Cathepsin B promotes Aβ proteotoxicity by modulating aging regulating mechanisms

Abstract

While the activities of certain proteases promote proteostasis and prevent neurodegeneration-associated phenotypes, the protease cathepsin B (CTSB) enhances proteotoxicity in Alzheimer’s disease (AD) model mice, and its levels are elevated in brains of AD patients. How CTSB exacerbates the toxicity of the AD-causing Amyloid β (Aβ) peptide is controversial. Using an activity-based probe, aging-altering interventions and the nematode C. elegans, we discovered that the CTSB CPR-6 promotes Aβ proteotoxicity but mitigates the toxicity of polyQ stretches. While the knockdown of cpr-6 does not affect lifespan, it alleviates Aβ toxicity by reducing the expression of swsn-3 and elevating the level of the protein SMK-1, both involved in the regulation of aging. These observations unveil a mechanism by which CTSB aggravates Aβ–mediated toxicity, indicate that it plays opposing roles in the face of distinct proteotoxic insults and highlight the importance of tailoring specific remedies for distinct neurodegenerative disorders.

Introduction

Maintaining an accurate protein milieu is vital for cellular and organismal functionality and viability. To maintain protein homeostasis (proteostasis), a nexus of mechanisms acts in concert to preserve the integrity of proteins throughout their lifecycle. Among other functions, this proteostasis network (PN)¹ assists newly synthesized polypeptides in attaining their native conformations, ensures correct protein-protein interactions and directs terminally misfolded and damaged proteins for degradation². Nevertheless, environmental insults, inherited or spontaneous mutations and aging, gradually weaken the competence of the PN, thereby enabling aggregation-prone proteins to escape the cellular quality control mechanisms, and in some cases, form hazardous aggregates. The accumulation of such toxic protein aggregates (proteotoxicity) destabilizes the proteome³ and jeopardizes cellular viability. Accordingly, various human diseases emanate from proteostasis collapse and are characterized by the accumulation of protein aggregates⁴. These maladies, which are collectively termed “proteinopathies”⁵, include the subgroup of neurodegenerative diseases, such as Alzheimer’s disease (AD) and Huntington’s disease (HD)⁶.

The development of AD stems from the accumulation and aggregation of the amyloid β (Aβ) family of peptides. Dual digestion of the amyloid precursor protein (APP) by two proteases, the β-secretase (BACE) and the γ-secretases complex, produces Aβ peptides. The formation and accumulation of Aβ oligomers, rather than large fibrils^7,8, underlie synaptic dysfunction, neuronal loss and eventually lead to the manifestation of neurodegeneration and dementia⁹. It is important to note that AD has an additional prominent pathophysiological feature, the presence of Tau tangles in the brain. Tau is a microtubule‐associated protein that under normal conditions regulates the assembly and maintenance of microtubules. However, in the brains of AD patients, Tau becomes hyperphosphorylated, thereby leading to the disassembly of microtubules. Tau molecules that are released as a result of this process, aggregate and contribute to the pathological manifestations of the disease. Tau hyperphosphorylation and aggregation have been also observed in other neurodegenerative disorders¹⁰. Polyglutamine-expansion diseases manifest in individuals who carry abnormally long stretches of CAG repeats in disease-associated genes. These CAG stretches, which are translated to poly-glutamine (polyQ) expansions, render the proteins that harbor them aggregation-prone, and underpin the development of neurodegeneration¹¹. The group of polyQ-associated disorders includes Huntington’s disease (HD), caused by the aggregation of the protein huntingtin¹², and Machado–Joseph disease (MJD) (or spinocerebellar ataxia type 3), which results from the aggregation of Ataxin-3¹³.

Neurodegenerative disorders also share the feature of late onset¹⁴. The similar temporal emergence patterns of distinct neurodegenerative disorders raise the prospect that aging-associated decline in the efficiency of the PN enables protein aggregation to become toxic in late stages of life. Accordingly, the alteration of aging by the inhibition of the highly conserved insulin/IGF signaling cascade (IIS) protects model nematodes from Aβ-mediated toxicity^15,16 and from the aggregation of polyQ stretches^17,18. This protection is dependent on the activity of a subset of highly conserved transcription factors, including DAF-16/FOXO and SKN-1/NRF^16,19. Consistently, reduced IGF1 signaling protects model mice from AD-like disease^7,20 indicating that this protective effect is conserved in mammals. Dietary restriction, another aging-modulating metabolic pathway²¹, also mitigates the proteotoxicity conferred by both, Aβ and polyQ²².

The clearance of aggregated proteins is a pivotal function of the PN. Two key protein degradation mechanisms, the ubiquitin proteasome system (UPS) and autophagy, are branches of the PN that are critical for proteostasis maintenance^23,24. Thus, the finding that certain proteases mediate the degradation of Aβ and mitigate its toxicity²⁵ is in line with the theme that autophagy promotes proteostasis. Nevertheless, the roles of the cathepsin B (CTSB) family, a large class of lysosomal cysteine proteases, in Aβ-mediated toxicity are uncertain. On one hand, CTSB contributes to the clearance of Aβ²⁶, while on the other its knockout alleviates behavioral deficits and neuropathology of Alzheimer’s model animals²⁷. What roles are played by CTSBs as mediators of proteostasis, and whether these proteases are mechanistically linked with aging-regulating mechanisms are key unanswered questions.

Here we utilized the nematode Caenorhabditis elegans, a preferred model organism for the study of aging and proteostasis²⁸, and GB123, an activity-based chemical probe, which was designed to measure the activity of cathepsin B, L and S of mammals²⁹, to investigate the enigmatic links between cathepsin B, aging and proteostasis. GB123 is comprised of a peptide sequence that confers specificity and mediates binding to the active target, the cathepsin cysteine protease. This peptide sequence (Linker) is attached to a reactive ‘warhead’ that covalently binds to the target using an enzyme-catalyzed chemical reaction. An additional key element of the probe is a Cy5 reporter tag that allows the covalently bound cathepsin-GB123 complex to be detected by the visualization of the Cy5 fluorophore (free GB123 is not detected on gels as it migrates out due to its small size). We discovered that the activity of the nematode CTSB CPR-6, is the sole protease activity that is detected by the probe and that it is modulated by age and by aging-controlling pathways. Surprisingly, while the knockdown of cpr-6 protects model worms from the toxicity of Aβ, it aggravates the polyQ-associated proteotoxicity, indicating that this protease plays opposing roles in the face of distinct proteotoxic insults. The levels of certain IIS components were found to be modulated by the knockdown of cpr-6 indicating that this protease governs proteostasis, at least partly, through the modulation of this aging-regulating pathway.

