± 0.1), consistent with previous reports using serotonin-depletion mutants (Feng et al., 2025). cat-1 conditional KO animals that do not express Flippase in ADF neurons (0.07 ± 0.03) are indistinguishable from wild-type. Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; ** represents p<0.01; * represents p<0.05; and NS means “not significant”.

Mapping of co-transmitter neurons in the C. elegans nervous system

(A) Schematic of the approach used to find co-expression of two vesicular transporters in the same cells. Flippase drivers (Element #1, blue) and Flippase-dependent cassettes that result in fluorescence (Element #2, orange) are used. Magenta cells are seen when both elements are co-expressed (in green box). (B) Strategy to track the co-expression of the vesicular glutamate and acetylcholine transporter in combination with the 4 most common neurotransmitters in C. elegans: Acetylcholine, GABA, Serotonin and Dopamine. “Elements #1” or “Element #2” refers to the genetic elements in the schematic in Fig 6A. (C-D) Specific example of the genetic strategy outlined in Fig 6A, for the vesicular transporters of glutamate (EAT-4) and acetylcholine (UNC-17). (C) We repurposed the eat-4 conditional KO strain (kySi76 kySi77) (Figure 1B) (Lopez-Cruz et al., 2019) in which the eat-4 gene coding sequence is flanked by two FRT sites and followed by cytosolic mCherry (Element #1). We crossed this line with a strain that has an endogenously inserted self-cleaving peptide sequence (T2A) followed by Flippase before the STOP codon in the unc-17 gene locus (Table 1). (D) (Top) Co-expression of EAT-4 and UNC-17 results in cytosolic mCherry. (Bottom) Cytosolic mCherry was detected in the head of three neurons: (i) AFDR, AFDL and M5; and in two neurons in the tail region: (ii) PVN, and DVA. (E) Fluorescence microscopy shows neurons with co-expression of (E) VGLUT/EAT-4 and VGAT/UNC-47; (F) VGLUT/EAT-4 and dopamine synthesis gene DAT-1; (G) VAChT/UNC-17 and VGAT/UNC-47; and (H) VAChT/UNC-17 and the serotonin synthesis gene TPH-1. All scale bars = 10μm.

Visualizing the distribution of cholinergic and serotonergic vesicles in ADF neurons.

(A) Dual-labeling of the endogenous acetylcholine (UNC-17, magenta) and serotonin (CAT-1, green) vesicular transporters in ADF neurons. This panel was created using BioRender.com. (B) UNC-17::mRuby and CAT-1::GFP overlap along the ADF axon. White arrowheads denote overlap of both signals. (B’) UNC-17::mRuby and CAT-1::GFP intensity profile. (C) Example of UNC-17::mRuby and CAT-1::GFP when they partially do not overlap along the ADF axon, green arrowheads point to CAT-1-only puncta. (C’) UNC-17::mRuby and CAT-1::GFP intensity profile. Scale bar = 10μm. (D) Schematic of the dual-labeling of endogenous CAT-1::GFP or UNC-17::GFP with endogenous active zone protein UNC-13::mScarlet along the ADF axon. This panel was created using BioRender.com. (E) Live imaging of the ADF axon reveals UNC-13::mScarlet puncta that lack CAT-1::GFP. Scale bar = 10 μm. (E’-E’’) Zoom-in region of E. White arrowheads show overlapping vesicular transporter tagged (green) with the active zone protein UNC-13 (magenta) in individual puncta. Magenta arrowhead points to UNC-13::mScarlet puncta that do not overlap with CAT-1::GFP. (F) Live imaging of the ADF axon reveals UNC-13::mScarlet puncta that lack UNC-17::GFP. Scale bar = 10 μm. (F’-F’’) Zoom-in region of F. White arrowheads show overlapping vesicular transporter tagged (green) with the active zone protein UNC-13 (magenta) in individual puncta. Magenta arrowhead points to UNC-13::mScarlet puncta that do not overlap with CAT-1::GFP. Scale Bar = 2 μm. (G) AiryScan imaging of dual-labeled UNC-17::mRuby and CAT-1::GFP along the ADF axon show localization in the same synaptic bouton but with distinct enrichment areas. Green arrowhead head points to CAT-1 enrichment and magenta arrowhead points to UNC-17 enrichment. Scale Bar = 2 μm.

