Evidence for declining lysosome function during aging in C. elegans neurons

To test the hypothesis that neuronal lysosomal dysfunction is a pathology of aging in vivo, we examined the posterior ventral process D (PVD) neurons. These two bilaterally symmetric, glutamatergic sensory neurons provide an accessible model to interrogate lysosomal function with high resolution. Lysosomal degradation occurs predominantly in the neuron cell body, and as in other organisms, the lysosomal compartments within C. elegans neurons mostly localize to the cell body [43]. We therefore focused on the PVD neuron cell body and assessed two established features of lysosomal dysfunction: increased luminal pH and buildup of degradative cargo within lysosomal compartments [44].

First, we developed a tool to measure relative lysosomal acidity. Healthy lysosomes have a luminal pH of about 5, and a higher pH indicates lysosomal dysfunction [3,45]. NUC-1, the ortholog of human DNAse II, is a luminal lysosome resident and a previously validated marker for lysosomal compartments—lysosomes, endolysosomes, and autolysosomes—in C. elegans [46]. We co-expressed NUC-1::GFP and NUC-1::RFP alongside a BFP morphology marker in PVD neurons (Figs 1A, S1A, and S2A). GFP fluorescence is partially quenched in the acidic environment of lysosomal compartments, while RFP intensity is unaffected by the lysosomal pH and is used for normalization. Therefore, the GFP/RFP ratio represents relative acidity: a higher ratio indicates reduced acidity (i.e., higher pH). We first validated our tool using wild-type (WT) worms continuously treated with 0.03 μM concanamycin A, which inhibits vacuolar H⁺ ATPase, from Day 0 of adulthood. Relative lysosomal acidity is decreased in Day 6 adults post-treatment compared to untreated animals, demonstrating the serviceability of our tool (Fig 1B). Lysosomal acidity in Day 3 concanamycin A-treated adults is not significantly different from controls, indicating that this low dose of concanamycin A requires time to get into neurons and impede lysosomal function. We also injected Day 3 adult WT worms with 25 μM concanamycin A, and we found that the relative lysosomal acidity is decreased 3 h post-concanamycin A injection compared to mock drug injection. This indicates that a high, acute dose of concanamycin A is able to disrupt lysosomal acidity in early adulthood, as expected, further validating the tool (Fig 1C). We then examined WT worms at different ages and found that relative neuron lysosome acidity decreases in aged worms by Day 9 of adulthood (Fig 1D). For comparison, the median life span of worms is about 2 weeks (S2B Fig) [19,47–49].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. PVD neurons show reduced lysosome function during aging and in the hlh-30 mutant.

(A–D) A fluorescent reporter for luminal acidity of neuronal lysosomal compartments shows that acidity decreases with aging and in the hlh-30 mutant. (A) Example images of the PVD neuron cell body in a Day 3 adult carrying the carEx6 transgene, which stably expresses all three constructs shown. White dashed lines indicate the edge of the cell body, blue dashed lines indicate the edge of the lysosomal compartments. Scale bar = 10 μm. (B) Quantification of NUC-1::GFP/RFP ratio using the carEx6 transgene. ***P < 0.001, **P < 0.01, *P < 0.05, pairwise comparisons not labeled show no significant difference, Two-way ANOVA with Sidak post-test. (C) The NUC-1::GFP/RFP ratio is increased 3 h after injecting concanamycin A into early-adult (Day 3) worms compared with mock drug injection. *P < 0.05, two-tailed Student t test. (D) The NUC-1::GFP/RFP ratio increases with age in WT. ****P < 0.0001, **P < 0.01, ns: not significant, One-way ANOVA with Tukey post-test. (E–H) Images and quantification of a representative degradative cargo for lysosomes, the synaptic vesicle transmembrane protein SNT-1, within lysosomal compartments. (E) Representative images of WT animals. White dashed lines indicate the edge of the cell body, blue dashed lines indicate the edge of the lysosomal compartments. Scale bar = 10 μm. (F) SNT-1::GFP shows increased accumulation in lysosomal compartments with age. ***P < 0.001, *P < 0.05, linear mixed-effects model with animal as a random effect and genotype/treatment and age as fixed effects, followed by pairwise comparisons with Tukey correction. Statistical comparisons were performed on log-transformed data to meet model due to heteroscedasticity in the untransformed data. (G) The amount of degradative cargo SNT-1 in individual lysosomal compartments shows a bimodal distribution. Threshold = 529 AU. (H) Aging, chronic concanamycin A exposure, and the hlh-30 mutant each result in an increased portion of lysosomal compartments with an atypically high amount of SNT-1::GFP. ** P < 0.01, * P < 0.05, pairwise comparisons not labeled show no significant difference, Fisher’s exact test with Holm correction. ConcA, concanamycin A. The data underlying the graphs shown in the figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002957.g001

Next, we aimed to assess the efficiency of lysosomal degradation by quantifying the accumulation of a degradative cargo, the synaptic vesicle protein Synaptotagmin/SNT-1, within lysosomal compartments in the neuron cell body. We measured the fluorescence intensity of endogenously tagged SNT-1::GFP that co-localized with NUC-1::RFP, which represents SNT-1 protein that has been sorted into a lysosomal compartment but has not been fully degraded, and the GFP has not been fully quenched (Figs 1E–1H and S1B) [50,51]. The intensity of SNT-1::GFP in lysosomal compartments increases with age, which could be indicative of age-related lysosomal dysfunction (Fig 1F). Chronic exposure to concanamycin A does not exacerbate this accumulation (Fig 1F). This suggests that the age-associated increase of SNT-1::GFP within lysosomal compartments may not, in itself, be a pathology caused by lysosome dysfunction. On the other hand, autophagosome maturation and retrograde transport of lysosomal degradative cargoes requires acidification by the vacuolar H⁺ ATPase, meaning that concanamycin A may not only reduce lysosome acidity but also disrupt delivery of retrograde cargoes, including SNT-1 to the lysosome [52].