Results

GB123 is bound by active CPR-6

Similarly to many other proteases, cathepsins are synthesized as inactive precursors that are activated post-translationally³⁰. Therefore, cathepsin activity, rather than expression levels, should be measured to unravel their physiological roles. To directly measure cathepsin activity, we tested whether the fluorescent activity-based probe GB123 (Fig. 1a and Supplementary Fig. 1a), which targets active cysteine proteases of the cathepsin family²⁹, can detect cathepsin activity in homogenates of C. elegans. To address this, we utilized animals that carry two temperature-sensitive mutations which render them feminized upon exposure to 25^oC during development (strain CF512). This feature enabled us to examine whether GB123 is bound by active cathepsins that originate from adult tissues with no background from developing embryos. Synchronized worm populations were homogenized at either day 1, 5 or 9 of adulthood. Each homogenate was divided into two tubes; one was pre-incubated with a potent cathepsin inhibitor GB111-NH₂, to block cathepsin activity³¹, while the other was incubated with the vehicle (DMSO). Next, the samples were incubated with GB123 and separated by polyacrylamide gel electrophoresis. Two fluorescent bands of ∼28 kDa and 30 kDa were observed where homogenate of day 1 old worms was loaded (Fig. 1b, lane 1) (20 µg total protein was found to be sufficient to produce prominent bands, Supplementary Fig. 1b). Interestingly, changes in intensities and additional GB123 bands appeared when homogenates of 5 or 9-day old worms were loaded (bands 2 and 3 respectively). First, the intensity of the lower band (band 2) was largely increased as the animals aged, indicating a major rise in the activity of the corresponding protease in post-reproductive adulthood. In addition, in older animals two upper bands of ∼32 kDa and 40 kDa (bands 4 and 5 respectively) appeared, suggesting that the detected cathepsin interacts with distinct partners at older ages. All bands were greatly weakened when GB111-NH₂ was incubated with the homogenates prior to the reaction (lanes 4–6), confirming the specificity of the GB123 signal to cathepsin activity.

Fig. 1: Active CPR-6 binds GB123.

a An illustration of the activity-based probe GB123. This compound includes a Cy5 tag, a linker, and a warhead that covalently binds active CTSB. b GB123 is bound by a C. elegans protease. Pre-treatment with the competitive inhibitor GB111-NH₂ abolishes the signal. The pattern of bands this protease produces differs in homogenates of young (day 1) and post-reproductive animals (days 5 and 9 of adulthood). c The inhibition of CTSB by the inhibitor CA-074 but not of CTSL (by CAS 108005-94-3) reduces the binding of GB123 to the active protease (n = 1). d A directed RNAi screen indicates that the knockdown of two CTSBs, cpr-6 and cpr-9, but not of other cysteine proteases, abolishes GB123 signal (n = 3). e Highly specific RNAi constructs that target the 3’UTR of cpr-6 or of cpr-9, indicate that CPR-6 is the protease that binds GB123 (n = 2). f A cpr-6 deletion mutant exhibits remarkably reduced GB123 binding in different ages (lanes 3, 6 and 9). Similarly, reduced signal was observed when wild type worms were treated with cpr-6 RNAi (lanes 2, 5 and 8).

We next sought to identify the specific protease, or proteases, whose activities are detected by GB123. Using homogenates of day 2 old worms, we conducted an activity assay in the presence of either one of two selective inhibitors; CA-074 that inhibits the activity of CTSBs or CAS108005-94-3 which reduces the activity of CTSLs. The inhibition of CTSB, but not of CTSL, largely weakened the GB123-specific bands (Fig. 1c) suggesting that GB123 binds CTSB. To further scrutinize this notion and to identify the specific gene product that binds GB123, we conducted a directed RNA interference (RNAi)-based screen. CF512 worms were left untreated (fed with bacteria that harbor the empty RNAi vector (EV)), or treated from hatching with RNAi towards one of the different mammalian orthologs of cysteine protease-type cathepsins found in C. elegans genome. The worms were homogenized at day 2 of adulthood and a cathepsin activity assay was performed. Our results indicated that only the knockdown of two CTSB-encoding genes, cpr-6 (C25B8.3) and cpr-9 (F57F5.1) largely reduced the GB123 signal (Fig. 1d). These results confirm that the signal results from the binding of GB123 with an active CTSB, but with neither CTSL nor cathepsin Z or cathepsin O.

Two models may explain the abolishment of GB123 signal by two RNAi species. One suggests that the probe is bound by two different active CTSB proteases, while the other proposes that the two RNAi species knockdown the expression of the sole CTSB which binds GB123. To distinguish between these possibilities, we conducted a quantitative real-time PCR (qPCR) experiment using primer sets that specifically recognize either cpr-6 or cpr-9. We found that the two RNAi strains of the Vidal library exhibit cross-silencing effects, wherein RNAi towards either cpr-9 or cpr-6 efficiently knockdown the expression of both genes (Supplementary Fig. 1c). Thus, to specifically knockdown the expression of each of these genes, we created RNAi species that target the highly variable 3’ untranslated regions (3’UTR) of cpr-6 or cpr-9. A qPCR experiment confirmed the specificity of these new RNAi species towards either cpr-6 or cpr-9 (Supplementary Fig. 1c). A GB123-based activity assay showed that, while the knockdown of cpr-6 by the highly specific 3’UTR RNAi, largely reduced the fluorescent signal, treating the worms with cpr-9 3’UTR RNAi showed no such effect (Fig. 1e).

To further test whether CPR-6 is the CTSB that reacts with GB123, we utilized worms that harbor a 157 base pairs deletion which removes part of the last intron and the last exon of cpr-6 (strain tm12210) and thus, are expected to partially lose CPR-6 activity. Untreated wild-type nematodes (strain N2) and tm12210 worms, as well as cpr-6 3’UTR RNAi-treated wild-type animals, were harvested at three ages: pre-fertile adulthood, day 1 and day 5 of adulthood. The worms were homogenized and subjected to a GB123 activity assay. In all age groups, cpr-6 RNAi-treated worms and cpr-6 mutants exhibited a remarkable reduction in the GB123 signal (Fig. 1f), supporting the theme that GB123 is bound by CPR-6.

Finally, we sought to directly identify the CTSB that binds GB123. A homogenate of day 5 old CF512 worms was incubated with GB123 and an anti-Cy5 antibody (Cy5 is a component of GB123, Supplementary Fig. 1a) was used to pulldown the proteins that are bound by the probe. Mass spectrometry (MS) analysis (complete dataset is available at GEO: GSE232889) indicated that the sole cathepsin that was bound by GB123 was CPR-6 (Table 1) supporting our previous results. Intriguingly, endoplasmic reticulum (ER)-resident proteins such as the chaperone HSP-4, as well mitochondrial proteins also sediment with GB123. It is tempting to speculate that CPR-6 interacts with these organelles, possibly through mitochondria-associated membranes (MAMs)³², membrane structures that were reported to play roles in the pathophysiology of AD³³, however, this theme requires further elucidation.

