Cold activates a BH2 metabolic network promoting longevity and proteostasis

Emerging evidence demonstrates that moderately cold temperatures delay ageing and promote health across species¹. Lowering body temperature extends lifespan in both poikilotherms, including C. elegans, Drosophila melanogaster and various fish species, and homeotherms, including rodents^{1,2,3,4,5,6,7,8}. In adult C. elegans, a shift from standard (20 °C) to moderately low temperature (15 °C) markedly extends lifespan^2,9,10,11,12. In rodents, a mild decrease in body temperature by only around 0.5 °C prolongs lifespan, whereas a comparable increase shortens it^3,4. Lower body temperature also correlates with longevity in humans^13,14. Notably, average human body temperature has declined since the Industrial Revolution, raising the possibility that changes in body temperature may have contributed to increased longevity over recent decades¹³. Together, these findings suggest that moderate cold engages evolutionarily conserved programs that promote longevity.

Despite being recognized for more than a century¹⁵, the mechanisms by which moderate cold temperature promotes longevity remain largely unclear. Historically, cold-induced lifespan extension was interpreted as a consequence of passive thermodynamic effects, including reduced metabolic rate and slower biochemical reaction kinetics. However, studies in C. elegans revealed that cold-induced longevity is an actively regulated process^2,9,10,11,12. For example, the temperature-sensitive TRPA-1 ion channel promotes longevity at 15–20 °C but not at warmer temperatures, indicating that low temperature is sensed and transduced into longevity mechanisms^11,12. Moreover, cold temperature induces transcriptional programs that enhance proteostasis^2,9.

Despite these advances, a fundamental question remains of what systemic signals coordinate cold-induced responses across tissues. Small metabolites are ideally suited for this role because they can act as rapidly diffusible regulators integrating metabolism, enzymatic activity and intertissue signalling. We therefore hypothesized that cold temperature induces metabolite-based signals that coordinate metabolic, proteostatic and longevity pathways across tissues. We further hypothesized that such metabolites could function as temperature-independent effectors capable of recapitulating the beneficial effects of cold temperature even at higher temperatures.

Here we identify BH2 as a cold-induced metabolite of which the synthesis, intertissue transport and conversion into the bioactive enzymatic cofactor BH4 constitute a multi-tissue regulatory system that promotes longevity and maintains proteostasis across tissues. Through metabolomic, genetic and physiological analyses, we find that elevated BH2 synthesis in muscle and its conversion to BH4 in distal tissues promotes tyrosine synthesis, thereby mediating key aspects of cold-induced longevity. This pathway can be leveraged to confer longevity and proteostasis benefits at normal and elevated temperatures in both C. elegans and human cells.

Exposure of C. elegans to moderately low temperature (15 °C) after development induces a considerable extension of lifespan^2,10,11,12. To assess metabolic changes associated with this effect, we raised wild-type worms at the standard temperature (20 °C) until adulthood and subsequently transferred them to 15 °C, 20 °C or 25 °C. Untargeted metabolomics analysis on day 5 of adulthood identified 64 metabolites that were significantly increased at 15 °C relative to both 20 °C and 25 °C (Fig. 1a and Supplementary Table 1). An independent experiment confirmed 45 of these metabolites, with strong concordance between datasets (Extended Data Fig. 1a–d and Supplementary Table 2).

Among the 45 reproducibly cold-induced metabolites, several were linked to BH4 synthesis and its function as an enzymatic cofactor, including GTP, BH2, tryptophan, phenylalanine and tyrosine (Fig. 1a,b). As BH2 was among the most strongly induced metabolites at cold temperature (Fig. 1a), we focused subsequent analyses on this metabolite. Cold exposure after development increased BH2 levels from day 1 of adulthood, and BH2 remained elevated at 15 °C relative to age-matched animals at 20 °C until at least day 10 (Fig. 1c and Supplementary Table 3). BH2 induction was not a general consequence of longevity, as reduced insulin signalling, a distinct longevity pathway, did not increase BH2 levels at 20 °C (Extended Data Fig. 1e and Supplementary Table 4).

Loss of the temperature-sensitive trpa-1 channel shortens lifespan under both standard and cold temperatures¹¹. However, worms still exhibit substantially extended lifespan at 15 °C compared with at 20 °C, indicating that cold-induced longevity is not solely mediated by TRPA-1-dependent pathways. Loss of trpa-1 did not reduce BH2 levels at cold temperature (Fig. 1d and Supplementary Table 1), demonstrating that elevated BH2 synthesis is induced independently of canonical TRPA-1 temperature sensing.

We next examined whether BH2 reflects a regulated pathway rather than a passive metabolic consequence of reduced temperature. Quantitative proteomics identified GTP cyclohydrolase I (GCH1 in humans, CAT-4 in C. elegans), the first and rate-limiting enzyme in BH2 synthesis^16,17, as one of the most strongly induced proteins at 15 °C (Fig. 1e and Supplementary Table 5). Cold temperature also increased cat-4 mRNA amounts, indicating that CAT-4 induction occurs, at least in part, at the transcriptional level (Extended Data Fig. 2a). MED15 (human)/MDT-15 (C. elegans), a key transcriptional regulator of metabolic responses, is required for cold-induced longevity⁹. Importantly, knockdown of mdt-15 reduced cat-4 transcript levels at 15 °C (Extended Data Fig. 2b), indicating that cold-induced CAT-4 expression is part of a regulated transcriptional program.

Although BH2 is an inactive metabolite, it functions as a key intermediate in the synthesis and intertissue transport of its reduced, bioactive form, BH4^16,17. As BH4 is chemically unstable and readily oxidized, the more-stable BH2 form serves as a major organismal reservoir of this bioactive metabolite^16,17. Consequently, cat-4-mutant worms are severely depleted of BH4¹⁸. Consistent with the essential role of BH4 as a cofactor for multiple enzymes, these worms exhibit increased fragility compared with wild-type worms¹⁸. Accordingly, cat-4-mutant worms were short lived under both cold and standard temperatures (Extended Data Fig. 2c–e). Knockdown of cat-4 from early development also shortened lifespan at both 15 °C and 20 °C, although the effect was more pronounced under cold temperatures (Extended Data Fig. 2f).

As cold temperature during adulthood was sufficient to induce CAT-4 expression, BH2 synthesis and lifespan extension, we next examined the effects of cat-4 knockdown after development. As BH4 could not be detected by metabolomics because of its chemical instability, we quantified BH4 using an enzyme-linked immunosorbent assay (ELISA). Cold temperature induced a significant but modest increase in BH4 levels (Fig. 1f), consistent with BH2 serving as the major intermediate reservoir for this metabolite^16,17. Knockdown of cat-4 after development decreased BH4 levels under both standard and cold temperatures, although this reduction was moderate (Fig. 1f). Importantly, cat-4 RNA interference (RNAi) decreased BH4 at cold temperature only to levels comparable to those observed at standard temperature without knockdown (Fig. 1f). Thus, BH4 levels remained within a physiological range, enabling us to determine whether elevated CAT-4 activity during adulthood is required for cold-induced longevity. Notably, knockdown of cat-4 after development diminished the lifespan extension induced by cold temperature, but had no significant effect on lifespan at 20 °C or 25 °C (Fig. 1g).

To further dissect the adult-specific requirement for elevated CAT-4 in cold-induced longevity, we generated an auxin-inducible degron (AID) strain enabling conditional degradation of CAT-4 protein in somatic tissues (Extended Data Fig. 2g). CAT-4 degradation from early development substantially reduced lifespan at 20 °C (Extended Data Fig. 2h). By contrast, degradation of CAT-4 after development did not affect lifespan at 20 °C, but specifically shortened longevity at 15 °C (Extended Data Fig. 2i). These results further indicate that CAT-4 activity is essential during development for adult viability, whereas increased CAT-4 activity during adulthood specifically contributes to cold-induced longevity.