By plotting the SNT-1::GFP fluorescence intensity of all lysosomal compartments from WT, we observed a bimodal distribution, suggesting the presence of two distinct lysosomal compartment populations: typical lysosomal compartments and atypical lysosomal compartments (Fig 1G). The percentage of atypical lysosomal compartments increases by Day 6 of adulthood (Fig 1H). Chronic concanamycin A treatment caused an increase in the percentage of atypical lysosomal compartments by Day 3 of adulthood, to a degree similar to that of older, untreated WT animals (Fig 1H). Considering that both aging and concanamycin A treatment led to an increased percentage of atypical lysosomal compartments, we speculate that this phenotype arises from reduced degradative functionality of the lysosomes. An alternative interpretation is that the same phenotype arises with concanamycin A treatment and older WT animals via two distinct mechanisms. In this alternative model, the increased percentage in older animals could arise from an increase in the amount of degradative cargo delivered to the lysosomes. These two models—age-associated lysosome dysfunction versus increasing degradative flux—are considered further below.

HLH-30 is required for adult neuronal lysosomal function

We next aimed to understand how neurons generate adequate lysosomal function in early adulthood. TFEB is a master regulator of lysosome and autophagosome biogenesis in mammalian cells, but it has been predominantly shown to function during starvation and stress [53]. Whether and how TFEB functions in healthy neurons in vivo is unknown. We first asked whether loss of hlh-30, the C. elegans homolog of TFEB, affects the late endosome, endolysosome, and lysosome population in the neuron. The hlh-30(tm1978) allele is a frameshifting deletion and putative null allele. The hlh-30(tm1978) mutant, hereafter referred to as the hlh-30 mutant, is viable and displays WT animal morphology and behavior [19]. The morphology of the PVD neuron in the hlh-30 mutant appears indistinguishable from WT at Day 1 of adulthood (Fig 4A–4C). The RAB-7 GTPase is a trafficking factor of late endosomes and autophagosomes, and it localizes to late endosomes, endolysosomes, and autolysosomes [38]. Using PVD-expressed mCherry::RAB-7, we found that the number of RAB-7-labeled compartments in the neuron cell body increases between Day 0 of adulthood (the L4 larval stage) to Day 3 of adulthood (Fig 2A and 2B) [54]. In the hlh-30 mutant, the number of RAB-7 labeled compartments is similar at Day 0 but significantly reduced at Day 3 of adulthood compared to WT (Fig 2B). This reduction at Day 3 can be rescued by expressing hlh-30::gfp specifically in the PVD neurons, indicating that HLH-30 regulates the lysosomal compartment abundance cell-autonomously (Fig 2C). We observed the same trends when we quantified the abundance of endogenously-tagged GFP::RAB-7 compartments in WT versus hlh-30 mutant animals, indicating that the accumulation of RAB-7-labeled lysosomal compartments in adult worms and their reduced abundance in hlh-30 mutants are not due to an overexpression artifact (S2C Fig ) [54]. These data suggest that HLH-30 promotes late endosome, endolysosome, and/or autophagosome abundance in early adulthood but is dispensable in development.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. HLH-30 functions cell-intrinsically to promote neuronal lysosomal function in early adulthood.

(A–C) The hlh-30 mutant has fewer RAB-7::mCherry-labeled endosomes in the PVD neuron cell body by Day 3 of adulthood compared to WT, and this reduction can be rescued by PVD-specific expression of HLH-30. White dashed lines indicate the edge of the cell body. Scale bar = 10 μm. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns: not significant, One-way ANOVA with Tukey post-test. (D–G) Quantification of cell body lysosomal compartment number and size using super-resolution fluorescence microscopy of NUC-1::RFP. (D) Example images of lysosomal compartments in PVD neuron cell body in WT, hlh-30 mutant, and WT with chronic concanamycin A treatment at adult Day 3. White dashed lines indicate the edge of the cell body. Scale bar = 10 μm. (E) Number of lysosomal compartments varies between individual neurons within the same age/genotype/treatment group and does not show strong changes by age, genotype, or treatment. *P < 0.05, pairwise comparisons not labeled show no significant difference, Two-way ANOVA with Tukey post-test. (F) The size of the largest lysosomal compartments increases in early adulthood, between Day 1 and 3, then remains stable in WT animals. Both the hlh-30 mutant and concanamycin A treatment cause reduced size of the largest lysosomal compartments by mid-adulthood. ***P < 0.001, *P < 0.05, pairwise comparisons not labeled show no significant difference, linear mixed-effects model with animal as a random effect and genotype/treatment and age as fixed effects, followed by pairwise comparisons with Tukey correction. Statistical comparisons were performed on log-transformed data to meet model assumptions due to heteroscedasticity in the untransformed data. (G) Total lysosomal compartment volume in the neuron cell body increases between adult Day 1 and Day 3 in WT animals but not in the hlh-30 mutant. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, pairwise comparisons not labeled show no significant difference. Statistical comparisons were performed using a linear regression model on square root-transformed data to meet model assumptions, followed by pairwise comparisons with Tukey correction. (H) Neuronal depletion of HLH-30 shows increased lysosomal compartment luminal pH compared to WT at adult Day 6. ****P < 0.0001, **P < 0.01, One-way ANOVA with Tukey post-test. (I) SNG-1::ARGO turnover rate at the synapses, calculated from one-phase exponential decay curves. ****P < 0.0001, Extra sum-of-squares F test. ConcA, concanamycin A. The data underlying the graphs shown in the figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002957.g002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Basal HLH-30 activity is not accompanied by enriched nuclear localization and declines during neuronal aging.