Table 1 A partial list of the most abundant proteins that co-sediment with GB123

The knockdown of cpr-6 differentially modulates proteostasis

Having identified CPR-6 as the CTSB which is bound by GB123, we sought to test whether reducing the levels of this protease affects Aβ toxicity in model worms. To test this, we employed nematodes that express the AD-causing aggregative peptide, Aβ_3-42³⁴ in their body wall muscles (strain CL2006, hereafter “Aβ worms”). Aβ aggregation results in a progressive paralysis within the worm population³⁵, a phenotype that we followed by the “paralysis assay”²⁸. Aβ worms were either grown on EV bacteria, or treated with cpr-6 RNAi or daf-2 RNAi (a positive control for reduced Aβ toxicity), and rates of paralysis were recorded daily. Our results indicated that the knockdown of cpr-6 by RNAi of the Vidal library (Fig. 2a) or by the highly specific cpr-6 3’UTR RNAi (Fig. 2b) significantly protects the worms from Aβ-mediated proteotoxicity. Similarly, when Aβ worms were crossed with cpr-6^-/- animals (tm12210), which exhibit loss of CPR-6 activity (Fig. 1f), we observed a trend of protection from Aβ (Supplementary Fig. 2a, strain EHC151).

Fig. 2: Opposing effects of cpr-6 on Aβ- and polyQ35-YFP-mediated toxicity.

a, b The knockdown of cpr-6 by RNAi of the Vidal library (a) or the 3’UTR cpr-6 RNAi (b) protects Aβ worms from proteotoxicity as measured by the paralysis assay. n = 3, Bars represent average daily rates of paralysis of the population ± SEM. Statistical test used: unpaired, one-tailed Student’s t-test, *p < 0.05, **p < 0.01, ***p < 0.001. c, d. WB analysis using CL2006 worms and an Aβ antibody shows that cpr-6 RNAi treatment reduces Aβ aggregation in the worm debris (c) in a reproducible manner (d). Bars represent average signal intensity of 4 replicates ± SEM. Statistical test used: unpaired, two-tailed Student’s t-test. e A thrashing assay indicates that cpr-6 RNAi treatment exacerbates polyQ35-YFP proteotoxicity in day 6 old worms but not in day 4 old animals. In contrast, IIS reduction by daf-2 RNAi protects the worms in both ages. Bars represent median thrashing rate within the population ± 95% confidence intervals. Statistical test used: 2-way ANOVA (Tukey’s multiple comparisons test; *p < 0.05, **p < 0.01). f–i The knockdown of cpr-6 results in an elevated polyQ35-YFP aggregation level as measured by native agarose gels (f), increased number of polyQ35-YFP containing foci, seen by fluorescent microscopy (g, scale bar 60μm), and measured by foci quantification (h) as well as by comparison of total fluorescence levels in untreated and cpr-6 3’UTR RNAi-treated animals (i). Bars represent average number of foci and average YFP fluorescence levels ± SEM. Statistical test used: unpaired, two-tailed Student’s t-test. j cpr-6 3’UTR RNAi does not affect lifespan of wild-type worms (strain N2, n = 3).

We next wondered how the knockdown of cpr-6 protects from Aβ-mediated proteotoxicity. First, we treated Aβ worms with 3’UTR cpr-6 RNAi or left them untreated (EV) and compared the rates of Aβ aggregation by Western blot analysis. We found that the knockdown of cpr-6 remarkably reduces Aβ aggregation (Fig. 2c, d). Since the proteostasis network can differentially respond to distinct proteotoxic insults³⁶, we next examined whether the knockdown of cpr-6 also protects worms from polyQ-mediated toxicity. Using worms that express polyQ35-YFP in their body wall muscles (strain AM140) and the thrashing assay²⁸, we found that knocking down cpr-6 aggravates the toxicity of polyQ35-YFP in 6 days old worms, but not of their younger counterparts. This enhancement of proteotoxicity was contradictive to the effect of daf-2 RNAi treatment which ameliorated the toxicity of polyQ35-YFP in both ages, day 4 and 6 of adulthood (Fig. 2e). This observation implies that cpr-6 plays opposing roles in the face of proteotoxicity that stems from Aβ and polyQ.

Utilizing native agarose gel electrophoresis (NAGE³⁷) we observed increased rate of polyQ35-YFP aggregation upon the knockdown of cpr-6 in day 5 old animals (Fig. 2f and Supplementary Fig. 2b). cpr-6 RNAi also significantly increases the number of polyQ35-YFP-containing foci in the worms (Fig. 2g, h) as well as the total YFP fluorescence levels (Fig. 2i).

Interestingly, the knockdown of cpr-6 did not affect the lifespan of wild-type worms (Fig. 2j, n = 3 and Supplementary Fig. 2c), further supporting the notion that lifespan and proteostasis are not necessarily coupled^15,38,39.

Age-associated differences of CPR-6 activity and distribution by protein modulators

The well-established roles of the IIS⁶ and DR²² in proteostasis regulation and the involvement of CTSB in the development of AD-like disease in model animals²⁷, have prompted us to examine whether CPR-6 expression and activity are modulated by aging. First, we characterized the changes in CPR-6 activity in early and late stages of life. CF512 animals were cultured on EV bacteria and harvested at either one of the following ages: larval stages L1, L2 or, L4; young, pre-fertile adults; or at day 1, 5, or 9 of adulthood. We also harvested eggs of CF512 worms, right after extraction from fertile worms. The worms and eggs were homogenized and subjected to GB123 activity assay. Band 1 was much more prominent in homogenates of developing larvae (larval stages L1-L4) and of eggs, compared to homogenates of adult worms. In contrast, band 2 was weak in larvae but largely intensified in adult worms and during embryogenesis, whereas band 4 and 5 were predominantly detectable in post reproductive adult worms and in embryos. Surprisingly, the pattern of active CPR-6 bands produced by the egg extract was similar to that seen in elder animals (Fig. 3a and Supplementary Fig. 3a).

Fig. 3: CPR-6 activity is modified with aging by a temperature-sensitive modulator.

a An analysis of the age-dependent changes in CPR-6 activity patterns from embryonic development through larval development and adulthood (n = 3). b Mixing homogenates of L4 larvae (lane 3) and day 5 old worms (lane 4) shows changes in the intensities of the different bands. Bands 2 and 3 are largely intensified whereas bands 4 and 5 are greatly weakened (lane 5). These effects are abolished when the L4 homogenate was heat inactivated (lane 6) or when the homogenates were not incubated prior to the analysis (lane 7) (n = 3). c Different mixing ratios indicate that 25% of L4 homogenate is sufficient to greatly reduce the intensities of bands 4 and 5 (lane 7) (n = 2). d Protein identification by mass spectrometry indicates that CPR-6 is the sole protein that appears in all five bands. e–h Analysis of proteins that sediment with each CPR-6-containg band shows an enrichment of cytosolic and mitochondrial proteins in all bands. Interestingly, lysosomal proteins were less abundant.