Prompted by these findings, we next examined whether CAT-4 activity also contributes to lifespan extension in other longevity paradigms. Knockdown of cat-4 in mutant worms with reduced insulin signalling did not affect their extended lifespan at standard temperature, but only shortened lifespan at 15 °C (Extended Data Fig. 3a,b). By contrast, knockdown of cat-4 shortened lifespan in long-lived worms with reduced mitochondrial electron-transport chain activity or reduced food intake at both cold and standard temperatures (Extended Data Fig. 3c–h), suggesting that CAT-4 activity contributes to other longevity paradigms beyond cold temperature.

Downstream of activity of GCH1/CAT-4, bioactive BH4 can be generated through three pathways: the de novo, salvage and recycling pathways^16,17 (Fig. 1b). In the de novo pathway, a GCH1/CAT-4-dependent intermediate is converted to BH4 by sepiapterin reductase. In the salvage pathway, this intermediate is instead converted to BH2 by either aldose reductase (AKR1B1 (human), Y39G8B.1 (C. elegans)) or carbonyl reductase 1 (CBR1 (human), DHS-31 (C. elegans))^16,19. This pathway operates alongside de novo BH4 synthesis, but is particularly important in mammalian tissues with low sepiapterin reductase activity and in organisms such as C. elegans, which lack a sepiapterin reductase orthologue^16,18. Knockdown of the BH2-producing enzyme Y39G8B.1 reduced lifespan specifically at 15 °C, whereas loss of dhs-31 produced a smaller, non-significant effect (Extended Data Fig. 4a). In the salvage pathway, BH2 is further converted into bioactive BH4 by dihydrofolate reductase (DHFR (human), DHFR-1 (C. elegans)). Importantly, knockdown of dhfr-1 after development decreased longevity at cold temperature but did not shorten lifespan at standard temperature (Fig. 1h). These results indicate that, downstream of CAT-4 activity, the conversion of BH2 into bioactive BH4 by DHFR-1 contributes to cold-induced longevity.

A recycling mechanism regenerates BH4 after its use as an enzymatic cofactor^16,18 (Fig. 1b). In C. elegans, the recycling pathway is not required to maintain steady-state BH4 levels and becomes rate-limiting only under conditions of low BH4 synthesis, such as in reduction-of-function cat-4 mutants¹⁸. Subsequently, knockdown of recycling enzymes did not shorten the lifespan of wild-type worms at cold temperature (Extended Data Fig. 4b). Likewise, mutant worms defective in BH4 recycling were not short lived at 15 °C compared with wild-type worms (Extended Data Fig. 4c,d), indicating that this pathway is dispensable for cold-induced longevity. Together, our findings support a central role of CAT-4 activity and the salvage pathway in elevated BH4 synthesis for cold-induced longevity.

To determine the tissues in which CAT-4 is expressed, we generated a GFP-based fluorescent reporter driven by the endogenous cat-4 promoter. Knockdown of cat-4 reduced the GFP signal, validating the reporter (Extended Data Fig. 5a). At standard temperature, CAT-4 was mostly expressed in the anterior intestine, and cold exposure further increased CAT-4 levels in this region (Extended Data Fig. 5a,b). In contrast to the intestine, CAT-4 was barely detectable in muscle at 20 °C (Fig. 2a). However, cold temperature triggered robust expression of cat-4 in the muscle, an effect that was abolished by cat-4 RNAi treatment (Fig. 2a,b). To assess whether elevated BH2 synthesis in these tissues promotes organismal longevity at cold temperature, we performed tissue-specific knockdown experiments. Notably, knockdown of cat-4 in the muscle alone reduced cold-induced longevity, whereas knockdown in the intestine had no effect on lifespan (Fig. 2c,d). Likewise, knockdown of cat-4 in neurons, epidermis or germline did not shorten lifespan at 15 °C (Fig. 2e–g).

To further investigate the link between cold-induced CAT-4 expression in muscle and longevity, we generated worms in which auxin-inducible degradation of CAT-4 was restricted to muscle. We confirmed that auxin treatment specifically depleted CAT-4 in the muscle of these animals, whereas ubiquitous somatic degradation reduced CAT-4 levels in both muscle and intestine (Extended Data Fig. 5c,d). Notably, muscle-specific degradation of CAT-4 after development was sufficient to fully phenocopy the longevity-shortening effects of ubiquitous somatic CAT-4 degradation at 15 °C (Extended Data Fig. 5e). This effect was specific to cold temperature, as muscle-specific CAT-4 degradation did not affect lifespan at standard temperature (Extended Data Fig. 5f). Together, these results identify muscle as a critical tissue in which elevated BH2 synthesis promotes cold-induced longevity.

BH4 is essential for the function of amino acid hydroxylases (TPH1 (human), TPH-1 (C. elegans); TH (human), CAT-2 (C. elegans); PAH (human), PAH-1 (C. elegans) and alkylglycerol monooxygenase (AGMO (human), AGMO-1 (C. elegans)). However, these enzymes are not expressed in muscle cells¹⁸, raising the possibility that BH4 acts in a cell non-autonomous manner. BH2 is the principal transport intermediate for BH4 distribution between tissues, enabling the rapid generation and utilization of bioactive BH4 in distal tissues^16,17. We therefore hypothesized that cold temperature increases BH2 synthesis in muscle cells, which is then released and transported to distal tissues, where it is converted into BH4 to activate its target enzymes. In humans, the uptake of BH2 is mediated by the transmembrane nucleoside transporters ENT1 (SLC29A1) and ENT2 (SLC29A2)¹⁶. C. elegans expresses seven orthologues of human ENT1 and ENT2 (ent-1 to ent-7). We found that single knockdown of different orthologues, such as ent-1, ent-5 and ent-6, specifically shortens lifespan at 15 °C but not 20 °C (Fig. 2h,i).

Once BH2 is taken up by distal cells, it can be converted into bioactive BH4 through the salvage pathway by the DHFR-1 enzyme^16,20. Consequently, BH4 redistribution efficiency depends on DHFR-1 activity, which regulates local BH4 availability. While BH2 synthesis through CAT-4 activity in muscle cells is required for cold-induced longevity (Fig. 2c), we found that muscle-specific knockdown of dhfr-1 does not shorten lifespan at 15 °C (Fig. 2j). Similarly, dhfr-1 knockdown in the epidermis had no effect on longevity at 15 °C (Fig. 2k). By contrast, tissue-specific knockdown of dhfr-1 in neurons, intestine and germline reduced cold-induced longevity (Fig. 2l–n). Moreover, neither neuron-specific knockdown of dhfr-1 nor muscle-specific knockdown of cat-4 further reduced the shortened lifespan of ent-1 mutant animals at cold temperature (Extended Data Fig. 5g,h). These results support the notion that cold temperature induces BH2 synthesis in muscle cells, after which, BH2 is released and converted to bioactive BH4 in distal tissues such as neurons, intestine and germline, where it promotes cold-induced longevity.

To further examine the cell non-autonomous effects of BH2, we focused on the intestine, a tissue that progressively loses structural integrity during ageing^21,22. To this end, we assessed the intermediate filament IFB-2 because this protein is expressed exclusively in the intestine and undergoes age-related changes that can be readily monitored²². In young animals, IFB-2 is confined to the apical cytoplasm of intestinal cells. With age, IFB-2 levels increase and the protein becomes mislocalized throughout the cytoplasm, where it forms aggregates that contribute to intestinal decline²². We found that cold temperature prevents IFB-2 mislocalization and aggregation during ageing (Extended Data Fig. 6a). However, loss of cat-4 abolished these protective effects (Extended Data Fig. 6a). Notably, muscle-specific knockdown of cat-4 was sufficient to block the cold-induced suppression of intestinal IFB-2 aggregation (Extended Data Fig. 6b). Collectively, these findings indicate that cold-induced BH2 synthesis in muscle exerts cell non-autonomous effects, such as preserving intestinal integrity during ageing.