(A, B) Endogenous HLH-30, visualized with split GFP, is mainly localized in the neuron’s cytoplasm in unstressed adulthood. Scale bar = 10 μm. ****P < 0.0001, ns: not significant, Fisher’s exact test. (C) The bHLH binding sequence is similar between human TFEB and Caenorhabditis elegans HLH-30 [61]. (D) High score of HLH-30 binding sequences is found in the rab-7 promoter-plus-enhancer region. (E, F) A fluorescent reporter of rab-7 transcription shows that PVD-specific rab-7 transcription increases in early-adults (Day 3) and decreases during aging in WT. In the hlh-30 mutant, the PVD-specific rab-7 transcription level is lower compared to WT at the same age, and there is no difference between Days 0 and 3. There is still a slight decrease between Days 3 and 6. Scale bar = 10 μm. The outline of the PVD neuron was generated using the PVD > bfp fluorescence in this strain, which is not shown. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns: not significant, Two-way ANOVA with Tukey post-test. (G) Relative rab-7 mRNA abundance from whole-worm transcriptomic data shows reduced rab-7 mRNA level in aging WT worms. **P < 0.01, ns: not significant, One-way ANOVA with Tukey post-test [62]. (H) Relative hlh-30 mRNA abundance from whole-worm transcriptomic data shows reduced hlh-30 mRNA level in aging WT worms. *P < 0.05, ns: not significant, One-way ANOVA with Tukey post-test [62]. The data underlying the graphs shown in the figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002957.g003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Loss of HLH-30 accelerates age-related dendrite degeneration and sprouting.

(A–E) In WT adults, the PVD dendrite shows progressive dendrite degeneration (B and D) and sprouting (C and E) with age, and these phenotypes manifest earlier in the hlh-30 mutant. Note that WT Day 11 shows similar penetrance and severity as the hlh-30 mutant Day 6. In (A), the blue and red dashed boxes on the hlh-30 Day 6 animal, which correspond to the zoomed-in images on the right with the same outline colors, show examples of severe dendrite degeneration and severe sprouting, respectively. Scale bar = 50 μm (whole worm) or 10 μm (zoom in). ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns: not significant, Fisher’s exact test. Arrowheads point to degeneration or sprouting (not all instances are indicated). (F, G) Pan-somatic expression of hlh-30 rescues the dendrite degeneration (F) and sprouting (G) phenotypes in Day 6 hlh-30 mutants, and PVD-specific or pan-neuronal hlh-30 expression partially rescues the dendritic defects. ****P < 0.0001, ***P < 0.001, *P < 0.05, ns: not significant, Fisher’s exact test. (H, I) WT animals exposed to concanamycin A treatment and cup-5(lf) mutant animals both show increased percentage and severity of dendrite degeneration (H) and sprouting (I) compared to WT at Day 6. ****P < 0.0001, ***P < 0.001, **P < 0.01, Fisher’s exact test. (J, K) Depleting HLH-30 in the whole worm (except the germline) phenocopies the hlh-30 mutant, showing increased dendrite degeneration (J) and sprouting (K) compared to WT at the same age. Note that for these experiments, all animals were cultivated in the presence of auxin, and they were scored at a later age than (A–I) because the auxin caused slowed growth and aging. (J) Pan-neuronal, muscle, and hypodermal-specific HLH-30 depletion all show increased degeneration compared to WT. **P < 0.01, ns: not significant, Fisher’s exact test. (K) Pan-neuronal HLH-30 depletion causes increased sprouting compared to WT. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns: not significant, Fisher’s exact test. (L, M) PVD-specific overexpression of hlh-30 in the WT genetic background does not have a significant impact on dendrite degeneration (L) and reduces severity of aging-associated sprouting (M). ****P < 0.0001, ns: not significant, Fisher’s exact test. In all panels of this figure, the hlh-30(tm1978) mutant is labeled as hlh-30(o) for simplicity. The data underlying the graphs shown in the figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3002957.g004

Chronic exposure to concanamycin A did not result in significantly fewer mCherry::RAB-7-labeled endosomes in the cell body (Fig 2B). This may be because it takes more time for this low dose of concanamycin A to get through the cuticle and impact lysosomal compartments.

Next, we quantified NUC-1::RFP-labeled lysosomal compartment number and volume in WT versus hlh-30 mutants (Figs 2D–2G and S2D). We used super-resolution microscopy for the improved ability to detect clusters of closely spaced lysosomal compartments versus large compartments compared with standard confocal microscopy (S2E Fig).

In WT, we observed a wide range of lysosomal compartment numbers and sizes between neurons in age-matched animals (Fig 2E and 2F). Within the neuronal cell body, both small lysosomes and larger endolysosomes or autolysosomes can be observed at any given time (Figs 2D and S2E). The mean size of individual NUC-1::RFP-labeled lysosomal compartments increases between Day 1 and 3 of adulthood, then remains stable through Day 11 (Fig 2F). This difference is driven by the larger size of the largest lysosomal compartments in older adults. The total lysosomal compartment volume within the neuron cell body also increases between Day 1 and 3 of adulthood, by close to 6-fold (from 1.1 μm³ to 6.4 μm³, P < 0.0001) (Fig 2G). For comparison, the neuron cell body size is similar between Day 1 and Day 3 adults but increases significantly between Day 3 and Day 6 (S2F Fig).