The age-associated differences in CPR-6 activity among larvae and adult worms (Fig. 3a), raised the prospect that an aging-associated modulator changes CPR-6 activity during the worms’ lifecycle. We tested this hypothesis by mixing homogenates of L4 larvae and day 5 adult worms and asking whether they functionally interact. As expected, pure homogenates of L4 larvae (Fig. 3b, lane 1 and 3) exhibited low CPR-6 activity level that was apparent by a faint band 1 and relatively weak bands 2 and 3. In contrast, homogenate of day 5 adult worms showed much higher levels of activity wherein bands 2, 3, 4 and 5 exhibited elevated intensities (lane 2 and 4). This elevated activity was observed despite a reduced cpr-6 expression level in day 5 of adulthood as measured by qPCR using CF512 worms (Supplementary Fig. 3b) and CL2006 animals (Supplementary Fig. 3c). When equal protein amounts of the two samples were mixed and incubated (30 min at 25 °C), the intensity of band 1 did not notably change (lane 5), but band 3 was largely intensified and the intensities of bands 4 and 5 were greatly reduced. These phenomena were abolished when the L4 homogenate was heat inactivated (95 °C for 10 min) prior to mixing (lane 6), and when the samples were mixed immediately prior to the CPR-6 assay (lane 7). These results indicate that the molecules, which modulate CPR-6 activity with age, are heat-labile, most likely proteins. Mixing homogenates of L4 larvae and day 5 old worms in different ratios (Fig. 3c), we found that 25% of L4 homogenate is sufficient to reduce the intensities of bands 4 and 5, albeit not to the same extent as the addition of 50% of this homogenate.

To verify that the different bands represent sub-populations of active CPR-6 we prepared homogenates of untreated (EV) L4 larvae and of day 5 adult CF512 animals. The homogenates were supplemented with GB123, pulled down using a Cy5 antibody, and the bands were separated on a gel and sliced individually. Proteins that sediment with each of the five bands were identified by mass spectrometry (Supplementary Table 1). We identified total of 305 proteins in all samples. Importantly, CPR-6 was the sole protein that was identified in all five bands (Fig. 3d). The identification of CPR-9 in four of the bands (Supplementary Table 1) suggests that CPR-9 may physically interact with CPR-6 and that protein identification in this experiment was more sensitive than in the experiment that is displayed in Table 1. It is important to note that all cathepsins that we identified in this experiment were much less abundant in the samples than CPR-6 (Supplementary Table 2). Interestingly, cytosolic and mitochondrial proteins were more abundant than lysosomal proteins in bands 1, 2, 3 and 4. For instance, among the 253 proteins that sediment with band 1 and 2, we found 58 mitochondrial proteins (as analyzed by the DAVID bio-informatic source), including the MAM-resident proteins VDAC-1 and the enzyme fumarase (FUM-1), and only 9 lysosomal proteins (Supplementary Table 2). In addition, we identified in these samples, the proteostasis-promoting molecular chaperones Hsp-110 and the Hsp-70 family member, Hsp-1. The abundance of cytosolic and mitochondrial proteins in bands 1, 2, 3 and 4 were also demonstrated by the clustering of the sediment proteins according to their cellular localizations (Fig. 3e–h, respectively). Using the STRING tool (https://string-db.org/) to predict protein-protein interaction networks we found that while proteins that were sedimented with all bands included components of the mitochondria and ribosome (Supplementary Fig. 3d–g), a network of lysosomal proteins was only detected in band 1 (Supplementary Fig. 3d). It is noteworthy that the proteasomal protein network was predicted in bands 2, 3 and 4 (Supplementary Fig. 3e-g). These results further suggest that ER, mitochondria, proteasomes and lysosomes may interact within the cell and that CPR-6 may be present in these contact sites. However, further research is needed to scrutinize these predicted interactions.

The knockdown of cpr-6 modulates gene expression and protein stability to promote proteostasis

The aging-associated changes in CPR-6 activity (Fig. 3a) have led us to ask whether aging-regulating pathways control the expression and activity levels of this protease. To scrutinize this hypothesis, we tested whether IIS reduction affects the pattern of CPR-6 activity. CF512 worms were treated with daf-2 RNAi or left untreated, harvested at day 2 of adulthood and subjected to GB123-based, CPR-6 activity assay. daf-2 RNAi treated worms exhibited a pattern of CPR-6 activity which is similar to that of older untreated animals, namely intensification of bands 4 and 5 (Fig. 4a and Supplementary Fig. 4a). Importantly, testing how the knockdown of daf-2 affects the CPR-6 activity signature in different ages unveiled that while day-2-old worms exhibit a pattern which is more similar to that of older untreated worms, in later stages of life daf-2 RNAi treatment appears to slow the aging-driven modulation of CPR-6 activity signature (Supplementary Fig. 4b). We also employed worms that carry a weak daf-2 allele (strain CB1370) and tested whether the knockdown of one of the IIS-regulated transcription factors, skn-1, daf-16 or hsf-1 changes the CPR-6 activity pattern in these animals. The worms were homogenized and subjected to a GB123 assay side by side with wild-type worms of the same ages. Our results (Supplementary Fig. 4c) show no difference in the CPR-6 signature at day 1 of adulthood. However, at day 9 of adulthood CB1370 worms exhibited an enhanced intensity of band 5 (arrowhead) compared to wild-type animals. This phenomenon was reduced upon the knockdown of daf-16 suggesting that this transcription factor may be involved in the regulation of CPR-6 promoted activity.

Fig. 4: CPR-6 activity and aging are interrelated.

a, b The alteration of aging by knocking down daf-2 (a) or by bacterial deprivation (b) changes CPR-6 activity patterns in homogenates of day 2 adult worms (n = 3). c skn-1 is needed for the cpr-6 RNAi-mediated protection from Aβ proteotoxicity as the mix of skn-1 RNAi and EV bacteria (red line) and a concurrent knockdown of skn-1 and cpr-6 (dashed red line) similarly enhance the toxicity of Aβ. No such effect was seen when cpr-6 and daf-16 were concomitantly knocked down (dashed green line). d Three independent paralysis assays confirm the requirement of skn-1 for cpr-6 RNAi to counter Aβ proteotoxicity. Bars represent the average daily rates of paralysis in the three replicates ± SEM (n = 3). Statistical test used: unpaired, one-tailed Student’s t-test (*p < 0.05, **p < 0.01, ***p < 0.001). e RNA-seq experiment indicates that the knockdown of cpr-6 results in the elevated expression of 46 genes and reduced expression of 29 genes (log2 ≤ 0.58 or ≥0.58, p-value ≤ 0.05). f A heat map of prominently upregulated and downregulated genes. g The knockdown of swsn-3 by RNAi enhances paralysis of CL2006 worms and a concurrent knockdown of swsn-3 and of cpr-6 prevents cpr-6 RNAi from alleviating Aβ proteotoxicity. Bars represent the average rates of paralysis in the three replicates ± SEM. Statistical test used: unpaired, one-tailed Student’s t-test (*p < 0.05, **p < 0.01, ***p < 0.001). h Cellular localization analysis unveils that most genes whose expression levels are affected by cpr-6 RNAi encode membrane proteins.