BH2 supplementation is often more effective at increasing BH4 levels than direct BH4 administration^16,20. However, C. elegans exhibits a limited capacity for biopterin uptake, including BH2, and exogenous supplementation results in only a modest increase in biopterin levels¹⁸. Despite this limitation, BH2 supplementation in adult worms led to a slight but significant extension of cold-induced longevity (Extended Data Fig. 7a). Although BH2 supplementation did not extend lifespan at standard temperature, it delayed the age-related decline in motility, a key indicator of healthspan (Extended Data Fig. 7b,c). Although BH2-treated worms did not display overt signs of sickness, we further assessed the possibility that these beneficial effects arise from a hormetic stress response, such as that induced by mitochondrial dysfunction^23,24. However, BH2 supplementation did not reduce brood size and did not induce the mitochondrial unfolded protein response (Extended Data Fig. 7d,e), two commonly used indicators of physiological stress.

Given the limited uptake of exogenous biopterins in worms, we used a genetic approach to assess the full potential of these metabolites to extend longevity. Ubiquitous overexpression of cat-4 in somatic tissues resulted in a substantial increase in cold-induced longevity (Extended Data Fig. 7f). Notably, overexpression of cat-4 in the muscle alone was sufficient to increase organismal levels of bioactive BH4 and extend lifespan not only at 15 °C, but also at 20 °C and 25 °C (Fig. 3a,b and Extended Data Fig. 7g–i). Moreover, muscle-specific cat-4 overexpression also extended lifespan in loss-of-function trpa-1 mutant worms across different temperatures (Extended Data Fig. 7j–l). These findings indicate that increasing CAT-4 activity in muscle promotes longevity independently of both temperature and the cold-sensitive channel TRPA-1.

Consistent with a systemic protective effect, muscle-specific cat-4 overexpression prevented age-related accumulation and aggregation of intestinal IFB-2 protein at standard temperature (Fig. 3c and Extended Data Fig. 7m). As IFB-2 aggregation contributes to loss of intestinal integrity and subsequent bacterial colonization during ageing²², we examined whether CAT-4 prevents this deleterious process. Indeed, muscle-specific cat-4 overexpression reduced bacterial invasion of the intestine in aged worms at standard temperature (Fig. 3d,e). We then tested whether muscle-derived BH2 also acts in neurons. Neuronal-specific downregulation of either BH2 uptake (using ent-1 RNAi) or its conversion to BH4 through the salvage pathway (through dhfr-1 RNAi) significantly reduced the lifespan extension conferred by muscle-specific cat-4 overexpression at standard temperature, but had no effect in control worms (Extended Data Fig. 7n,o). These results further demonstrate that elevated BH2 production in muscle extends lifespan in a cell non-autonomous manner and requires both BH2 uptake and its conversion into bioactive BH4 in distal tissues, including the intestine and neurons.

We next examined whether increasing CAT-4 levels directly in the tissues that use BH2 is sufficient to promote longevity. While muscle-specific cat-4 overexpression had the strongest longevity effect, intestinal overexpression also markedly extended lifespan at both 15 °C and 20 °C (Fig. 3f,g). Although to a lesser extent, cat-4 overexpression in neurons also increased lifespan under both cold and standard temperatures (Fig. 3f,g). Together, these data suggest that, while elevated BH2 synthesis in muscle has a central role in longevity, increasing cat-4 expression in other tissues partially mimics its beneficial effects.

To determine how utilization of BH4 as an enzymatic cofactor contributes to cold-induced longevity, we performed a targeted RNAi screen against the four BH4-dependent enzymes expressed in C. elegans: phenylalanine hydroxylase (PAH-1), tryptophan hydroxylase (TPH-1), tyrosine hydroxylase (CAT-2) and alkylglycerol monooxygenase (AGMO-1)¹⁸. Among them, knockdown of pah-1 after development shortened lifespan, whereas knockdown of the other BH4-dependent enzymes did not affect cold-induced longevity (Extended Data Fig. 8a). Consistently, loss-of-function tph-1, cat-2 and agmo-1 mutants were not short-lived, whereas pah-1 mutants exhibited reduced longevity specifically at cold temperature (Extended Data Fig. 8b–d).

Among the tissues involved in cold-induced longevity, we found that PAH-1 activity in the intestine is particularly important for lifespan extension (Fig. 4a and Extended Data Fig. 8e–g). Consistent with this, knockdown of pah-1 increased both the levels and aggregation of the intestinal intermediate filament IFB-2, thereby abolishing the protective effects of cold temperature on age-associated IFB-2 aggregation (Fig. 4b and Extended Data Fig. 8h). While muscle-specific overexpression of cat-4 was sufficient to prevent IFB-2 aggregation during ageing at standard temperature, this protective effect was lost after pah-1 knockdown (Fig. 4c and Extended Data Fig. 8i).

PAH-1 catalyses the synthesis of tyrosine, a metabolite that we found to be upregulated at cold temperature (Fig. 1a). Exogenous supplementation with tyrosine extended lifespan not only at cold temperature, but also at standard temperature (Fig. 4d,e). Importantly, tyrosine supplementation rescued the reduction in cold-induced longevity caused by cat-4 knockdown (Extended Data Fig. 8j), further supporting a functional role of regulated tyrosine metabolism in mediating the longevity effects of elevated BH2 synthesis.

To gain insights into the downstream mechanisms by which tyrosine promotes longevity, we performed over-representation analysis (ORA) of the differentially abundant proteins in worms at 15 °C relative to either 20 °C or 25 °C (Supplementary Table 5). Notably, we found strong enrichment for proteins involved in small-molecule and amino acid metabolic processes (Extended Data Fig. 9a,b). Within these pathways, we identified increased levels of 4-hydroxyphenylpyruvate dioxygenase (HPD (human), HPD-1 (C. elegans); Extended Data Fig. 9c)—a key enzyme in the canonical tyrosine catabolic pathway that ultimately generates acetoacetate and fumarate, thereby linking tyrosine metabolism to central carbon metabolism²⁵. Moreover, analysis of published transcriptomic datasets also revealed increased hpd-1 mRNA levels at cold temperature¹² (Extended Data Fig. 9c). We found that loss of hpd-1 significantly reduced cold-induced longevity (Fig. 4f), supporting a functional contribution of tyrosine catabolism to lifespan extension at low temperature. Although hpd-1 knockdown also caused a slight decrease in lifespan at 20 °C and 25 °C, these effects did not reach statistical significance (Fig. 4f).

To identify molecular pathways that are responsive to elevated tyrosine levels, we performed quantitative proteomic analyses after tyrosine supplementation at both 15 °C and 20 °C (Supplementary Table 6). These analyses revealed 54 commonly upregulated proteins and 42 commonly downregulated proteins after tyrosine supplementation across both temperatures (Extended Data Fig. 9d,e). ORA of the differentially abundant proteins identified significant enrichment for lipid metabolic processes and enzyme regulator activity (Extended Data Fig. 9f,g). Together, these data suggest that tyrosine promotes longevity through specific downstream regulatory and metabolic pathways, including HPD-1-dependent tyrosine catabolism.

Loss of proteostasis during ageing is a major risk factor for diseases characterized by pathological protein aggregation²⁶. Moderately cold temperatures suppress disease-related protein aggregation across tissues, including polyQ-expanded proteins in C. elegans models of Huntington’s disease^2,9. Knockdown of cat-4 abolished this protective effect and increased aggregation without altering total polyQ levels (Extended Data Fig. 10a–c), indicating that CAT-4 activity modulates aggregation rather than overall protein amounts.