We predicted that the number of NUC-1::RFP-labeled compartments would be decreased in the hlh-30 adults, showing a similar trend as mCherry::RAB-7-labeled late endosomes, endolysosomes, and autophagosomes, but this was not the case (Fig 2E). The average individual lysosomal compartment size in the hlh-30 mutant appeared similar to WT at adult Day 1, and the total lysosomal compartment volume within each neuron was, unexpectedly, significantly greater in the Day 1 hlh-30 mutant (3.3 μm³) compared to WT (1.1 μm³, P < 0.01) (Fig 2F and 2G). Unlike in WT, however, the hlh-30 mutant did not substantially increase either the size of individual lysosomal compartments or the total lysosomal compartment volume between Days 1 and 3 (Fig 2F and 2G). These results suggest that HLH-30 is required to expand lysosomal compartment capacity in the neuron cell body between Days 1 and 3 of adulthood.

We quantified the number and size of NUC-1::RFP-labeled lysosomal compartments in WT animals chronically treated with concanamycin A to assess how lysosomal compartment behavior changes when lysosomal acidity is disrupted. Similar to the hlh-30 mutant, concanamycin A treatment led to a reduction in the average volume of individual lysosomal compartments by Day 6 of adulthood (Fig 2F). At the same time point, concanamycin A-treated animals showed an increase in lysosomal compartment number, suggestive that concanamycin A may impair lysosome fusion dynamics (Fig 2E).

Based on these results, we hypothesized that HLH-30 is required to generate adequate lysosomal capacity in adult neurons. If this is the case, then hlh-30 mutants might exhibit accelerated neuronal lysosomal dysfunction in adulthood compared to WT. Using the PVD lysosome acidity reporter (Fig 1A–1D), we found that the NUC-1::GFP/RFP ratio is higher in hlh-30 mutants compared to WT at Day 3 and Day 6 of adulthood, indicating that lysosomal compartment acidity is indeed disrupted in hlh-30 mutants (Fig 1B). To test whether the impact of HLH-30 on lysosomal acidity maintenance is cell-autonomous, we used the Auxin-inducible degradation (AID) system, wherein we inserted the AID sequence into the hlh-30 genomic locus to tag the C-terminus of HLH-30 (S1F Fig). When the hormone auxin is added, the tissue-specifically expressed TIR1 E3 ubiquitin ligase recognizes the AID-tagged HLH-30 and promotes its degradation [55]. This technique allows us to deplete HLH-30 in specific tissues. Neuronal depletion of HLH-30 is able to deplete most of HLH-30 protein in PVD neurons (S3A–S3C Fig). Lysosomal acidity under neuronal HLH-30 depletion is significantly reduced compared to WT (Fig 2H). The most parsimonious interpretation of this result is that HLH-30 promotes neuronal lysosomal compartment acidity cell-autonomously, though it is possible that it functions non-cell-autonomously from other neurons (Fig 2H).

Next, we assessed accumulation of degradative cargo SNT-1::GFP in the lysosomal compartments of hlh-30 mutant neurons (Fig 1F and 1H). We found that the percentage of lysosomal compartments with atypically high levels of SNT-1::GFP is increased in the hlh-30 mutant by Day 3 of adulthood (Fig 1H). This is consistent with the model that lysosomal degradation is less efficient in the absence of HLH-30 function, though an alternate interpretation is that the amount of cargo that is delivered to lysosomal compartments for degradation is increased, in which case the efficiency of degradation could be unchanged.

To determine which of these models is more accurate, we set out to directly test the notion that the HLH-30-mediated expansion of lysosomal functionality is necessary for efficient turnover of degradative cargo. To do this, we applied our recently developed Analysis of Red-Green Offset (ARGO) method, wherein we endogenously tagged another characteristic cargo of lysosomal degradation, the synaptic vesicle protein Synaptogyrin/SNG-1, with both GFP and RFP (S1C and S4A Figs) [56]. The ARGO method involves a pulse-chase component in which the gene encoding gfp is excised via Cre/LoxP recombination with the pulse, and then the neuron is periodically imaged to quantify the ratio of GFP/RFP intensity at the synapses during the chase. A one-phase exponential decay curve was fitted to these data to calculate the presynaptic half-life of SNG-1. We induced the visualization of SNG-1 specifically in the cholinergic motor neuron Dorsal A type motor neuron 9 (DA9) using FLP/FRT recombination. We selected the DA9 neuron for this experiment because A) it allows us to investigate the turnover dynamics of a synaptic vesicle protein, and the DA9 neuron serves as a well-established model for examining presynapses; and B) we aimed to determine whether HLH-30 promotes lysosomal capacity broadly across neurons or specifically in PVD neurons.

First, we examined steady-state SNG-1::ARGO fluorescence intensity in control animals that were not “pulsed.” We observed no difference in presynaptic localization, organization, or number between hlh-30 mutants and WT animals (S4B and S4C Fig). The presynaptic fluorescence intensity of SNG-1::ARGO RFP is slightly decreased in the hlh-30 mutant compared to WT, suggesting a slight reduction in synaptic vesicle numbers per synapse (S4B and S4D Fig). Consistent with the framework that endolysosomal protein degradation occurs in the neuron cell body, we have observed that endosomes in the cell body show a lower SNG-1::ARGO GFP/RFP fluorescence intensity ratio compared to the presynapses in WT animals (S4E and S4F Fig, comparison not showed) [56]. In these cell body-localized endosomes, the GFP fluorescence from SNG-1::ARGO is partially quenched, indicative of the low pH environment of degradative endolysosomes (S4E Fig) [56].

To test this interpretation, we injected Day 3 adults with concanamycin A, imaged them 3 h post-injection, and quantified SNG-1::ARGO GFP/RFP ratio at the DA9 neuron cell body and synapses. Indeed, concanamycin A injection increases SNG-1::ARGO GFP/RFP ratio of endosomes in the cell body (S4G Fig). We also noticed that the GFP/RFP ratio in the cell body is lower than at the synapses in hlh-30 mutants, suggesting that the location of SNG-1 degradation—lysosomal compartments in the neuron cell body—is not altered in the hlh-30 mutant (S4E, S4F, S5A, and S5C Figs, comparison not shown). Furthermore, concanamycin A injection increases the GFP/RFP ratio at the synapses (S4H Fig). Previous research indicates that synaptic vesicle proteins are sorted for degradation into late endosomes at the presynapses [7,8,57]. Our result likewise indicates that SNG-1 is sorted for degradation into acidic compartments—likely late endosomes—at the presynapses, causing some GFP to quench.

