Identifying neuronal aging targets in WT worms using neuron-specific RNA- sequencing.

(a) Wild-type learning and 1hr memory results on Day 1 and Day 7. Learning and memory results are represented as learning index (LI). Details of the LI calculation are explained in the methods. Learning, N = 10, memory, N = 5. ****: p < 0.01. Student’s t-test. (b) PCA plot for Day 1 (orange) and Day 8 (blue) neuronal bulk RNA-seq samples. (c) Volcano plot comparing age-associated differentially-expressed genes in WT neurons. Genes downregulated with age (orange) and upregulated with age (blue) were obtained by neuron-specific RNA sequencing of adult wild-type animals with neuron-specific GFP expression. Adjusted p-value < 0.001, log2(Fold-change) > 2. N = 6 biological replicates per age. 1146 genes were significantly downregulated with age (higher in young neurons) and 2016 genes were upregulated with age (higher in old neurons) (d) Tissue prediction scores for genes higher in young neurons. (e) GO terms of genes that decline with age in wild-type neurons. Synaptic and signaling GO terms are enriched in neuronal genes. p-value calculated using hypergeometric distribution probability. (f) Comparison of whole-body higher-in-young genes and neuronal higher-in-young genes. GO Terms and representative genes were performed using g:Profiler software. P-value of overlapping regions were calculated using a hypergeometric calculator. (g) Normalized reads of ins-6, unc-4, mec-7, folt-2, fbf-1, and madd-4, in Day 1 and Day 8 neurons in our dataset. P- adjusted values were calculated from DESeq2 software.

Genes that increase with age cause behavioral defects.

(a) Tissue Query for wild-type genes expressed at higher levels in aged worms show lower nervous system and neuron prediction score. (b) GO terms of genes expressed higher in aged neurons highlight transcription regulation and proteolysis. GO term analysis was done using Wormcat 2.0. (c) Normalized reads of utx-1 on Day 1 and Day 8. (d) Short-term associative memory (STAM) assay shows that adult-only utx-1 knockdown improves 1hr and 2hr memory of wild-type worms on Day 2. (e) Normalized reads of ins-19 on Day 1 and Day 8. (f) ins-19 mutation improves learning and memory in STAM on Day 3 of adulthood. (g) Normalized reads of nmgp-1 on Day 1 and Day 8. (h) nmgp-1 RNAi knockdown improves memory in STAM on Day 2. P-adj value of normalized count change generated from DEseq2 analysis. N = 5 plates in each behavioral experiment. Representative result of 3 biological repeats is shown. *: p<0.05. **: p<0.01. ***: p<0.001. ****: p<0.0001

Identifying neuronal IIS/FOXO targets in aged worms using neuron-specific RNA- sequencing.

(a) daf-2 mutants show better learning maintenance with age compared to N2 and daf-16;daf-2 worms. N = 10 plates in each condition. (b) daf-2 mutants show better memory maintenance with age compared to N2 worms. daf-16;daf-2 worms do not have 1hr memory on Day 1 of adulthood. N = 10 plates in each condition. (c-d) daf-2 mutants have a slightly larger learning span/lifespan ratio and memory span/lifespan ratio than N2 (wild type). Lifespan shown in Figure S3c. (e) PCA plot of Day 8 N2, daf-2, and daf-16;daf-2 neuronal RNA sequencing results. (f) Volcano plot of neuronal daf-2-regulated, daf-16-dependent up- and downregulated genes on adult Day 8 (Adjusted p-value < 0.05, log2(Fold-change) > 0.5, N = 6 biological replicates per strain). 570 genes were significantly upregulated and 814 genes were downregulated in daf-2 neurons compared with daf-16;daf-2. (g) Volcano plot of whole-worm daf-2 vs daf-16;daf-2 differentially-expressed genes during aging. 3154 genes are higher in daf-2, 1289 genes are higher in daf-16;daf-2 (log2[Fold-change(daf-2 vs daf-16;daf-2)] >1.5, p- adjusted <0.01). (h) Comparison of neuronal and whole-worm Day 8 daf-2 differentially-expressed genes (overlap p = 3.34E-63, hypergeometric test). Neuron-specific and shared daf-2 upregulated genes with the highest fold-changes are labeled.

List of top daf-2 vs daf-16;daf-2 upregulated genes with orthologs that have neuroprotective functions

Neuronal IIX/FOXO aging targets regulate memory decline with age in daf-2 worms.

(a) Comparison of neuronal Day 1 and Day 8 daf-2 vs daf-16;daf-2 upregulated genes. All shared genes and top Day 8-specific daf-2 upregulated genes are labeled. (b) daf-2-regulated fold-change profile of candidate genes. All candidates are upregulated in daf-2 mutants. (c) Description of candidate genes. log2(Fold-change) and p-adjusted values from the daf-2 vs daf-16;daf-2 comparison unless stated otherwise. (d) Candidate gene knockdown effects on Day 6 adult daf-2 learning (0hr after conditioning). Two candidate genes, dod-24 and F08H9.4, show a significant decrease in learning ability. N = 5 plates in each condition, the representative image of 3 biological repeats shown. (e) Candidate gene knockdown effects on Day 6 adult daf-2 short-term memory (1hr after conditioning). C44B7.5, dod-24, F08H9.4, mtl-1, and alh-2 showed significant decreases in memory. N = 5 plates in each condition, the representative image of 3 biological repeats shown. *: p < 0.05. **: p < 0.01. ****: p <0.0001.