Consistent with the cell non-autonomous effects of muscle CAT-4 activity, loss of cat-4 increased aggregation at cold temperature not only in muscle, but also in worms expressing polyQ-expanded proteins exclusively in the intestine or neurons (Extended Data Fig. 10a–c). In neurons, polyQ67 aggregation causes behavioural deficits during ageing, including impaired chemotaxis and nose-touch avoidance compared with wild-type or polyQ19 controls^27,28. Cold temperature partially rescued these neuronal deficits, whereas loss of cat-4 abolished this protection (Extended Data Fig. 10d,e). Conversely, muscle-specific overexpression of cat-4 reduced neuronal polyQ67 aggregation under both standard (20 °C) and warmer (25 °C) temperatures, preventing deficits in chemotaxis and nose-touch avoidance (Extended Data Fig. 10f–h). These results support a cell non-autonomous role for muscle CAT-4 activity in maintaining proteostasis in distal tissues.

We next investigated how elevated CAT-4 activity enhances proteostasis capacity. Cold temperature induces the proteasome activator PA28γ (human, encoded by PSME3)/PSME-3 (C. elegans), increasing trypsin-like proteasome activity². Knockdown of psme-3 attenuated the ability of muscle-specific cat-4 overexpression to suppress neuronal polyQ67 aggregation, without altering total polyQ levels (Extended Data Fig. 11a), suggesting a broader defect in proteostasis rather than impaired degradation of polyQ proteins per se. Thus, CAT-4 activity appears to influence proteostasis through additional mechanisms. Besides proteasome activity, cold temperature also induces expression of different subunits of the TRiC/CCT complex¹⁰, a chaperonin that prevents polyQ-expanded aggregation^29,30,31. Knockdown of cct-2 subunit completely abolished the protective effects of muscle-specific cat-4 overexpression at standard temperature (Extended Data Fig. 11b). By contrast, knockdown of fkb-6, a cold-induced co-chaperone of HSP90¹⁰, had no effect on polyQ67 aggregation (Extended Data Fig. 11c). We also tested hsf-1, a key transcription factor that protects against protein misfolding and aggregation³². Loss of hsf-1 partially attenuated the protective effects of muscle-specific cat-4 overexpression (Extended Data Fig. 11d), although less strongly than cct-2 or psme-3 knockdown.

Finally, we assessed whether PAH-1 activity contributes to these proteostasis mechanisms. Indeed, loss of pah-1 reduced the cold-induced upregulation of cct-2 and psme-3 expression, as well as the increase in trypsin-like proteasome activity (Extended Data Fig. 11e–g). Together, these results indicate that elevated CAT-4 activity and subsequent BH4-dependent tyrosine synthesis enhance proteostasis through different downstream mechanisms, including proteasome activity and the TRiC/CCT chaperonin complex.

We next examined whether BH2 supplementation prevents disease-related protein aggregation in human cells at standard temperature (37 °C). To this end, we generated HEK293 cells expressing mutant huntingtin exon 1 containing polyQ74 repeats, which is prone to aggregation^33,34 (Fig. 5a). We found that supplementation with 10 µM BH2 markedly reduces polyQ aggregation without affecting overall protein levels (Fig. 5a). To determine whether BH2 also protects against aggregation of other disease-associated proteins, we assessed mutant variants of TDP-43 and FUS associated with amyotrophic lateral sclerosis (ALS). In contrast to their wild-type counterparts, the ALS-causing mutant TDP-43(A382T) and FUS(P525L) variants aggregated in HEK293 human cells (Fig. 5b,c). However, BH2 supplementation prevented aggregation of both mutant proteins (Fig. 5b,c). Consistent with this mechanism, exogenous tyrosine supplementation also reduced pathological protein aggregation (Fig. 5d).

Prompted by these findings, we examined whether BH2 attenuates disease-related neurodegeneration in human cells. Motor neurons differentiated from induced pluripotent stem (iPS) cells expressing the severe ALS-linked FUS(P525L) mutation exhibit elevated apoptosis compared with isogenic control cells³⁵. Notably, supplementation with either BH2 or tyrosine reduced apoptosis in ALS neurons (Fig. 5e,f). ALS motor neurons also showed elevated phosphorylation of receptor-interacting protein (RIP) kinase, a marker of necroptotic cell death, which was reduced by BH2 supplementation (Fig. 5g), further supporting a protective effect against disease-related neurodegeneration.

We then examined whether BH2 and tyrosine enhance proteostasis capacity through mechanisms similar to those activated by cold temperature in C. elegans. Indeed, both treatments caused a slight but significant increase in trypsin-like proteasome activity in HEK293 cells at standard temperature (Extended Data Fig. 12a,b). By contrast, neither treatment increased LC3 lipidation, a marker of autophagosomes (Extended Data Fig. 12c,d). As BH2 suppressed aggregation without significantly altering total levels of the pathological proteins (Fig. 5a–c), we examined additional proteostasis mechanisms. Consistent with our findings in worms, knockdown of the TRiC/CCT chaperonin subunit CCT2 reduced the ability of BH2 to suppress pathological protein aggregation, identifying TRiC/CCT-complex-dependent folding capacity as a key contributor to BH2-mediated proteostasis enhancement in human cells (Extended Data Fig. 12e).

In contrast to in C. elegans, mammalian cells express nitric oxide synthases (NOSs), which require BH4 as a cofactor^16,18. Under conditions that decrease intracellular BH4 availability, such as oxidative stress—which promotes excessive oxidation of BH4 to BH2—or impaired DHFR activity, BH2 can compete with BH4 for binding to endothelial NOS (eNOS) in the vascular system^36,37,38. This shifts eNOS activity from nitric oxide production toward superoxide generation, a process known as eNOS uncoupling³⁸. Importantly, eNOS uncoupling is associated primarily with reduced BH4 availability and a consequent decrease in the BH4/BH2 ratio, rather than with elevated BH2 levels per se or BH2 supplementation. In fact, when cells retain an intact DHFR-dependent salvage pathway, BH2 supplementation increases intracellular BH4 levels and alleviates eNOS uncoupling under oxidative stress^36,37. In contrast to paraquat-induced oxidative stress, neither 10 µM BH2 nor a 20-fold higher concentration increased the percentage of cells with detectable superoxide levels (Extended Data Fig. 12f). Moreover, even high doses of BH2 reduced pathological protein aggregation (Extended Data Fig. 12g), supporting the proteostasis benefits of BH2 supplementation without detectable superoxide production indicative of BH4 imbalance.

Given that oxidative stress promotes BH4 oxidation to BH2, we hypothesized that such conditions might limit the therapeutic potential of BH2 supplementation. Indeed, paraquat-induced oxidative stress abolished the protective effect of BH2 on protein aggregation (Extended Data Fig. 12h). By contrast, tyrosine supplementation retained its protective effect under oxidative stress (Extended Data Fig. 12i), suggesting that this downstream effector of the pathway can bypass conditions in which BH2 cannot be efficiently converted into BH4.

Here we identify BH2 as a cold-induced metabolite that links low temperature to systemic regulation of longevity and proteostasis (Extended Data Fig. 12j). Moderate cold increases CAT-4-dependent BH2 synthesis in muscle, from which BH2 is transported to distal tissues and converted into bioactive BH4 by DHFR-1. This establishes muscle as a metabolic hub that produces a pro-longevity signal of which the activity can be regulated locally in recipient tissues. BH4-dependent PAH-1 activity promotes tyrosine synthesis, and tyrosine supplementation recapitulates the longevity benefits of this pathway even at normal temperature, identifying the PAH-1–tyrosine axis as an important downstream mediator of the cold-induced BH2 pathway. The pathway also enhances organismal proteostasis: increasing BH2 synthesis in muscle suppresses pathological protein aggregation and preserves neuronal and intestinal function in distal tissues. Together, these findings expand the concept of cold-induced longevity beyond a predominantly cell-autonomous response to a systemic metabolic communication mechanism that coordinates longevity and proteostasis across tissues.