Next, with the ARGO pulse at Day 2 of adulthood, we found that the SNG-1 presynaptic turnover rate is slower in the hlh-30 mutant (half-life = 4.9 d, 95% C. I. 4.3–5.7 d) compared to WT (half-life = 3.1 d, 95% C. I. 2.7–3.5 d) (Figs 2I, S5B, S5C, S5F, and S5G). Interestingly, the steady-state SNG-1::ARGO GFP/RFP ratio (in the absence of ARGO activation for pulse-chase imaging) is lower at the presynapses in the hlh-30 mutant compared to WT at Day 2 and 4 of adulthood (Figs 2I and S4F). This result suggests that in the hlh-30 mutant, a larger proportion of presynaptic SNG-1 resides in late endosomes rather than in synaptic vesicles compared to WT. Considering this change in steady-state presynaptic SNG-1::ARGO GFP/RFP along with the slower SNG-1 turnover rate in the hlh-30 mutant, it suggests that there is a backup in the clearance of sorted-for-degradation SNG-1 from the presynapses in the hlh-30 mutant.

The slower turnover of SNG-1 in the hlh-30 mutant compared to WT motivated us to take a closer look at the steady-state fluorescence of SNG-1::ARGO-labeled compartments in the cell body, which are lysosomal compartments based on their low pH (S5A, S5D, and S5E Fig). Notably, in the hlh-30 mutants, there are fewer and smaller SNG-1::ARGO-labeled endolysosomes in the neuron cell body, suggesting reduced delivery of SNG-1 to lysosomal compartments (S5D and S5E Fig). Further, the hlh-30 mutants show a higher SNG-1::ARGO intensity ratio in the neuron cell body compared to WT, suggestive of dysfunctional endolysosomal degradation (S4E Fig).

Taken together, these results indicate that HLH-30 is required for maintaining neuronal lysosomal function in early adulthood, whereas loss of hlh-30 leads to inadequate lysosomal degradation.

Basal HLH-30 activity in neurons operates without nuclear enrichment and declines in aging

During starvation and stress, TFEB/HLH-30 translocates from the cytoplasm to the nucleus [20–22,58]. The subcellular localization of TFEB has been used as a key indicator of its activity status, wherein nuclear enrichment means it is active and cytosolic enrichment means it is inactive. We therefore wondered whether neuronal HLH-30 shows enriched nuclear localization when it is functioning in well-fed, unstressed young adults to expand neuronal lysosomal capacity. We used split GFP to visualize the sub-cellular localization of endogenously tagged HLH-30::GFP11 in neurons (Figs 3A and S1D) [59]. Neuronal HLH-30::GFP11 shows enriched nuclear localization upon heat stress, consistent with findings in other cell types (Fig 3A and 3B) [20,21]. However, HLH-30::GFP11 avoids the nucleus in the steady-state of both early- (Day 3) and mid-adult (Day 6) PVD neurons (Fig 3A and 3B). HLH-30::GFP11 is not completely absent from the nucleus, though, so it can still be affecting some transcriptional regulation (Fig 3A and 3B). When the animals are starved for 1 day, neuronal HLH-30::GFP11 also avoids the nucleus (Fig 3A and 3B). We speculate that the PVD neurons might be protected from starvation, preventing them from experiencing or sensing it during our starvation manipulation. Recent research suggests that HLH-30 activity is not always correlated with its nuclear localization [60]. Our data suggest that, although acute stress can increase neuronal HLH-30 activation by promoting its nuclear localization, the basal HLH-30 activity that expands lysosomal capacity in healthy, unstressed adulthood occurs in the absence of nuclear enrichment.

We hypothesized that basal HLH-30 activity declines in aged neurons, contributing to the aging-associated reduced lysosomal function. To test this hypothesis, we generated a real-time fluorescent transcriptional reporter of rab-7, wherein we express GFP11::nls::PEST from the endogenous promoter-plus-enhancers of rab-7 and use neuronally-expressed GFP1-10 to reconstitute GFP fluorescence specifically in the neurons (S1E Fig). RAB7 is a known transcriptional target of TFEB in mammalian cells [12]. HLH-30 shares a similar bHLH binding sequence as TFEB, and there are several predicted binding sites of HLH-30 in the promoter and enhancer region of rab-7 (Fig 3C and 3D) [61]. The GFP11::nls::PEST is transcribed together with rab-7 into the same mRNA, which is then alternatively spliced to separate the GFP11::nls::PEST mRNA from the rab-7 mRNA, to be translated separately. The PEST sequence targets protein for rapid degradation, so the fluorescence intensity of GFP11 reports the transcription level of rab-7 within a short period, instead of the cumulative transcription level [63]. The NLS (nuclear localization signal) directs GFP11 to the nucleus, facilitating image quantification [64]. The transcription level of rab-7 in WT PVD neurons increases in early adulthood (Day 3), corresponding with the expansion of the number of late endosome and lysosomal compartments. Then, it decreases in older neurons (Day 6) by 40 ± 5%, anticipating the age-associated decline of lysosomal function (Fig 3E and 3F). In hlh-30 mutants, the transcription level of rab-7 in the neuron is always lower compared to WT at the same age (Fig 3F). Furthermore, in contrast to WT animals, we detected no difference in rab-7 transcription level between Day 0 and Day 3 in hlh-30 mutant (Fig 3F). The fluorescence intensity of the reporter decreases between Day 3 and 6 in the hlh-30 mutant by 26 ± 4%. This indicates that some of the decrease observed in WT is due to reduced HLH-30 activity, and there is also an hlh-30-independent mechanism that contributes to the declining rab-7 expression between Day 3 and 6 (Fig 3F).

