https://doi.org/10.5061/dryad.1c59zw43k

Data for the two major experiments from Rahmani et al., eLife 2025, are shared here:

Chemotaxis assays

Chemotaxis assays were performed to test animals for their capacity to perform salt associative learning (Figs. 1, 4, 5, 6, S7, & S8) or salt aversive learning (Fig. S9). This involved manually counting populations of nematodes on chemotaxis assay plates for each experiment/figure. The relevant figures are listed below:

Figure 1: We tested salt associative learning for non-transgenic animals with ‘no TurboID’ (from line N2) versus transgenic animals with ‘TurboID’ (from line YLC207). Overall, TurboID enzyme expression in nematodes did not perturb the capacity to perform the learning paradigm. These animals were hence used in mass spectrometry experiments to screen neurons for proteins present during memory encoding of salt associative learning.

Figure 4: Three acetylcholine components were assessed for their importance in salt associative learning following TurboID experiments, by testing related single mutant lines. We found that mutations in genes acc-1, acc-3, and lgc-46 affected this form of learning.

Figure 5: Protein kinase A regulator KIN-2 has been linked to memory in nematodes previously through an olfactory learning paradigm, and was suggested here by TurboID to potentially affect salt associative learning. This was validated through testing a single mutant line (Fig. 5A) and a transgenic line overexpressing mutant kin-2 in neurons (with a wild-type background, in Fig. 5B).

Figure 6: Two proteins also assessed following TurboID experiments were armadillo-domain containing protein C30G12.6 and putative arginine kinase F46H5.3. This was done by assessing non-backcrossed and backcrossed single mutants, the latter of which were generated by backcrossing with wild-type/N2. We found that a mutation in F46H5.3, and not C30G12.6, was sufficient to impact salt associative learning capacity.

Figures S7 & S8: 20 additional proteins were explored in this study since they were suggested to be present by TurboID during memory encoding, through assessing salt associative learning in 20 respective single mutant lines. These animals learnt the same as wild-type.

Figure S9: Single mutants seen to display learning phenotypes significantly different from wild-type were then tested for salt aversive learning. F46H5.3 mutants displayed a learning phenotype in the same direction as salt associative learning, whereas other mutant lines displayed a wild-type phenotype.

Data description and File structure

File structure: The files are included in a zip folder: Chemotaxis_assay_data.zip. Spreadsheets can be opened using Microsoft Excel or a similar program. The data is separated by its respective figure, in that each file represents data for one figure (e.g., the file ‘Figure1B_data’ contains raw data for Figure 1B). Each figure/file contains up to eight biological replicates of chemotaxis assay data, separated as individual worksheets named ‘Replicate 1’, ‘Replicate 2’, …, and ‘Replicate 8’. Each biological replicate contains three technical replicates per experimental group, where data for one chemotaxis assay plate is equivalent to a technical replicate.

Data: Titles for columns are provided in row 1 for each worksheet (in black and bold). Chemotaxis assay data involves three experimental groups for each C. elegans line: (1) naïve (grey), (2) high-salt control (blue) or mock-conditioned (purple), and (3) trained (orange) or conditioned (yellow). Borders are used to annotate data specific to a C. elegans line (annotated in column A).

The number of animals (#) counted in regions of low salt, the origin, high salt, and outside regions are distinguished as individual columns (i.e., columns D, E, F, & G). A schematic is provided in Figure S2 of Rahmani et al. 2025 to visually define these regions for each chemotaxis assay plate. Counting data for a specific technical replicate is provided on the same row (i.e., from rows ≥ 2, with the technical replicate number in column C).

The total number of worms on each chemotaxis assay plate is provided in column G. The CI (or chemotaxis index) is determined for each technical replicate. This is done by calculating the difference between the number of worms on high salt versus low salt, weighted by the total number of animals on the plate, excluding those on the origin. This resulted in three CI values (for three technical replicates), used to determine the average CI for each experimental group per biological replicate – this is the data plotted in Figures 1, 4, 5, 6, S7, S8, & S9.