We also examined whether bacterial deprivation, a well-established aging-regulating manipulation²¹, modulates the pattern of CPR-6 activity and found that this treatment has also led to the weakening of band 3 (Fig. 4b and Supplementary Fig. 4d). qPCR indicated that IIS reduction by daf-2 RNAi does not affect cpr-6 expression levels in day 1 of adulthood but slightly reduces it on day 5 (Supplementary Fig. 4e). These results imply that CPR-6 activity is foremost regulated by this pathway post-translationally.

The modulation of CPR-6 activity by aging-controlling pathways has led us to speculate that transcription factors downstream of the IIS may be involved in protection from Aβ proteotoxicity conferred by cpr-6 RNAi. To test this hypothesis, Aβ worms were fed from hatching with mixtures of RNAi bacteria to concurrently knockdown the expression of cpr-6, and of either daf-16 or skn-1 and subjected to paralysis assays (50% dilution of cpr-6 RNAi bacteria with EV or with RNAi toward other genes, does not reduce the knockdown efficiency of cpr-6 RNAi (Supplementary Fig. 4f) nor of the RNAi towards the other tested genes (Supplementary Fig. 4g–i)). We found that when skn-1 (Fig. 4c, d), but not daf-16 (Fig. 4c and Supplementary Fig. 4j), is knocked down, reducing cpr-6 expression can no longer protect the worms from Aβ proteotoxicity.

The requirement for SKN-1 for cpr-6 RNAi-conferred counter proteotoxic effect strongly suggests that this treatment modulates gene expression. Thus, we used RNA-sequencing to compare gene expression profiles of day 6 old untreated Aβ worms and their counterparts which were cultured for two generations on cpr-6 RNAi (The NGS raw data is available at GEO: GSE232889). 46 genes showed significantly elevated expression and 29 genes exhibited reduced expression levels (p < 0.05) upon the knockdown of cpr-6 (Fig. 4e). We focused on genes that had at least 20 reads and whose expression levels were increased by at least 50% or decreased to a level which is 67% or less of the expression level seen in untreated animals (log2 > 0.58 or <-0.58, see Supplementary Tables 3 and 4, p-value < 0.05). Among the genes which showed reduced expression (Fig. 4f) we identified swsn-3, a chromatin modulator that interacts with DAF-16 and is needed for the full longevity of daf-2 mutant animals⁴⁰. Of note, the knockdown of swsn-3 by RNAi enhanced paralysis of CL2006 worms and prevented cpr-6 RNAi from further mitigating proteotoxicity (Fig. 4g). Using tm12210 worms that were crossed with CL2006 and treated with swsn-3 RNAi (Supplementary Fig. 5A) we found that swsn-3 is crucial for proteostasis of worms that are challenged by Aβ. Similarly, treating these mutant worms with RNAi toward either skn-1 (Supplementary Fig. 5b) or smk-1 (Supplementary Fig. 5c) confirmed that these genes are required for proteostasis of worms that are challenged by Aβ.

The expression of ubc-1, a ubiquitin conjugating enzyme, which reduces polyQ aggregation and prevents ubiquitin and proteasome localization to aggregates⁴¹, was also reduced upon the knockdown of cpr-6. Among the upregulated genes, we identified spp-17 and f56d6.9, both are regulated by the IIS. These findings strengthen the notion that CPR-6 is involved in the regulation of proteostasis, at least partially, by modulating the levels of IIS-controlled genes. Functional annotation (Fig. 4h) pointed at genes encoding membrane proteins to be prominently affected by cpr-6 RNAi.

To evaluate the possibility that CPR-6 modulates proteotoxicity post-translationally, we compared the proteomic landscapes of Aβ worms that were treated for two generations with either cpr-6 3’UTR RNAi, Aβ RNAi or left untreated. The worms were homogenized at either L4 larval stage or day 5 of adulthood and their proteomes were compared to these of untreated animals (EV) of the same ages, using MS. We focused on proteins that exhibited at least two-fold increase or decrease in at least two of the three independent repeats. 77 proteins exhibited a significant increase and 52 showed decreased levels in cpr-6 RNAi-treated worms in samples of L4 larvae and the levels of 204 were elevated and of 42 were reduced in day 5 adult animals (Fig. 5a, b). Functional clustering indicated that proteins which are involved in splicing and import into nucleus were most abundant among the upregulated (Fig. 5c). Among the downregulated proteins, we also identified proteins that are involved in transport (Fig. 5d).

Fig. 5: cpr-6 RNAi modulates the proteome of larvae and adult worms.

a, b Mass spectrometry indicates that the knockdown of cpr-6 by the 3’UTR RNAi leads to an increase in the levels of 77 proteins and to decreased levels of 52 proteins in L4 larvae. 204 proteins showed increased levels and 42 exhibited reduced amounts in day 5 adult worms upon the knockdown of cpr-6. c, d Clustering of the identified proteins by biological function shows an enrichment of proteins that are involved in splicing, nuclear import and organization in day 5 old worms. e The most prominent proteins that showed at least two-fold elevated (upper table) or decreased (lower table) levels in day 5 old CL2006 worms (p-value was calculated using T-test, 2 tails, unpaired). f A paralysis assay indicates that smk-1 is crucial for the cpr-6 RNAi-mediated protection from Aβ proteotoxicity. g Three independent repeats of the paralysis assay as in (f). Bars represent average daily rates of paralysis in the population ± SEM (n = 3). Statistical test used: unpaired, one-tailed Student’s t-test (*p < 0.05, **p < 0.01).

Sixteen proteins were found to exhibit two-fold elevated levels and two proteins two-fold reduced levels in all three samples of worms that were treated with cpr-6 RNAi (Fig. 5e). Among the proteins that were found to be significantly upregulated in cpr-6 RNAi treated worms, we recognized SMK-1, a regulatory subunit of the PP4 protein phosphatase complex, which is needed for the full longevity and resistance to pathogenic bacteria phenotypes of daf-2 mutant worms⁴². SMK-1 is also needed for DR-mediated longevity⁴³. Thus, we tested whether smk-1 is required for cpr-6 RNAi-mediated protection from Aβ, by treating CL2006 worms with the following mixtures of RNAi bacteria: EV+cpr-6, EV+smk-1, cpr-6+smk-1 or left them untreated (EV) (the knockdown efficiency of smk-1 RNAi is not affected by the mixture with cpr-6 RNAi (Supplementary Fig. 4i)). The worms were subjected to a set of paralysis assays which indicated that smk-1 is critically needed for the promotion of proteostasis by cpr-6 RNAi (Fig. 5f, g). No such effect on paralysis was seen when we knocked down the expression of R05D3.9, the protein that exhibited the most significant elevated levels upon the knockdown of cpr-6 (Supplementary Fig. 5b).

Several additional interesting proteins were identified by our MS experiment. The levels of GPA-7, the orthologue of the human protein GNAO1 that is involved in the developmental and epileptic encephalopathy 17⁴⁴, and of DAO-5 which was reported to be negatively regulated by DAF-2 and to be a lifespan regulator⁴⁵, were also affected by the knockdown of cpr-6. Finally, W08F4.3, the worm orthologue of the human protein SIGMAR1 that is involved in the development of Amyotrophic Lateral Sclerosis (ALS) with dementia⁴⁶ showed low levels in cpr-6 RNAi treated worms.