We analyzed pre-existing RNAseq data from whole animals to assess whether the changes we observed in neuronal rab-7 transcription during adulthood extend to the whole animal [62]. The whole-animal rab-7 mRNA level decreases from Day 3 to Day 6, consistent with the decrease observed with our neuron-specific transcriptional reporter (Fig 3G). On the other hand, there is no increase in the rab-7 mRNA level in early adulthood, suggesting that the expansion of lysosomal capacity during early adulthood may be neuron-specific. We next examined the mRNA levels of five additional genes, cpr-1, cpr-5, vha-5, lmp-2, and syx-17, that are well-validated transcriptional targets of HLH-30 [65]. mRNA levels of cpr-5, lmp-2, and syx-17 show a statistically significant decrease during aging. Therefore, 4 out of 6 HLH-30 transcriptional targets show reduced expression in aging, by bootstrapping analysis, only 3.5% of randomly selected groups of 6 genes have at least 4 that decline in aging (S6A–S6F Fig). Moreover, the level of hlh-30 mRNA itself also decreases during aging (Fig 3H). HLH-30 is predicted to regulate its own expression based on ChIP-seq data [66], echoing the self-transcriptional regulation of mammalian TFEB [67]. Together, these results suggest that HLH-30 transcriptional activity decreases with age, likely contributing to the age-associated neuronal lysosomal dysfunction.

Loss of HLH-30 accelerates age-related decline in dendrite morphology maintenance

In WT worms, the PVD neuron displays a highly branched and organized dendritic arbor. The structure is developed by Day 0, and the PVD neuron can fully maintain organization in middle-aged adult, Day 6 worms (Fig 4A). However, in Day 6 hlh-30 mutants, we observed disorganized dendrite outgrowths, what we refer to as the “sprouting” phenotype, and dendrite degeneration, in approximately 75% and 50% of the population, respectively (Fig 4A, 4D, and 4E). We did not observe any difference in degeneration between WT and the hlh-30 mutant at Day 1 or 3 (Fig 4B). The sprouting phenotype is noticeably increased in the hlh-30 mutant compared to WT by Day 3 of adulthood (Fig 4C). Previous works have shown that the PVD dendrite exhibits progressive degeneration in WT as they age [68,69]. We likewise observed progressive dendrite degeneration with age in WT animals, and the hlh-30 mutant at Day 6 exhibits a similar level of dendrite degeneration as WT at Day 11 (Fig 4D). We also observed that the sprouting phenotype increases with age in WT animals, and the percentage of animals showing sprouting at Day 11 in WT is similar to that in the hlh-30 mutant at Day 6 (Fig 4E). These results suggest that loss of HLH-30 causes early-onset and accelerated neuronal aging in PVD neurons. Expression of wild-type hlh-30 from a pan-somatic promoter was able to rescue the level of both sprouting and dendrite degeneration in the hlh-30 mutant to those observed in WT at Day 6, confirming that HLH-30 is necessary for maintaining dendrite morphology in aging (Fig 4F and 4G).

In addition to regulating lysosomal genes, both TFEB and HLH-30 regulate genes that are not involved in the autophagy or lysosomal pathway [65,70–72]. Given our results showing that HLH-30 promotes lysosomal functionality in C. elegans neurons (Figs 1, 2, S4, and S5), we hypothesized that both the dendrite sprouting and degeneration phenotypes result from decline of lysosomal function. To test this, we examined the PVD dendrite morphology in WT worms treated with V-ATPase inhibitor concanamycin A, which causes reduced acidity of lysosomal compartments (Fig 1B and 1C). Indeed, WT worms treated with concanamycin A show an increased percentage and severity of both sprouting and degeneration compared with controls (Fig 4H and 4I). CUP-5 is an ortholog of mammalian lysosomal channel protein MLN1/TRPML1; complete loss of CUP-5 is lethal, and partial loss of cup-5 function results in reduced lysosomal functions [73,74]. We also assessed the PVD dendrite morphology in cup-5(lf) mutant animals (partial loss-of-function). Indeed, the cup-5(lf) mutant phenocopies the hlh-30 mutant by showing the sprouting and dendrite degeneration phenotypes at Day 6 (Fig 4H and 4I). These results indicate that both dendrite sprouting and degeneration are caused, at least in part, by decline of lysosomal function.

To determine the spatial requirement for HLH-30 to preserve dendrite morphology during aging, we used the Auxin-Inducible Degradation (AID) system, wherein we inserted the AID sequence into the hlh-30 genomic locus to tag the C-terminus of HLH-30 (S1F Fig). When the hormone auxin is added, the tissue-specifically expressed TIR1 E3 ubiquitin ligase recognizes the AID-tagged HLH-30 and promotes its degradation [55]. This technique allows us to deplete HLH-30 in specific tissues (S3 Fig). Auxin delays worm growth, and the progression of age-related dendrite phenotypes irrespective of the presence of TIR1, so we performed all comparisons for this experiment in Day 9 adults grown on auxin (Fig 4J and 4K). We used pan-somatically expressed TIR1 to eliminate HLH-30 in the whole worm except the germline, which phenocopies the dendrite degeneration and sprouting phenotypes observed in the hlh-30 mutant (Figs 4J, 4K, and S3A). Next, we depleted HLH-30 from four individual somatic tissues: neurons, hypodermis, muscle, and intestine (S3A–S3C Fig). Neuron, hypodermis and muscle-specific degradations all show a significant increase in dendrite degeneration compared to WT, and the severity is similar to that of the hlh-30 mutant (Fig 4J). This suggests that HLH-30 is required in neurons, as well as in the two tissues that physically contact the PVD dendrite, to protect against dendrite degeneration. For the dendrite sprouting phenotype, pan-neuronal HLH-30 depletion shows an intermediate sprouting phenotype between hlh-30 mutant and WT (Fig 4K). Muscle-specific HLH-30 degradation also results in a trend of higher percentage of the population exhibiting sprouting compared to WT, though it is not statistically significant (Fig 4K). These data indicate that HLH-30 is necessary in neurons and perhaps also in muscle to maintain organized PVD dendrites during aging.