Aged daf-2 neurons upregulate neuroprotective genes to maintain improved cognitive behaviors.

During normal neuronal aging, neuron-specific genes decrease in expression, while proteolysis and epigenetic regulators are upregulated, resulting in neuron dysfunction and cognitive function loss. In aged daf-2 mutants, upregulation of neuroprotective genes including dod-24, F08H9.4, C44B7.5, alh-2, and mtl-1 contribute to daf-2’s improved cognitive function. The diagram was generated using Biorender.

Aged neuron-specific sequencing.

(a-b) FACS results of neuron isolation. Over 99.94% of the cells collected are GFP+ neurons. 100,000 cells are collected for each biological replicate, 6 biological replicates for each condition. (c) Workflow of neuron isolation, library generation and sequencing. (d) Number of genes detected in Day 1 and Day 8 wild-type neurons. Genes with log2(TPM)>0.5 are considered expressed. (e) Down-sampling analysis for N2 Day 1 vs Day 8 indicates downsampling 30% of the data will still yield good results, indicating sufficiency of sequencing depth. (f) Normalized reads of txt-12, flp-33, and srd-23 in Day 1 and Day 8 neurons. P-adjusted values were calculated from DESeq2 software.

Whole-worm RNA-sequencing identifies whole-body changes during aging.

(a) Volcano plot of Day 1 vs Day 8 differentially-expressed genes during aging. 264 genes are expressed at higher levels in young worms, 1626 genes are higher in aged worms (log2[Fold-change (Day 1/Day 8)] >2.0, p-adjusted <0.001). (b) GO terms of genes that are expressed at higher levels in young (wild-type) whole animals highlight collagen and metabolism. (c) GO terms of genes that are expressed at higher levels in aged (wild-type) whole animals. GO terms were generated using Wormcat 2.0. (d) Tissue query for whole-worm aged-related genes highlights the alimentary system. (e) Full image of comparison of top wild-type neuronal and whole-worm differentially expressed with age genes. Related to Figure 2D.

Neuron-specific sequencing of Day 8 daf-2 and daf-16;daf-2 mutants.

(a-b) FACS results of neuron isolation. Over 99% of the cells collected are GFP+ neurons. 100,000 cells are collected for each biological replicate, 6 replicates for each genotype. (c) Representative image of N2 and daf-2 lifespan. (d) Number graph of lifespan, learning, and memory function. Related to Figure 4C and 4D. (e) Number of genes detected in Day 8 N2, daf-2, and daf-16;daf-2 neurons. (f) Ribosomal RNA depletion during sequencing. We used the library generation protocol with C. elegans-specific ribosomal RNA depletion kit (Tecan Genomics) successfully depleted rRNA to less than 20% of total reads. (g-h) Downsampling analysis for daf-2 vs daf-16;daf-2 and daf-2 vs N2. Both shows sufficient depth.

Whole-worm RNA-sequencing identifies changes in aged daf-2 mutants.

(a-b) GO term analysis of whole-worm daf-2-regulated genes shows enrichment in stress-resistant genes. (c) Comparison of neuronal and whole-worm Day 8 daf-2 differentially-expressed genes show high overlap (∼30%), but also identify genes specific to neurons and to the whole body. Related to Figure 3h. (d) Comparison of neuronal Day 1 and Day 8 daf-2 differentially expressed genes identifies a set of consistent and sets of changed genes.

DAF-16-dependent and -independent daf-2-regulated genes show different features.

(a) Tissue query for daf-2 vs daf-16;daf-2 differentially expressed genes. (b) Tissue query for daf-2 vs N2 differentially expressed genes. (c) Volcano plot of daf-2 vs N2 differentially expressed genes during aging. 1036 genes are more highly expressed in daf-2 mutants, 1285 genes are higher in N2 (log2[Fold-change(daf-2 vs N2)] >0.5, p-adjusted <0.05). (d-e) Neuronal Day 8 daf-2 vs N2 differentially expressed GO Terms. (f) Full image of comparison of neuronal Day 8 daf-2 vs daf-16;daf-2 and daf-2 vs N2 differentially expressed genes. Related to Figure 5B.

Comparison with recent sequencing datasets.

(a) Correlation between our whole-worm Day 8 daf-2 and daf-16;daf-2 sequencing and Class I and Class II gene rank from Tepper et al., 201345. High correlation indicates consistency between our isolated neuron RNA-sequencing results with former microarray results, despite the differences in approaches and ages. (b) Comparison with CeNGEN (L4 stage) gene expression data28 shows high correlation, with many genes only detected in our isolated neuron bulk-sequencing dataset (orange; wild-type Day 1 neuron sequencing data used in the comparison). Genes with an average log2(TPM) > 0.5 are considered detected. (c) Venn diagram showing genes detected in this dataset (orange), CeNGEN dataset (brown) and both (gray); only wild-type Day 1 neuron sequencing data were used in the comparison. (d) STAM with and without FUdR shows that 3 Days on FUdR does not affect learning and memory ability. (e-f) Comparison with gene expression data from the age-matched neuron cells from the Roux et al., dataset27 shows high correlation, with many genes only detected in our bulk-sequencing dataset. (g-h) Venn diagram showing genes detected in this dataset (orange or blue), Roux et al., dataset (brown) and both (gray). Only age-matched neuron cells from the Roux et al., dataset was used for gene expression analysis.