These findings suggest that cpr-6 modulates proteostasis, at least partially, by changing the levels of IIS-associated proteins.

Discussion

While the over-expression of the metallopeptidase neprilysin or of the insulin-degrading enzyme protects model mice from AD-like disease²⁵, CTSB, a papain-like cysteine protease that is synthesized as a preproenzyme⁴⁷, promotes the development and progression of AD. Hence, the knockout of CTSB improves the cognitive performance of AD model animals²⁷. How CTSB impairs proteostasis and enhances AD-like phonotypes is an unsolved enigma.

Here we employed the nematode C. elegans and the specific activity-based probe GB123, to study the links between CTSB activity, aging and proteostasis. We found that GB123 is bound by the nematode CTSB, CPR-6 (Fig. 1), that CPR-6 activity is modulated by aging (Figs. 1b and 6i) and by aging-regulating pathways, namely IIS reduction and DR (Fig. 4a, b). While the knockdown of cpr-6 protects model worms from Aβ-mediated toxicity (Fig. 2a, b), it aggravates proteotoxicity of polyQ35-YFP (Fig. 2e and 6ii), indicating that CPR-6 activity plays opposing roles in the face of distinct proteotoxic challenges. Protection from Aβ by cpr-6 RNAi is dependent upon the transcription factor SKN-1 and the phosphatase regulatory subunit SMK-1 (Fig. 6iii, iv), both critically needed for the longevity phenotype of daf-2 mutants^42,48 and of dietary restricted animals^43,49. It is important to note that our data correlate SMK-1 and SKN-1 but do not provide sufficient evidence for a mechanistic link between these two regulatory proteins. The observation that the knockdown of cpr-6 does not affect lifespan further strengthens the notion that longevity and proteostasis are separable^15,38,39. Transcriptomic analysis unveiled that the knockdown of cpr-6 modulates the levels of several lifespan-associated genes including the chromatin modulator swsn-3 (Fig. 6v), whose knockdown aggravates Aβ-mediated proteotoxicity and prevents cpr-6 RNAi from protecting the worms. This result is counter-intuitive as reducing the expression of cpr-6 mitigates proteotoxicity and reduces the expression of swsn-3 (Fig. 4f), despite the fact that the knockdown of swsn-3 enhances proteotoxicity (Fig. 4g). This apparent contradiction may involve an activation of SWSN-1, which was shown to be cooperate with DAF-16 as a regulator of lifespan⁴⁰.

Fig. 6: A model.

The activity of CPR-6 (Cathepsin B) is modulated by aging (i). CPR-6 protects worms from the toxicity of polyQ35 stretches (ii) but exacerbates Aβ proteotoxicity. Accordingly, the knockdown of cpr-6 mitigates the toxicity of Aβ. The counter-proteotoxic effect of cpr-6 RNAi is linked with increased levels of SMK-1 protein (iii), which probably cooperates with SKN-1 (iv), and with reduced expression of the swsn-3 gene (v) which encodes for a chromatin modulator. SWSN-3 is predicted to be important for the proteostasis-maintaining activity of SKN-1 (vi) by modulating gene expression (vii). This modulation mitigates Aβ proteotoxicity (viii), possibly by the modification of nuclear import and organization.

These results culminate to substantiate a mechanistic link between CPR-6, proteostasis and aging and raise the prospect that this protease functions as a component of a SKN-1 proteostasis-regulating axis (Fig. 6vii, viii), which is controlled by both: IIS and DR. This theme is in line with the evidence that the activation of NRF2, the mammalian orthologue of SKN-1, can potentially counter neurodegenerative disorders⁵⁰. It is important to note that SMK-1, SKN-1 and SWSN-3 are not expected to physically interact with CPR-6 but to affect its counter-proteotoxic features by functional interactions. However, the relations between CPR-6 and these proteins require further research and clarification.

Surprisingly, we found that the knockdown of daf-2 by RNAi or subjecting the worms to dietary restriction, two treatments that are known to slow aging and to mitigate proteotoxicity⁶, modulate the activity of CPR-6. Upon these treatments, CPR-6 exhibits early-in-life, activity patterns that are similar to those of elder worms (Fig. 4a, b). This may indicate that the alteration of aging activates protective mechanisms that are normally induced at later ages to reduce the accumulation of aging-associated damage. This notion is supported by the report that IIS reduction activates components of the unfolded protein response of the endoplasmic reticulum in the absence of aggregated proteins within the lumen of this organelle⁵¹. Nevertheless, further investigation is required to scrutinize this theme.

Interestingly, mixing homogenates of young and old worms (Fig. 3b, c) implied that these samples functionally interact, as homogenates of L4 larvae changed the typical band pattern of homogenate of day 5 old worms, plausibly by a heat-sensitive protein (Fig. 3b). It will be interesting to identify this unknown protein and test whether its expression or activity are regulated by aging and by aging-governing pathways. Furthermore, manipulating the activity of the human orthologue of this CPR-6 regulator could have therapeutic potential.

We have recently reported that knocking down the expression of tor-1/2, the C. elegans orthologues of human torsinA, protects worms from Aβ proteotoxicity but exacerbates the toxicity of polyQ35-YFP. These opposing effects are mediated by neuropeptide signaling which modulates SKN-1 activity³⁶. The differential effects of cpr-6 RNAi on Aβ and polyQ35-YFP toxicity, the lack of effect on lifespan and the requirement of SKN-1 for cpr-6 RNAi-mediated protection from Aβ, suggest that, similarly to torsins, CPR-6 might function in neurons to activate neuropeptide signaling which governs SKN-1 activity and promote proteostasis in the soma. This theme may be supported by the observation that although cpr-6 is foremost expressed in the intestine, it is also present in a subset of sensory neurons, URX, AQR and PQR⁵², that were shown to be regulators of innate immunity and proteostasis⁵³.

An additional interesting observation shows that although the knockdown of both; cpr-6 and daf-2 by RNAi enhances the aggregation rates of polyQ35-YFP molecules (Fig. 2f), the knockdown of daf-2 mitigates proteotoxicity that stems from this chimeric protein while cpr-6 RNAi treatment aggravates it. This apparent contradiction can be explained by the complex relations of hyper-aggregation and proteotoxicity. Early in life, hyper-aggregation can serve as a protective activity that sequesters highly toxic oligomers from the cellular environment thereby mitigating proteotoxicity. In contrast, in late stages of life, large aggregates can release small toxic conformers that enhance toxicity⁴. This idea is supported by the finding that the yeast chaperone HSP104 confers protective disaggregation when the concentrations of aggregative proteins are low, but induces hyper-aggregation when the challenge of aggregation exceeds a certain threshold⁵⁴. In any case, further research is needed to elucidate the roles of hyper-aggregation in the context of CPR-6 activity.