Since HLH-30 functions cell-autonomously to expand lysosomal compartment abundance (Fig 2A–2C), we asked whether HLH-30 in the PVD neuron is likewise sufficient to protect against age-associated dendrite morphology defects. Indeed, overexpressing HLH-30 either specifically in PVD neurons or pan-neuronally can partially rescue the sprouting and dendrite degeneration phenotypes in the Day 6 hlh-30 mutant (Fig 4F and 4G). These results are consistent with the results from the AID experiment (Fig 4J and 4K), and they indicate that HLH-30 is working in both cell-autonomous and non-cell-autonomous pathways to preserve adult dendrite morphology. The hlh-30 mutant is reported to have a wild-type life span at 20 °C and a slightly shortened life span at 25 °C (S2B Fig) [19,75]. Importantly, the shortened life span at 25 °C cannot be rescued with neuron-specific HLH-30 expression [75]. Therefore, the rescue of the hlh-30 mutant’s dendrite degeneration and sprouting by PVD-specific or pan-neuronal hlh-30 expression is not an indirect effect of life span differences between the strains.

Although PVD-specific overexpression of hlh-30 in WT worms has no statistically significant impact on dendrite degeneration, it reduces the severity of sprouting at Day 11 (Fig 4L and 4M). These data indicate that HLH-30 is sufficient to partially protect against age-associated dendrite morphology defects, supporting a model in which the basal activity of HLH-30 during adult homeostasis is essential for preserving lysosomal function and thereby maintaining neuron integrity during aging (S7 Fig).

C. elegans strains and maintenance

For maintenance, C. elegans strains were grown on nematode growth medium (NGM) seeded with E. coli OP50 at room temperature (22 °C) [105]. Aged worms are defined by the number of days after the L4 stage, “Day 0” of adulthood. All experiments were performed at room temperature (22 °C), except for those in Figs 1E–1H, 2I, S2A, S4, and S5, which were performed at 20 °C. Wild-type N2 and JIN1375 hlh-30(tm1978) IV strains were obtained from the Caenorhabditis Genetics Center (CGC), which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). Strains used in this study are listed in S1 Table.

Cloning and strain generation

Cloning was carried out in the pSM vector, a derivative of pPD49.26, or pPD117.01 (Addgene, Watertown, MA, USA) using standard restriction enzyme cloning technique (NEB, Ipswich, MA, USA). Podr-1::gfp, Podr-1::rfp, Pdes-2::bfp, and Punc-122::rfp vectors are kindly provided by the Kang Shen lab, Stanford University. nuc-1, hlh-30, and rab-7 cDNA were amplified from a cDNA prepared from total RNA isolated from N2 worms. Sequence of the cloning and genotyping primers are listed in S2 Table. Transgenes expressed from extrachromosomal arrays or integrated arrays were generated by microinjection of the constructs [106]. Genome editing was carried out by CRISPR-Cas9 [107]. The hlh-30 syb allele was generated at SunyBiotech (Fuzhou, Fujian, China) by CRISPR-Cas9.

Confocal imaging and fluorescence microscopy

Worms were transferred to a slide with a 3% agarose pad made by melting agarose in M9 buffer. Worms were immobilized in 10 mM sodium azide in M9 buffer for experiments to examine HLH-30::GFP localization and 10 mM levamisole in M9 buffer for all other experiments. Images were captured using a Nikon eclipse Ti2 microscope paired with a CSU W1 SoRa confocal scanner unit and a YOKOGAWA spinning disc unit with a Plan Apo VC 60xA/1.20 WI objective. Data for Figs 2D–2G, S2D, and S2E were collected using the 4× magnifier for Super-Resolution by Optical Re-Assignment. Image settings (laser power and exposure time) were identical for all genotypes and treatment groups across experiments. Images were taken with 0.5 μm step size (most images) or 0.4 μm step size (images with SNG-1::ARGO) in z-stack for 10–25 μm range. Images were analyzed using ImageJ software. Fluorescence intensities and sizes of vesicles were quantified from images using “3D object counter” (Figs 1B–1D, 1F, 2F–2H, 3F, and S3C). Mean fluorescence intensities were used for all figures and calculations except for rab-7 transcription level (Fig 3F), which used maximum fluorescence intensities. Number of vesicles were quantified from images using “3D object counter” (Fig 2E), by hand (Figs 2B, 2C, and S2C) or using “find maxima” (S4C Fig). HLH-30 localizations were quantified from images by eye. PVD dendrite morphology were quantified by eye using the Nikon eclipse compound microscope.

For experiments using SNG-1::ARGO, maximum projection images were generated from z-stack of the whole presynaptic region of each animal. Presynapses were identified by generating a mask from the RFP image with default threshold settings followed by watershed. This mask was used to calculate the mean fluorescence intensity of each synapse in both the RFP and GFP images. The GFP/RFP was calculated for each presynapse detected, then averaged to generate one presynaptic GFP/RFP value for each animal.