(A) Graphical representation of conservation of the amino acid predicted structure for the Vesicular Glutamate Transporter across common model organisms (named in Figure S1B). Dark colors denote the least conserved region, while clear colors denote the highest conservation. Note the C-terminus end, where we introduced the GFP, is one of the darkest (least conserved) regions in the structure. (B) Sequence alignment of the amino acid sequences for the Vesicular Glutamate Transporter across common model organisms. Blue letters indicate conservation of the amino acid properties (basic, acid, neutral) but not identities. Gray represents columns with gaps and no conservation. (C) Schematic of chemotaxis assay. (Top) A concentration gradient of NaCl (∼60-85 mM) is established on assay plates (blue rim) by adding high concentration NaCl drops to the outer point of the assay plate at pre-determined times (See Methods). (Bottom) Larva stage 4 animals are picked, and twenty-four hours later, the young adult animals are transferred to NaCl 100mM training plates (red rim) for five hours. Animals are transferred to NGM buffer drops to wash off bacteria, before being placed on the assay plates (blue rim, placement at blue dot in schematic), where animal movement away from the highest [NaCl] is recorded (towards red dot in schematic). This panel was created using BioRender.com. (D) Displacement of trained wild type animals on a sham gradient (13.92 ± 5.3 mm) or untrained wild-type animals on a gradient of NaCl (13.84 ± 5.4 mm). Wild-type animals (7.76 ± 5.2 mm) migrate across the salt gradient like EAT-4::GFP FLP-on animals that express (8.36 ± 5.8 mm) flippase in ASE neurons. Animals that only express flippase in ASE neurons (7.61 ± 5.3 mm) migrate across the salt gradient similar to wild-type animals. Results represent the mean distance of each worm from the salt peak, averaged across the final minute of the assay, with each dot representing an individual animal. Plots are overlaid with Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; *** represents p<0.001; * represents p<0.05; and NS means “not significant”.

(A) Graphical representation of conservation of the Vesicular GABA Transporter amino acid predicted structure across common model organisms (named in Figure S2B). Dark colors denote the least conserved region, while clear colors denote the highest conservation. (B) Sequence alignment of the cytosolic loop between transmembrane domains 2 and 3 of the vesicular GABA transporter across common model organisms. Red letters indicate highly conserved columns (conservation of amino acid identity), and blue letters indicate conservation of the amino acid properties (basic, acid, neutral) but not identities. Gray represents columns with gaps. (C) Schematic of thrashing assay. “One thrash cycle” is scored when the animal bends as indicated in the schematic (from elongated-to-bent-to-elongated). We measured the number of thrashes per minute. This panel was created using BioRender.com. (D) unc-47(e307) mutant animals thrash significantly less (41.5 ± 21.35) than wild-type animals (109.2 ± 23), while over-expression of UNC-47::GFP (100.2 ± 16), CRISPR knock-in UNC-47::GFP (107.9 ± 29) and UNC-47::mKate2 (103.1 ± 18) CRISPR-tagged animals thrash similarly to wild-type animals. Plots are overlaid with Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; *** represents p<0.001; ** represents p<0.01; * represents p<0.05; and NS means “not significant”. (E-F) Fluorescent image of endogenously tagged (E) UNC-47::GFP (syb6990) and (F) UNC-47::mKate2 (syb7358). DNC = Dorsal Nerve Cord, VNC = Ventral Nerve Cord. Scale Bar = 10μm.

(A) All presynaptic sites from known GABAergic neurons in the C. elegans nerve ring (RIS, RME, and RIB), according to 3D electron microscopy of a Larva stage 4 (L4) wild-type animal (White et al., 1986) (image generated with NeuroSC (Koonce et al., 2025)). (B) 3D-rendering of UNC-47::GFP (green) puncta in the C. elegans nerve ring. A-P denotes anterior-posterior axis. Scale bar = 10 μm. (C) Addition of RIB::BFP into Figure S3B. Green arrowheads point to UNC-47::GFP puncta that overlap with RIB::BFP. Yellow arrowheads point to the synapses we interpret, based on our labeling and distribution, to belong to the RME neurons. (D) In-vivo reconstitution of UNC-47::GFP11×3 with an RIB-specific GFP1-10 construct. Scale bar = 10 μm. Note that the only synapses detected come from RIB and not the other GABAergic cells in the nerve ring (compare to Figure S3B). (E) In-vivo reconstitution of UNC-47::GFP11×3 in DD neurons when tagged with (Top) one or (Bottom) three copies of GFP11 (Pflp-13::GFP1-10) (He et al., 2019). (F-G) (F) Line scans and (G) quantification of reconstituted GFP fluorescence intensity with one (11.3 ± 6) or three copies (38.3 ± 28) of GFP11 in DD neurons. Scale bar = 10 μm. Plots are overlaid with Mean ± Standard Deviation. Mann-Whitney test. **** represents p<0.0001.