One of our surprising findings is that CPR-6, a primarily lysosomal protease⁴⁷, sediments with mitochondrial proteins. This observation coincides with the emerging theme that lysosomes physically and functionally interact with mitochondria to mediate a myriad of biological processes including signaling. Impairment of these interactions is typical to AD⁵⁵. Of note, we also identified the endoplasmic reticulum (ER) resident chaperone HSP-4, an orthologue of the human BiP chaperone, to sediment with GB123-bound CPR-6 (Fig. 1g, Table 1). This suggests that lysosomes may interact with Mitochondria-Associated ER Membranes, lipid micro-domains that were implicated in the etiology of AD³³. This theme raises the prospect that lipid micro-domains serve as scaffolds for the interaction and communication of cellular organelles, which regulate proteostasis³². An alternative explanation to the co-precipitation suggests that a subpopulation of CTSB molecules, which functions away from the lysosome, interact with the mitochondria to suppress proteostasis.

While this study supports the idea that CTSB inhibitors can be used as components of future therapeutic cocktails for the treatment of AD, the opposing effects of this protease in the face of distinct proteotoxic proteins, highlights the need to carefully tailor such interventions. It is critical to properly treat different disease-causing protein aggregation, according to the proteotoxic protein that jeopardizes proteostasis and underlies the development of the specific illness.

Methods

Worm strains used in this study are listed at Supplementary Table 5.

Primers used for cloning of RNAi plasmids and genotyping are listed at Supplementary Table 6.

Primers used for qPCR are listed at Supplementary Table 7.

Caenorhabditis elegans and RNA interference

All the experiments using C. elegans were conducted at 20 °C (except when using CF512 strain). Strains used in this paper are provided in Supplementary Table 5. Worms were grown on Nematode Growth Medium (NGM) plates supplemented with 100 µg/ml ampicillin and seeded with E. Coli HT115 bacterial cultures. The worm population was synchronized using sodium hypochlorite (bleach) and potassium hydroxide. To induce RNAi, the seeded plates were supplemented with 100 mM isopropyl β-d-1-thiogalactopyranoside (IPTG; final concentration of 4 mM) to induce the expression of the dsRNA. All RNAi plasmids were created by amplifying the relevant sequences (Supplementary Table 6) using PCR and cloning into the L4440 plasmid using the restriction enzymes Nhl-1 and Xho-1.

Worm homogenization and GB123 labeling

Worms were collected in M9 buffer and washed three times (settled with gravity) to remove small larvae and bacteria. Zirconium oxide beads were added to the worm pellet and these animals were homogenized using a bullet blender. Homogenates were separated into debris and post-debris supernatant by spinning (PDS) @ 13000 G for 10 min. PDS was quantified using BCA kit (PierceTM, #23225).

For each sample, an equal amount of protein was preincubated with either the inhibitor, GB111NH (5 μM), or control, DMSO (0.1%), for 1 h followed by incubating both mixes with the probe GB123 (1 μM) for 2 h at 37 °C. The reaction was stopped by addition of Laemmli sample buffer (x4) (40% glycerol, 0.2 M Tris/HCl pH 6.8, 20% beta-mercaptoethanol, 12% SDS and 0.4 mg/ml bromophenol blue) and boiled for 10 min. Samples were separated on a 12% SDS gel and scanned for fluorescence by a typhoon scanner (FLA 9500) at excitation/emission wavelengths of 635/670 nm.

Paralysis and lifespan assays

On the second day of adulthood, randomly picked animals were transferred onto 60 mm NGM plates seeded with respective bacteria (10 plates for each treatment), 12 animals per plate. Worms were gently tapped with a platinum wire daily. In lifespan experiments, worms that failed to move were scored as dead. Aβ worms that could not crawl away were scored as paralyzed. Paralysis assays were stopped at day 12 of adulthood before the impact of age becomes pronounced while the lifespan experiments were continued till the last worm dies.

Thrashing assay

25 randomly picked, day 2 animals were transferred onto each 60 mm NGM plates seeded with respective bacteria. On each time point, 20 randomly picked animals (for each treatment), were sequentially placed in an M9 buffer drop on top of a microscope slide and allowed 30 s of recovery. After recovery, the number of body bends of that worm was counted for 30 s.

Fluorescent microscopy

For polyQ35-YFP foci measurement, on the indicated time point, worms were collected and immobilized in 20 mM sodium azide on a glass slide containing a 2% agarose pad in the center, and imaged immediately by an Nikon SMZ18 stereoscope.

SDS-PAGE and Western blot analysis

Worms were homogenized as described above and the homogenate was separated into debris and post-debris-supernatant (PDS) by centrifuging at 1500 g (4000 RPM in an Eppendorf desktop centrifuge, 5424) for 5 min. For each sample, 50 ug of protein was loaded and separated on a 12% polyacrylamide gels and transferred onto a nitrocellulose membrane. The membranes were then blocked in 5% nonfat milk and probed with the antibody against Aβ peptide (Biolegend, Clone 6E10, #803001 1:2000). Chemiluminescence was detected and quantified. The membrane was stripped using a Glycine-HCl buffer for 30 min, re-blocked in 5% non-fat milk and probed with antibody against β-Actin (Sigma Aldrich, #A5441, 1:2000) for normalizing the Aβ signal.

Native agarose gel electrophoresis (NAGE)

On each time point, worms were homogenized as before and separated into debris and PDS by centrifuging at 1500 g for 5 min. For each sample, 30 µg of PDS was loaded onto a 1% agarose gel and ran at 4 °C at 25 V for 16 h in a NAGE-specific running buffer. YFP fluorescence intensities in the gels were visualized using a ChemiDoc Imaging system (Biorad).

Immunoprecipitation and Mass-spectrometric analysis

Cathepsins activity was labeled using 400 μg protein extracts for each sample in acetate buffer with 2 μM GB123 (final concentration) at 37 °C. Equal amount of the protein sample was incubated with only buffer acetate without probe, to be used as a background control for the proteins binding to the beads nonspecifically.These samples were then incubated with an anti Cy5-antibody (1:60) (Sigma Aldrich, #C1117) overnight, in rotation at 4 °C.

Protein A/G beads (Santa Cruz, CA) were washed with PBS and 100 µl were added to each sample for 2 h at 4 °C on a rotating mixer. Beads were washed with PBS and then 50 µL of 1X Laemmli buffer was added to the beads and boiled for 10 min. Fluorescently labeled proteins immunoprecipitated were then separated on a 12% SDS–polyacrylamide gel electrophoresis (PAGE) at 4 °C for 10 h and the gel was transferred to 50% methanol:water before visualizing by Typhoon FLA 9500 scanner at 635/670 nm excitation/emission.

All 5 distinct bands were carefully cut out from the gel and stored at -20 °C, until LC-MS/analysis.

In-vitro chymotrypsin-like proteasome activity assay

In-vitro chymotrypsin-like proteasome activity assays were conducted using Suc-LLVY-AMC (Biomol # ABD-13453) according to the manufacturer instructions. 20 µM MG132 was used to inhibit proteasome activity. Measurements were performed using Infinite M200 PRP TECAN (Neotec Scientific Instrumentation Ltd.) with the Magellan Software at 37 °C.