The ability of BH2 and tyrosine to suppress disease-associated protein aggregation and neurodegeneration-related phenotypes in human cells further supports functional conservation of this pathway. However, its translational potential may depend on the cellular context in which BH2 is converted to BH4. In mammals, BH4 is also required by NOS, and oxidative stress or impaired DHFR activity can disrupt the BH4/BH2 balance and promote eNOS uncoupling in endothelial cells^36,37,38. BH2 supplementation did not increase detectable superoxide production in our cellular models, even at high concentrations, although these experiments do not directly address potential effects on NOS uncoupling in vivo. Consistent with the importance of BH2-to-BH4 conversion, we found that oxidative stress limits the protective effect of BH2, whereas tyrosine retains its protective effect. Thus, tyrosine may provide an alternative means of engaging this pathway when BH2-to-BH4 conversion is impaired. More broadly, our metabolomic analyses identified numerous additional metabolites induced by cold, raising the possibility that BH2 is one component of a larger network of systemic metabolic signals that coordinates the beneficial effects of low temperature. Defining these pathways may reveal additional strategies to reproduce the health-promoting effects of cold without altering body temperature.

C. elegans strains and maintenance

C. elegans were maintained at 20 °C on standard nematode growth medium (NGM) seeded with OP50 E. coli³⁹. In all the experiments, worms were raised at the standard temperature (20 °C) until adulthood and then transferred to the temperatures indicated in the corresponding figures (15 °C, 20 °C and 25 °C). All of the experiments were performed using hermaphrodite adult worms at the ages indicated in the corresponding figures and figure legends.

Wild-type (N2), TQ233 (trpa-1(ok999) IV, outcrossed 6× to wild-type N2 worms), LC141 (him-8(e1489) IV; cat-4(e3015) V, outcrossed 6× to N2), LC35 (cat-4(ok342) V, outcrossed 3× to N2), LC81 (cat-4(tm773) V, outcrossed 5× to N2), LC143 (pcbd-1(tm5924) I, outcrossed 4× to N2), LC87 (qdpr-1(tm2337) III, outcrossed 2× to N2), MT15620 (cat-2(n4547) II, outcrossed 6× to N2), MT14984 (tph-1(n4622) II, outcrossed 6× to N2), LC144 (agmo-1(e3016) III, outcrossed 6× to N2), LC74 (pah-1(ok687) II, outcrossed 5× to N2), MQ887 (isp-1(qm150) IV, outcrossed 3× to N2), DA1116 (eat-2(ad1116) II, outcrossed 2× to N2), DA597 (phm-2(ad597) I, outcrossed 2× to N2), SJ4100 (zcIs13[hsp-6p::GFP]) V, outcrossed 3× to N2) and AM141 (rmIs133[unc-54p::Q40::YFP] X, outcrossed 5× to N2) strains were obtained from the Caenorhabditis Genetics Center at the University of Minnesota (CGC), which is supported by the NIH Office of Research Infrastructure Programs (P40 OD010440). CF1041 (daf-2(e1370) III) and CF512 (fer-15(b26) II; fem-1(hc17) IV) were provided by C. Kenyon. BJ49 (kcIs6[ifb-2p::ifb-2a::CFP] IV)⁴⁰ was a gift from R. E. Leube. The AM23 (rmIs298[F25B3.3p::Q19::CFP]) and AM716 (rmIs284[F25B3.3p::Q67::YFP]) strains were provided by R. I. Morimoto²⁷. MAH602 (sqIs61[vha-6p::Q44::YFP + rol-6(su1006)]) was gifted by M. Hansen⁴¹.

For tissue-specific RNAi experiments, we used either rde-1- or sid-1-mutant worms in which wild-type rde-1 or sid-1 genes were rescued using tissue-specific promoters, respectively. The strains WM118 (muscle RNAi, rde-1(ne300) V; neIs9[myo-3p::HA::RDE-1 + rol-6(su1006)])⁴², IG1846 (epidermis RNAi 1, rde-1(ne300) V; frSi21[col-62p::rde-1 3′UTR] II; frIs7[nlp-29p::GFP + IV)⁴³, NR222 (epidermis RNAi 2, rde-1(ne219) V; kzIs9[lin-26p::NLS::GFP + + rol-6(su1006)])⁴⁴, MAH677 (neuronal RNAi, sid-1(qt9); sqIs71 [rgef-1p::GFP + rgef-1p::sid-1]⁴⁵, IG1839 (intestine RNAi 1, rde-1(ne300) V; frSi17[mtl-2p::rde-1 3UTR] II; frIs7 [nlp-29p::GFP + col-12p::DsRed] IV)⁴³, VP303 (intestine RNAi 2, rde-1(ne219) V; kbIs7[nhx-2p::rde-1 + rol-6(su1006)])⁴⁶, and DCL569 (germline RNAi, mkcSi13[sun-1p::rde-1::sun-1 3′UTR + unc-119(+)] II; rde-1(mkc36) V) were provided by the CGC. The RB891 strain (ent-1(ok743) IV) was also obtained from the CGC and outcrossed four times to N2 wild-type worms to generate DVG397. Then, DVG397 was crossed with MAH677 and WM118 to generate DVG399 and DVG400, respectively.

The control DVG9 strain (N2, ocbEx9[myo3p::GFP]) was generated by microinjecting 20 ng μl⁻¹ pPD93_97 (myo-3p::GFP) into the gonads of adult N2 hermaphrodite animals⁴⁷. To generate DVG348 (N2, ocbEx280[sur-5p::cat-4, myo-3p::GFP]), we injected a DNA mixture containing 70 ng μl⁻¹ of the sur5-p::cat-4 plasmid and 5 ng μl⁻¹ pPD93_97 (myo3-p::GFP). Following this method, we generated the strains DVG349 (intestine-specific overexpression: N2, ocbEx281[gly-19p::cat-4, myo-3p::GFP]), DVG351 (neuron-specific overexpression: N2, ocbEx283[rgef-1p::cat-4, myo-3p::GFP]), and DVG352 (muscle-specific overexpression: N2, ocbEx284[myo-3p::cat-4, myo-3p::GFP]). Likewise, we used the neuronal-specific RNAi strain (MAH677) to generate DVG377 (sid-1(qt9); sqIs71[rgef-1p::GFP + rgef-1p::sid-1]; ocbEx301[myo-3p::cat-4, myo-3p::mCherry]) and the corresponding control DVG378 (sid-1(qt9); sqIs71[rgef-1p::GFP + rgef-1p::sid-1]; ocbEx302[myo3p::mCherry]). To generate DVG389 (trpa-1(ok999) IV; ocbEx309[myo-3p::GFP]) and DVG390 (trpa-1(ok999) IV; ocbEx310[myo-3p::cat-4, myo-3p::GFP]), we injected the gonads of TQ233 worms as indicated above.

To generate DVG373 (rmIs298[F25B3.3p::Q19::CFP]; ocbEx297[myo-3p::cat-4, myo-3p::mCherry]) and DVG375 (rmIs284[F25B3.3p::Q67::YFP]; ocbEx299[myo-3p::cat-4, myo-3p::mCherry]), we injected the gonads of AM23 and AM716 worms with a DNA mixture containing 100 ng μl⁻¹ of the myo3-p::cat-4 plasmid and 5 ng μl⁻¹ of pCFJ104 (myo-3p::mCherry). The corresponding control strains, DVG376 (rmIs298[F25B3.3p::Q19::CFP]; ocbEx300[myo-3p::mCherry]) and DVG380 (rmIs284[F25B3.3p::Q67::YFP]; ocbEx304[myo-3p::mCherry]), were generated by injecting 5 ng μl⁻¹ of the pCFJ104 plasmid (myo-3p::mCherry) alone.