(A) Graphical representation of the amino acid conservation of the Vesicular Acetylcholine Transporter (VAChT) across common model organisms. Dark colors denote the least conserved region, while clear colors denote the highest conservation. Note the C-terminus end is one of the darkest regions in the structure, where the fluorophore was added. (B-C) Amino acid sequence alignment of VAChT across model organisms. Black arrowhead points to insertion site of fluorescent tag. (B) Sequences between transmembrane domains 6 and 7, which was tagged in olaEx5704 array (See panel E). (C) Sequence at the C-terminus end. Red letters indicate highly conserved columns (exact conservation of the amino acid identity), and blue indicates conservation of the amino acid properties (basic, acid, neutral) but not identities. Gray represents areas of no conservation. (D) Fluorescence image of CRISPR-tagged UNC-17::mKate2 (ot907) animal. Scale Bar = 10μm. (E) Schematic of UNC-17 predicted topology (magenta) along the membrane (gray). Green represents the locations of fluorescent tags tested for thrashing assays. Alleles, either extrachromosomal arrays expressed in unc-17 (e245) mutants or CRISPR knock-in strains in the unc-17 locus, are listed based on where the fluorescent tag was inserted. (F) unc-17(e245) mutant animals barely swim (1.7 ± 2). Over-expression of untagged Punc-17::UNC-17 (olaEx5703) (114.8 ± 18) and N-terminus-tagged UNC-17::GFP (olaEx5400) (91.1 ± 22) swim like wild-type animals (99.4 ± 20). GFP-tag between TM6-7 (olaEx5704) (34.3 ± 19) or CRISPR-insertion of C-terminus tag (ot907) (77.7 ± 17) leads to reduced swimming behavior when compared to wild-type animals. Insertion of GFP-FLP-on cassette at the N-terminus end (syb7251) (82.9 ± 16) or insertion of the GFP-FLP-on cassette at the C-terminus (ola503) (101.9 ± 16) thrash as well as wild-type animals (109 ± 25). Mean ± Standard Deviation. Kruskal-Wallis test with Dunn’s multiple comparison post hoc test. **** represents p<0.0001; ** represents p<0.01; * represents p<0.05; and NS means “not significant”.

(A) To track glutamate-involved co-transmission, we repurposed the eat-4 conditional KO strain (kySi76 kySi77) (Lopez-Cruz et al., 2019) where the eat-4 gene coding sequence is flanked by two FRT sites and followed by cytosolic mCherry (Element #2, orange box). Crossing this line with a panel of flippase drivers (Element #1, blue box) results in activation of cytosolic mCherry only in cells were both elements were co-expressed (Readout, green box). Pdat-1::Flippase and Ptph-1::Flippase driver strains were made available from (Muñoz-Jiménez et al., 2017). (B) To track acetylcholine-involved co-transmission, we repurposed the unc-17::GFP conditional KI strain (ola503) (Figure 3B) for which, after the unc-17 gene coding sequence, there are FRT sites flanking the STOP codon and 3’ UT, followed by GFP (Element #2, orange box). Crossing this line with a panel of flippase drivers (Element #1, blue box) results in the protein fusion of UNC-17 with GFP in cells where both elements were co-expressed (Readout, green box). (C-D) Systematic mapping shows neurons with co-transmission of (C) glutamate in combination with acetylcholine, GABA, and dopamine. (D) Co-transmission of acetylcholine is found in combination with GABA and serotonin.

Live-imaging of co-expression strategy of the eat-4 conditional knockout strain (kySi76 kySi77) with (A) endogenous Punc-47::UNC-47::T2A::Flippase and (B) Pdat-1:: Flippase driver. Live-imaging of co-expression strategy of the unc-17 conditional GFP knock-in strain (kySi76 kySi77) with (C) Ptph-1:: Flippase driver, and with (D) endogenous Punc-47::UNC-47::T2A::Flippase. (Top) DIC images. (Bottom) Fluorescence imaging. All scale bars = 10μm.

Our results are consistent with previous studies in the field (Taylor et al., 2021; Wang et al., 2024) and help establish an atlas of co-transmission in the C. elegans nervous system. Co-transmitter neurons are present in the (i) head, mid-body region and (ii) tail of the animal.

(A) Schematic of genes required for serotonergic identity. Mammalian (bold) and C. elegans (italics) homologue genes are listed. Detectable expression of (A’) tph-1, (A’’) bas-1 and (A’’’) mod-5 in ADF neuron (blue). Scale Bar = 5 μm. (B) Schematic of genes required for cholinergic identity. Mammalian (bold) and C. elegans (italics) homologue genes are listed. Detectable expression of (B’) cho-1 in ADF neuron (blue). Scale Bar = 10 μm. (C) (Left) Presynaptic densities (green) on ADF axons as determined by serial electron microscopy reconstructions (White et al., 1986) (image generated with NeuroSC (Koonce et al., 2025). (Right) Live-imaging of endogenous UNC-17::GFP and CAT-1::GFP specifically in ADF neurons. Panels A and B were created using BioRender.com.