RNA isolation, next generation sequencing and quantitative real-time PCR

Total RNA was isolated using a NucleoSpin® RNA kit (MACHEREY-NAGEL; 740955). For each time-point, 10,000 synchronized eggs were placed on NG-ampicillin plates seeded with HT115 bacteria. The worms, once they became adults, were washed daily with M9 to get rid of progeny and adult worms were transferred to new plates. Worms were harvested in M9 and frozen -80 C. The worms were thawed and subjected to mechanical disruption using magnetic beads and a Bullet blender® (Adaptas solutions). Homogenates were transferred to micro centrifuge tubes and centrifuged at 14,000 × g for 5 min. The supernatants were transferred to NucleoSpin® Filter (NucleoSpin® RNA kit, Macherey-Nagel, Düren Germany), and total RNA was purified according to the manufacturer’s instructions. The RNA was quantified using a NanoDrop2000c spectrophotometer.

For NGS

we used RNA ScreenTape kit (catalog #5067-5576; Agilent Technologies, Santa Clara, CA), D1000 ScreenTape kit (catalog #5067-5582; Agilent Technologies), Qubit® RNA HS Assay kit (catalog # Q32852; Invitrogen, Carlsbad, CA) and Qubit® DNA HS Assay kit (catalog #32854; Invitrogen). mRNA libraries were prepared using KAPA Stranded mRNA kit with mRNA Capture Beads (KAPA Biosystems, KK8421). In brief, 1 µg was used for the library construction; library was eluted in 20 µl of elution buffer. All DNA samples libraries were pooled to 10 nM sample. Multiplex samples Pool were loaded on NovaSeq 6000 (Illumina), using NovaSeq 6000 SP Reagent Kit v1.5 100 cycles (cat# 20028401), with 122 cycles of single-end sequencing.

For qPCR analysis

cDNA was prepared by reverse transcription of the total RNA samples via random-priming using the iScript^TM cDNA Synthesis Kit (#170–8890; Bio-Rad, Hercules, CA, USA) as per the manufacturer’s protocol. Analyzes by qPCR were performed with EvaGreen SuperMix (catalog #186–4035; Bio-Rad) and expression levels were normalized to the expression levels of cdc-42 and pmp-3. Primer sequences are listed in Supplementary Table 7. The qPCR reactions for each gene were performed in triplicates.

Computational analyses of next generation sequencing data

Raw reads were processed for quality trimming and adapters removal using fastx toolkit v0.0.14 and cutadapt v2.10 (Marcel M. et al., EMBnet.journal 2011, 17.1:10-12). The processed reads were aligned to the Caenorhabditis elegans transcriptome and genome version WBcel235 with annotations from Ensembl release 106 using TopHat v2.1.1 (Kim D et al., Genome Biology 2013, 14:R36). Counts per gene quantification was done with htseq-count v2.01 (Anders S et al., Bioinformatics 2015, 31 (2):166-169). Normalization and differential expression analysis were done with the DESeq2 package v 1.36.0 (Love MI et al., Genome Biology 2014, 15:550). Pair-wise comparisons were tested with default parameters (Wald test), without applying the independent filtering algorithm. For generating the volcano plot and heatmap (fig), significance threshold was taken as p-adj < 0.1. In addition, significant DE genes were further filtered by the log₂ fold change value. This filtering was base mean-dependent and required a base mean above 5 and an absolute log₂ fold change higher than 5/sqrt (base mean) + 0.3 (for highly expressed genes this means a requirement for a fold-change of at least 1.2, while genes with a very low expression would need a 5.8-fold-change to pass the filtering). For gene enrichment analysis, thresholds were set as at least an average of 20 reads (across 4 repeats) and an increase in expression by at least 50% or a decrease in expression to a level which is 67% or less of the expression level seen in untreated animals (log₂ > 0.58 or <-0.58, see Supplementary Table 1, p-value < 0.05).

Mass-spectroscopy

Sample preparation

Coomassie blue stained bands were excised, cut into 1 mm pieces and destained in 50% acetonitrile (ACN) and 50% 25 mM Tris-HCl, pH 8.0 for 15 min at 50 C^o with gentile agitation. Destaining solution was discarded and gel pieces were incubated for 5 min in ACN. Gel pieces were treated with 10 mM Dithiothreitol (DTT) (Sigma Chem. Co.) for 30 min followed by alkylation with 55 mM iodoacetamide (Sigma Chem. Co.) for 30 min at in the dark. The proteins were digested overnight at 37^oC with gentle agitation by using 0.4 µg of Trypsin (Mass Spectrometry grade, from Promega Corp., Madison, WI, USA) in 200 µl of 25 mM Tris-HCl, pH 8.0. The released peptides were extracted by addition of ACN and formic acid at final concentrations of 50% and 2.5% respectively and 30 min incubation at 50^oC. Tryptic peptides were desalted on C18 Stage tips (Rappsilber J, Mann M, Ishihama Y. Protocol for micro-purification, enrichment, pre-fractionation and storage of peptides for proteomics using StageTips. Nat Protoc. 2007;2(8):1896-906.). Peptides were eluted using 80% ACN, 0.1% formic acid and Lyophilized.

LC MS/MS analysis

Lyophilized peptides were reconstituted in 0.1% Formic Acid. For LC MS/MS analysis, 0.45 µg peptides were used for each sample. MS analysis was performed using a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) coupled on-line to a nanoflow UHPLC instrument (Ultimate 3000 Dionex, Thermo Fisher Scientific). Peptides were loaded on to a C18 column (75 um ID, 2 um, 100 Å, Thermo PepMap®RSLC) at a flow rate of 0.3 µl/min and separated on a gradient of 6%ACN to 50%ACN over 52 min at a flow rate of 0.15 µl/min. The survey scans (380–2,000 m/z, AGC target 3E, maximum injection time 50 ms) were acquired followed by higher energy collisional dissociation (HCD) fragmentation (normalized collision energy 25) for up to 15 dynamically chosen most abundant precursor ions (isolation window 1.8 m/z). A resolution of 70,000 was used for survey scans and 35,000 for MS/MS scans (AGC target 1E5 and maximum injection time of 121 ms).

MS data analysis

Mass spectra data were processed using the MaxQuant computational platform, version 2.0.3.0 (Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372 (2008)). Peak lists were searched against the Caenorhabditis elegans proteome obtained from Uniprot (UP000001940) and common lab contaminants. The search included cysteine carbamidomethylation as a fixed modification and oxidation of methionine as variable modifications. Peptides with minimum of seven amino-acid length were considered and the required FDR was set to 1% at the peptide and protein level.

Statistical analyses

The statistical tests used, statistical significance, error bars, and sample sizes can be found in the corresponding figure legends. “Statistically significant” was defined as p-value < 0.05.

Resource availability

All unique reagents generated in this study are available from Prof. Ehud Cohen with a completed material transfer agreement.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.