The VDL20/PHX10108 strain expressing endogenous CAT-4 tagged with GFP (cat-4(syb10108)) was generated by SunyBiotech (http://www.sunybiotech.com/) using the following sgRNAs: sg1, CACCACAACATCCTGCATGTTGG; and sg2, CCGTGACGATCCAAAGACTCGCG. Genome editing was confirmed by sequencing the cat-4 gene fused to GFP (primers: 5′-CATTTAGTTCCATTTATGGG-3′ and 5′-TTTGGAGCTCGAGTCGCTTA-3′). From the VDL20 strain, we generated DVG387 (cat-4::SL2::GFP) by CRISPR–Cas9-mediated insertion⁴⁸ of an SL2 sequence using the crRNA AAAGGAGAAGAACTTTTCAC, enabling independent translation of CAT-4 and GFP from a single bicistronic pre-mRNA transcribed under the endogenous cat-4 promoter. The editing was confirmed using the following primers: 5′-CCGGAGTTGCAGTTGTTATTG-3′ and 5′-TCTCGAGAAGCATTGAACACC-3′.

For all C. elegans experiments, exact sample sizes are indicated in the corresponding figure legends. No statistical methods were used to predetermine sample size. Sample sizes were selected based on standards established in the field and were similar to or greater than those reported to be sufficient in previous studies using the same procedures (for example, lifespan, nose-touch, chemotaxis and motility assays)^{2,10,28,49,50,51,52}. Worm populations were synchronized either by allowing young hermaphrodites to lay eggs for 6 h or by bleaching⁵³. For synchronization by egg laying, young hermaphrodites were randomly selected from maintenance plates. For synchronization by bleaching, agar chunks containing mixed-stage animals were randomly selected from maintenance plates, and larvae were allowed to develop until sufficient young hermaphrodites were available for bleaching. After the 6 h egg-laying period or bleaching, larvae were raised to adulthood. For lifespan assays, adult worms were randomly selected from the synchronized population and allocated to the different experimental conditions. For all other experiments, adult worms were randomly distributed from single synchronized populations into the different experimental groups. Thus, allocation of worms to experimental conditions was randomized throughout the study. Samples from different experimental conditions were collected and lysed in a randomized order to minimize potential processing-order effects. Samples and experimental conditions were not processed in a blinded manner. To minimize potential bias and assess reproducibility, key experiments, including phenotypic assays, were independently repeated by at least two researchers involved in the study.

Metabolomics

Adult worms were collected at the ages indicated in the corresponding figures. Worms were lysed in an extraction solution consisting of 50% LC–MS–grade methanol (Thermo Fisher Scientific, 10284580), 30% LC–MS–grade acetonitrile (Thermo Fisher Scientific, 10001334), 20% ultrapure water and 5 μM valine-d8 (CK Isotopes, DLM-488), using a volume equivalent to 10× the worm pellet weight. The worm pellet was transferred to a Precellys tube and homogenized using a Precellys Evolution Touch Homogenizer (Bertin Technologies). The lysates were then transferred to Eppendorf tubes and incubated for 15 min at 4 °C on a shaker at maximum speed. After incubation, the samples were centrifuged at the maximum speed for 25–40 min and the supernatant was collected.

The worm extracts were normalized by weight for metabolomics analysis of all possible metabolites. Metabolites were separated chromatographically using a Millipore SeQuant ZIC-pHILIC analytical column (5 µm, 2.1 × 150 mm) with a matching 2.1 × 20 mm guard column, both with 5 µm particle size. A binary solvent system was used, consisting of 20 mM ammonium carbonate with 0.05% ammonium hydroxide as solvent A and pure acetonitrile as solvent B. Chromatography was performed at 40 °C with a flow rate of 0.200 ml min⁻¹, using the following gradient program: 80% solvent B from 0 to 2 min, a linear gradient to 20% solvent B from 2 to 17 min, a rapid return to 80% solvent B from 17 to 17.1 min, and a hold at 80% solvent B from 17.1 to 23 min. The autosampler tray was maintained at 4 °C, and 5 µl of each sample was injected. The samples were randomized to minimize analytical variability, and pooled quality-control samples, prepared by mixing equal volumes of all individual samples, were analysed regularly to ensure consistency.

Metabolite quantification was performed using a Vanquish Horizon UHPLC system coupled to the Orbitrap Exploris 240 mass spectrometer (Thermo Fisher Scientific) with a heated electrospray ionization (HESI) source. Ionization parameters were set to +3.5 kV (positive mode) and −2.8 kV (negative mode), with an RF lens value of 70, a heated capillary temperature of 320 °C and an auxiliary gas heater temperature of 280 °C. Sheath gas flow was set to 40, auxiliary gas to 15 and sweep gas was off. MS1 scans covered a mass range of m/z 70–900, with automatic gain control (AGC) targets and maximum injection times. Data were acquired in full-scan mode with polarity switching at a resolution of 120,000. The AcquireX Deep Scan workflow facilitated untargeted metabolite identification by iterative data-dependent acquisition. Fragmentation was conducted with resolutions of 60,000 for full scans and 30,000 for MS/MS, an intensity threshold of 5.0 × 10³, a 10 s dynamic exclusion, a 5 ppm mass tolerance, an isolation window of 1.2 m/z and stepped normalized collision energies of 30, 50 and 150. Mild trapping was enabled to improve signal quality.

Metabolite identification was performed using Compound Discoverer software (v.3.2, Thermo Fisher Scientific), with criteria including a precursor ion m/z within 5 ppm of the theoretical mass, fragment ions within 5 ppm of an internal spectral library, a minimum match score of 70 and retention times within 5% of purified standards analysed under identical conditions. For downstream analyses, we focused on annotated metabolites with spectra and retention times matching entries in our in-house library, which contains over 1,000 compound standards. Peak area integration and chromatogram review were carried out using TraceFinder software (v.5.0, Thermo Fisher Scientific). To correct any potential variations introduced from sample handling to instrument analysis, we normalized the detected metabolites in each sample using the probabilistic quotient normalization method⁵⁴. The data were processed using the R package MetaProViz⁵⁵ (https://github.com/saezlab/MetaProViz), with the only modification being that probabilistic quotient normalization was performed using 100% of the features from the multtest package (v.2.56.0)⁵⁶, instead of total ion count normalization. Downstream statistical analyses were performed using GraphPad Prism (v.10.6.0). Data were analysed with two-sided unpaired t-tests, followed by the two-stage step-up method of Benjamini, Krieger and Yekutieli to correct for multiple comparisons.

For correlation analysis of metabolite changes between independent datasets (experiments 1 and 2), metabolites were filtered to retain metabolites, log2[fold change] and FDR-adjusted P (q value). The datasets were merged using an inner join on metabolite names. Significance was defined as an FDR-adjusted P < 0.01. Pearson correlation was calculated using only metabolites that were significant in both datasets. Scatter plots showing log₂-transformed fold changes from experiments 1 and 2 were generated using the ggplot2 package (v.4.0.3) in R (v.4.5.2). Reference lines at zero (x and y axes) and a diagonal identity line (slope = 1) were included to aid interpretation. The Pearson correlation coefficient (r) and corresponding P value were annotated on the plot. Extreme outliers (for example, carnitine-C16) were excluded before visualization. For Venn diagram analysis, significantly upregulated metabolites were identified independently in each dataset using thresholds of fold change > 1 and FDR-adjusted P < 0.01. Venn diagrams were generated using the ggvenn package (v.0.1.19) in R (v.4.5.2). Dot plots comparing the 45 commonly and significantly upregulated metabolites (fold change > 1 and FDR-adjusted P < 0.01) across experiments 1 and 2 were generated using ggplot2 (v.4.0.3) in R (v.4.5.2).