Transgenic C. elegans lines assessed in Figures 1B & 5B expressed transgenes from an extrachromosomal array – as a consequence, not all animals in a worm population were transgenic (non-transgenic siblings were used as controls). This is why the number of transgenic animals in each technical replicate is provided in files for Figures 1B & 5B in column K, excluding the origin, since this region does not affect CI values. Finally, the percentage of transgenic animals (versus non-transgenic animals) is provided in files for these figures in column L.

Mass spectrometry experiments

Proteins were extracted from animals and processed for mass spectrometry runs to facilitate protein identification. The aim of this experiment was to determine the proteins present in neurons during memory encoding.

This involved expressing the enzyme TurboID in all C. elegans neurons, and training animals by classical conditioning in an appetitive gustatory learning paradigm (i.e., salt associative learning). Namely, (1) TurboID was used to label neuronal proteins during the training phase of this learning paradigm, (2) >3000 whole animal bodies were lysed per experimental group to facilitate protein extraction due to their small size, and (3) from total protein, labelled proteins were separated from non-neuronal proteins by pull-down, so they could be processed for mass spectrometry runs.

Protein identities were then compared between experimental groups through a simple qualitative approach to identify >700 proteins unique to trained TurboID animals.

Data description and File structure

File structure: The files are included in a zip folder: Mass_spectrometry_data.zip. Files can be opened on a web browser. Five biological replicates of mass spectrometry data were generated for this study. Data is hence separated by their replicate in their respective folders (named ‘Replicate 1’, ‘Replicate 2’, …, ‘Replicate 5’). For technical reasons, we used two mass spectrometers– the ThermoFisher Scientific Q-Exactive Orbitrap (QE) and ThermoScientific Orbitrap Exploris (Exploris). Samples from biological replicates 1 and 2 were run on the QE, replicates 4 and 5 were run on the Exploris, and replicate 3 was run on both machines. This is why there are two folders named ‘Replicate 3’, where those in replicate 3a were from the QE, and the Exploris was used for data in the replicate 3b folder.

Each folder contains files generated for a specific biological replicate, which are labelled with the appropriate replicate number as ‘Replicate1’, ‘Replicate2’, …, ‘Replicate5’. Each file is specific to an experimental group that has TurboID/‘TbID’ (from line YLC207) or ‘no TbID’ (from line N2) – these labels define the type of C. elegans used to define an experimental group. Animals underwent salt associative learning experiments as ‘trained’ or high-salt ‘control’ (as mock-training) – this defines the treatment strategy for the animals. Each experimental group has one file labelled ‘no-elution’ and another called ‘elution’ for reasons explained below (as the ‘sample type’). Each file is named from left to right as the replicate number, the C. elegans type, the treatment strategy, and then the sample type.

Total protein from these animals underwent pull-down to enrich proteins from neurons present during training or mock-training (i.e., labelled by TurboID), and digested by trypsin into peptides. The enrichment process involved binding labelled proteins to magnetic beads in a solution, and then the labelled proteins were exposed to trypsin to promote protein digestion. This was expected to cut peptides from proteins bound to the beads, such that the peptides fell into the solution that the beads were in, and peptides labelled by TurboID would remain bound to the beads. The solution containing fallen peptides was run into a mass spectrometer, and the data were labelled as ‘no elution’. Correspondingly, bound peptides were eluted from the beads, and then run through the same mass spectrometer – the data from this sample was called ‘elution’.

Data: Each file contains peptide summary report files (in .html format) exported from the software MASCOT, which assigns protein identities to peptides detected during mass spectrometry runs. From top-to-bottom, each file contains (1) a summary list of protein identities determined by MASCOT, ranked in ascending order based on the confidence that the software has that the identity was not a false positive, (2) tables separated by each protein identity, containing the individual peptide/s determined by MASCOT to suggest the protein identity, and (3) a list of peptides that could not be assigned a protein identity by MASCOT.