Images of HLH-30 localization were taken in the green channel (Image setting: 100 ms exposure time, 60% laser power). The look-up table (LUT) of green channel was adjusted to 30−800 in ImageJ before quantifying. Categories of HLH-30 localization were performed as follow: neurons with HLH-30::GFP enriched in the cytoplasm and has a clear boundary between nucleus and cytoplasm were counted as enriched in the cytoplasm (enrich in C); neurons with HLH-30::GFP intensity in the cytoplasm that is higher than in the nucleus, but no clear boundary between nucleus and cytoplasm were counted as cytoplasm higher than nucleus (N < C); neurons with no difference in HLH-30::GFP intensity between nucleus and cytoplasm were counted as nucleus equal to cytoplasm (N = C).

Quantification of dendrite degeneration and sprouting was performed blind to genotype/condition.

To quantify dendrite degeneration, one dendrite from each worm was assigned to a category as follows: worms with dendrite that shows several swellings and the dendrite between swellings are relatively thinner than normal dendrites were counted as “slight dendrite degeneration;” worms with dendrite that have swellings that are disconnected from the main part of the dendrite were counted as “medium dendrite degeneration;” worms with dendrite that have more than five branches that are disconnected from the main part of the dendrite were counted as “severe dendrite generation;” worms without any feature mentioned above were counted as “no dendrite generation.” Worms were always categorized by the most severe feature found in the whole dendrite.

To quantify dendrite sprouting, one dendrite from each worm was assigned to a category as follows: worms with disorganized dendrite outgrowth only at quaternary dendrites were counted as “slightly sprouting;” worms with disorganized dendrite outgrowth not only at quaternary dendrites, but restricted to one or two small areas were counted as “medium sprouting;” worms with disorganized dendrite outgrowth not only at quaternary dendrites among the whole dendrite were counted as “severe sprouting;” worms without any feature mentioned above were counted as “no sprouting.”

Concanamycin A treatment

NGM plates were supplemented to a final concentration of 0.03 μM concanamycin A (Chemcruz) from a 0.5 mM stock solution. Worms were moved onto seeded concanamycin A plates at the L4 stage/Day 0 of adulthood and were counted or imaged at Day 6 of adulthood. Of note, in performing experiments with concanamycin A-treated animals, we did not observe any detectable change in animal behavior caused by the drug. If the concanamycin A were having a strong impact on the neurons, we would expect it to impede the loading of neurotransmitter into synaptic vesicles, thereby causing aberrant, or uncoordinated movement. This is evidence that the amount of concanamycin A that the neurons experience is quite low.

For the concanamycin A injection, 25 μM concanamycin A in 5% DMSO, or 5% DMSO “mock drug,” was injected into the pseudocoelom of C. elegans, together with 100 g/L orange G as a color tracer. Worms were allowed to recover for 3 hours before they were imaged as described above.

Auxin treatment

NGM plates were supplemented to a final concentration of 1 mM K-NAA (PhytoTech Labs) from a 250 mM stock solution. Worms were synchronized by egg laying 15 gravid hermaphrodites for 2 h on seeded Auxin plates, and PVD dendrite morphology were counted at Day 9 unless otherwise noted.

Heat shock treatment

For measuring HLH-30 localization under heat stress, worms growing at room temperature were shifted to 35 °C for 3 h at Day 3 of adulthood and were imaged immediately afterward (Fig 3A and 3B) [21]. For protein turnover experiments using SNG-1::ARGO, worms growing at 20 °C were shifted to 34 °C for 1 h at Day 2 of adulthood, returned to 20 °C, and then imaged 0–5 days afterward (S4 and S5 Figs) [56].

Starvation treatment

Well-fed worms were moved to unseeded NGM plates at Day 2 of adulthood and were imaged at Day 3 of adulthood (Fig 3A and 3B).

RNAseq data analysis

RNAseq data were obtained from NCBI’s Gene Expression Omnibus. Sample “SRR19895471,” “SRR19895472,” “SRR19895473,” “SRR19895474,” “SRR19895475,” “SRR19895476,” “SRR19895477,” “SRR19895478,” and “SRR19895479” from project “PRJNA853940” were downloaded from NCBI [62]. Fastq reads were first trimmed and filtered for quality using TrimGalore (0.6.10). Reference genome were build using hisat2-build based on C. elegans Bristol N2 DNA reference. Each sample was aligned to the reference using hisat2 (2.1.0) and the results were organized using samtools sort (1.10). The raw gene hits were quantified using featureCounts (1.6.4). Raw gene hit count tables were normalized to total number of transcripts.

Statistical analysis

Statistical analyses were performed in either Microsoft Excel, GraphPad Prism 10, or Rstudio. For quantitative data, single parametric pairwise comparisons were analyzed using unpaired two-tailed Student t test, and non-parametric pairwise comparisons were analyzed using Kolmogorov–Smirnov test. Multiple parametric comparisons were analyzed using one-way analysis of variance (ANOVA) test with Tukey post-test or two-way ANOVA with Sidak post-test as indicated in the figure captions. Data for Figs 1F, 2E–2G, S4E, and S4H were analyzed in R studio using a linear mixed-effects model as indicated in figure captions. For qualitative data, Fisher’s exact test was performed between each individual comparison, and Bonferroni or Holm correction were used to determine the significance threshold for the P value. To compare SNG-1::ARGO half-lives between WT and hlh-30(tm1978), Prism 10 was used to generate one-phase exponential decay curves, constraining the plateau to the experimentally determined value based on background GFP fluorescence (which comes from autofluorescence from the worm cuticle). The curves were fitted using least squares regression with no weighting, and the comparison between WT versus hlh-30(tm1978) was performed using Extra sum-of-squares F test.

Bootstrapping

Two hundred groups of 6 genes per group were randomly chosen from the RNAseq raw gene hits list. Raw gene hit count tables were normalized to total number of transcripts.