ELISA quantification of BH4 levels

We quantified BH4 levels using a BH4 ELISA Kit (Novus, NBP2-74983), according to the manufacturer’s instructions. Worms were lysed at day 5 of adulthood in protein lysis buffer (50 mM Tris-HCl, pH 7.8, 150 mM NaCl, 0.25% sodium deoxycholate, 1 mM EDTA and protease inhibitor (Roche)) using a Precellys 24 homogenizer. Worm lysates were centrifuged at 10,000 rpm for 10 min at 4 °C and the supernatant was collected. Protein concentrations were determined with standard BCA protein assay (Thermo Fisher Scientific). In total, 650 µg of total protein was used for measurement.

Label-free quantitative proteomics

For comparison of protein abundance across temperatures (15 °C, 20 °C and 25 °C), synchronized day 5 adult worms were lysed in urea buffer (8 M urea, 50 mM triethylammonium bicarbonate (TEAB)) by glass-bead homogenization and cleared by centrifugation (13,000 rpm, 10 min). The supernatants were reduced with 5 mM dithiothreitol (DTT) for 1 h, alkylated with 40 mM chloroacetamide for 30 min and digested with trypsin (1:75 ratio, w/w) after dilution of urea to 2 M with 50 mM TEAB. For tyrosine-treatment experiments, day 3 adult worms were lysed in urea buffer as indicated above. The supernatants were reduced with 10 mM DTT for 30 min, alkylated with 40 mM chloroacetamide for 30 min and digested with Lysyl endopeptidase (Lys-C, 1:100 ratio) at 25 °C for 2 h. Proteins were then further digested with trypsin (1:50 ratio) after dilution of urea to 2 M with 50 mM TEAB. One day later, the samples were acidified with formic acid (1% final concentration), cleared by centrifugation (16,000 g for 5 min), purified using StageTips and subjected to mass spectrometry (MS) analysis.

MS was performed at the CECAD Proteomics Facility. Temperature-comparison samples were analysed on the Orbitrap Exploris 480 mass spectrometer equipped with a FAIMSpro differential ion mobility device coupled to an UltiMate 3000 (Thermo Fisher Scientific). Tyrosine-treatment samples were analysed on the Orbitrap Eclipse coupled to a Vanquish neo system (Thermo Fisher Scientific). The samples were loaded onto a precolumn (Acclaim 5 µm PepMap 300 µ Cartridge or Ionopticks NanoShield, respectively) before being reverse-flushed onto an in-house-packed analytical column (30 cm length, 75 µm inner diameter, filled with 2.7 µm Poroshell EC120 C18, Agilent) or an Aurora Ultimate 25 cm column (Ionopticks), respectively. Peptides were chromatographically separated at a constant flow rate of 300 nl min⁻¹ on a 90 min linear gradient running 0.1% formic acid against increasing amounts of 80% acetonitrile with 0.1% formic acid. For measurements where FAIMS pro was used, it was operated at constant −50 V compensation voltage and electrode temperatures of 99.5 °C for the inner and 85 °C for the outer electrode.

For the gas-phase fractionated library⁵⁷, a pool generated from all samples was analysed in six individual runs covering the range from 400 m/z to 1,000 m/z in 100 m/z increments. For each run, MS1 was acquired at 60,000 resolution with a maximum injection time of 98 ms and an AGC target of 100%. MS2 spectra were acquired at 30,000 resolution with a maximum injection time of 60 ms. Spectra were acquired in staggered 4 m/z windows, resulting in nominal 2 m/z windows after deconvolution using ProteoWizard⁵⁸.

For each sample acquired on the Orbitrap Exploris 480, MS1 scans were obtained from 399 m/z to 1,001 m/z at 15,000 resolution. The maximum injection time was set to 22 ms and the AGC target was set to 100%. MS2 scans ranged from 400 m/z to 1,000 m/z and were acquired at 15,000 resolution with a maximum injection time of 22 ms and an AGC target of 100%. DIA scans covering the precursor range from 400 to 1,000 m/z and were acquired in 60 × 10 m/z windows with an overlap of 1 m/z. All of the scans were stored as a centroid. For samples acquired on Orbitrap Eclipse, the MS1 and MS2 resolution was set to 30,000 with a maximum injection time of 54 ms and identical AGC targets. The same range of 400 to 1,000 m/z was covered using 20 windows of 30 m/z each.

For samples acquired on the Orbitrap Exploris 480 (temperature-comparison datasets), the gas-phase fractionated library was build using DIA-NN (v.1.8.1)⁵⁹ and the C. elegans proteomic database from WormBase (PRJNA13758.WBPS15, downloaded on 27 July 2021), with settings matching acquisition parameters. The samples were analysed in DIA-NN v.1.8.1 as well using the previously generated library and identical database. DIA-NN was run with the additional command line prompts ‘--report-lib-info’ and ‘--relaxed-prot-inf’. Further output settings were: filtered at 0.01 FDR, N-terminal methionine excision enabled, maximum number of missed cleavages set to 1, minimum peptide length set to 7, maximum peptide length set to 30, minimum precursor m/z set to 400, maximum precursor m/z set to 1,000 and cysteine carbamidomethylation enabled as a fixed modification. Next, the DIA-NN output was further filtered on a library q value and global q value of ≤ 0.01 and at least two unique peptides per protein using R (v.4.1.3). Then, label-free quantification values were calculated using the DIA-NN R package. The results were analysed using Perseus (v.1.6.15)⁶⁰. Proteins were filtered for data completeness in at least one replicate group and missing values imputed using sigma downshift approach with standard settings. Next, FDR-controlled two-sided t-tests were performed between the experimental groups and WormBase IDs mapped to UniProt IDs.

For samples acquired on Orbitrap Eclipse (tyrosine-treatment datasets), a directDIA approach using Spectronaut 20.5 (Biognosys) was used with the GPF runs added as supporting runs and searched against the WormBase C. elegans proteomic database (PRJNA13758.WBPS19, downloaded 4 January 2025). The Spectronaut output was imported into R (v.4.5.2) for downstream analysis. Protein groups were expanded to individual UniProt identifiers, mapped to gene symbols using the C. elegans annotation database (org.Ce.eg.db, v.3.22.0), and protein abundances were summarized at the gene level using median values. Proteins with >50% missing values across samples were excluded. Protein intensities were log₂-transformed after addition of a pseudocount of 1. Differential protein abundance analysis was conducted using the limma package (v.3.66.0) with two-sided empirical Bayes-moderated t-test. Proteins were considered differentially abundant when they exhibited an absolute log₂-transformed fold change of >0.58 (corresponding to a 1.5-fold change) and a nominal P value < 0.05.

ORA

ORA was performed using the enrichGO function from the clusterProfiler package in R (v.4.5.2) to identify Gene Ontology (GO) terms enriched among differentially abundant proteins. Analyses were conducted using the org.Ce.eg.db annotation database (v.3.22.0). GO Biological Process and Molecular Function ontologies were analysed using a one-sided hypergeometric test. Multiple-testing correction was performed using the Benjamini–Hochberg method, with significance thresholds of P < 0.05 and q < 0.2. For visualization, enrichment results were converted to gene ratios (GeneRatio), representing the proportion of input genes associated with each GO term. Dot plots were generated using ggplot2 (v.4.0.3), with the x axis representing the gene ratio, the y axis indicating GO terms, point size corresponding to the number of genes associated with each term (Count) and point colour representing statistical significance (−log₁₀[P]).

Analysis of transcriptomic data

The temperature-dependent gene expression data from day 3 adult C. elegans at 15 °C and 25 °C were obtained from the NCBI Gene Expression Omnibus (GEO) database under accession number GSE62297 (ref. ¹²). Raw data were reanalysed using the database’s in-built GEO2R web tool with default parameters, including multiple-testing correction using the Benjamini–Hochberg method to calculate FDR-adjusted P values.

qPCR

Total RNA was isolated from synchronized day-5 adult worms using RNAbee (Tel-Test, CS501-B). We then generated cDNA from isolated RNA using the qScript Flex cDNA synthesis kit (Quantabio). Quantitative PCR (qPCR) assays were performed with SYBR Green and a 1:20 dilution of cDNA on the CFX384 Real-Time System (Bio-Rad). Gene expression levels were normalized to the geometric mean of cdc-42 and Y45F10D.4 (ref. ⁶¹). Data were analysed using the comparative \({2}^{\varDelta \varDelta {C}_{{\rm{t}}}}\) method, which provides relative changes in gene expression to the control condition after correction for housekeeping genes⁶². Details about the primers used for qPCR are provided in Supplementary Table 8.

Imaging and quantification of the CAT-4 reporter

For imaging of the DVG387 (cat-4p::cat-4::SL2::GFP) reporter strain or AID-strains (cat-4::3×Flag::mIAA7::GFP), day 5 adult worms were immobilized with 0.1% sodium azide in M9 buffer and mounted on glass slides under coverslips. Fluorescence images of whole worms were acquired using the Zeiss Axio Zoom.V16 microscope. For quantification of GFP intensity, the anterior region of the intestine was outlined and quantified using ImageJ (v.1.54p). For muscle regions, higher-magnification images were acquired using the Zeiss ApoTome Axio Imager Z.1 fluorescence microscope.

RNAi constructs

Hermaphrodite worms were fed HT115 E. coli carrying either an empty control vector (L4440) or vectors expressing double-stranded RNAi. The RNAi constructs targeting cat-4, ent-1, ent-6 and hsf-1 were obtained from the Ahringer library. The mdt-15, Y39G8B.1, dhs-31, dhfr-1, qdpr-1, pcbd-1, ent-4, ent-5, cco-1, agmo-1, cat-2, pah-1, tph-1, hpd-1, cct-2 and fkb-6 RNAi constructs were obtained from the Vidal library. The daf-16 and psme-3 RNAi constructs were generated in previous studies^2,63. All RNAi constructs were sequence verified. A list of the RNAi sequences is provided in Supplementary Table 8.

AID

Genome editing was performed using a CRISPR protocol as previously described⁴⁸. The strain PP4642, cat-4(hh131[cat-4::3×Flag::mIAA7::gfp] V, was generated by inserting a C. elegans codon-optimized 3×Flag::mIAA7 sequence in-frame immediately upstream of the GFP sequence in the strain VDL20 cat-4(syb10108[cat-4::GFP]) V. The strain PP4640, wrdSi58[eft-3p::TIR1(F79G)::F2A::mTagBFP2::AID*::NLS::tbb-2 3′UTR] II, was generated by introducing the F79G mutation into TIR1 in strain JDW313. Similarly, PP4641 (reSi11[unc-54p::TIR1(F79G)::F2A::mTagBFP2::AID*::NLS::tbb-2 3′UTR] II) was generated by mutating TIR1(F79G) using DV3825 strain. All genomic modifications were confirmed by Sanger sequencing. A list of the oligonucleotide sequences used for strain generation and sequencing is provided in Supplementary Table 8.

PP4640, PP4641 and PP4642 were each outcrossed five times to N2 wild-type worms. The strains PP4634 (cat-4(hh131[cat-4::3×Flag::mIAA7::GFP] V;wrdSi58[eft-3p::TIR1(F79G)::F2A::mTagBFP2::AID*::NLS::tbb-2 3′UTR] II) and PP4635 (cat-4(hh131[cat-4::3×Flag::mIAA7::GFP] V; reSi11[unc-54p::TIR1(F79G)::F2A::mTagBFP2::AID*::NLS::tbb-2 3′UTR] II) were generated by crossing PP4642 with PP4640 and PP4641, respectively.

AID experiments were performed as described previously⁶⁴ with minor modifications. To induce CAT-4 degradation, age-synchronized animals were transferred to E. coli OP50-seeded NGM plates supplemented with 10 µM auxin (5-Ph-IAA, Biozol, T8885) at the ages indicated in the corresponding figures and maintained on auxin-containing plates until the end of the experiment.

Construction of cat-4 C. elegans expression plasmids

To construct the C. elegans plasmid for ubiquitous somatic overexpression of cat-4, pPD95.77 from the Fire Lab kit (Addgene) was digested with SphI and XmaI to insert 3.6 kb of the sur-5 promoter. The resulting vector was then digested with KpnI and EcoRI to excise GFP and insert a multicloning site. F32G8.6 (cat-4) was PCR amplified from cDNA to include 5′ XbaI and 3′ KpnI restriction sites and subsequently cloned into the vector described above. To generate the neuron-specific overexpression cat-4 plasmid, the pMS1 plasmid (rgef-1p::rpn-6.1::unc-54 3′UTR) containing an inserted multicloning site¹⁰ was digested with 5′ KpnI and 3′ApaI restriction sites. Then, cat-4 was PCR-amplified from cDNA to include 5′ KpnI and 3′ ApaI restriction sites and cloned into pMS1. Using the same strategy, cat-4 was cloned into pMS2 (gly-19p::rpn-6.1::unc-54 3′ UTR)¹⁰ to generate intestine-specific expression plasmid. For the muscle-specific expression plasmid, the pMS3 plasmid (myo-3p::rpn-6.1::unc-54 3′ UTR)¹⁰ was digested with 5′ XbaI and 3′ AscI restriction sites. All constructs were sequence-verified.

Lifespan assay

Synchronized larvae obtained by an egg laying protocol were raised on OP50 E. coli at 20 °C until day 1 of adulthood. Once hermaphrodite worms reached adulthood, they were shifted to a given temperature (15 °C, 20 °C or 25 °C) on plates seeded with HT115 E. coli carrying either empty vector or RNAi constructs. When indicated in the corresponding figure legends, RNAi treatment was initiated from hatching. For lifespan experiments with mutant and AID worms, the animals were fed OP50 E. coli throughout adulthood. Adult worms (n = 96 per condition) were scored every day or every other day⁴⁹. From the initial population, worms that were lost, burrowed into the medium, or exhibited a protruding vulva or bagging phenotype were censored. We used GraphPad Prism (v.10.6.0) and OASIS (v.2)⁶⁵ to determine median and mean lifespan, respectively. The P values were calculated with GraphPad Prism (v.10.6.0) using the log-rank (Mantel–Cox) method and refer to comparisons between experimental and control animals within a single lifespan experiment. Each lifespan graph represents a single, representative experiment. Supplementary Table 7 contains the number of total/censored worms and detailed statistical analyses for each replicate lifespan experiment.

Motility assay

C. elegans were synchronized on OP50 E. coli using the egg-laying method and grown to day 1 of adulthood, then randomly transferred to plates seeded with OP50 and vehicle control (DMSO) or BH2. At the ages indicated in the corresponding figures, worms were randomly picked and placed in 20 μl of M9 buffer on a microscope slide. After 30 s of adaptation, the number of body bends (that is, change in direction of the bend at the mid-body) was counted for 30 s^27,66.

Brood size

Wild-type worms were maintained on standard NGM plates seeded with E. coli OP50 until the late L4 stage. Animals were then transferred to plates supplemented with 500 µM BH2 or DMSO vehicle control. Three worms were placed onto each experimental plate and transferred daily to freshly prepared treatment plates until egg laying ceased. The numbers of viable progeny and unhatched eggs were scored every 24 h. Brood size was calculated as the total number of progeny produced per worm throughout the reproductive period.

Bacterial colonization assay

Bacterial invasion assays were performed as previously described^21,22. HT115 E. coli cultures were grown overnight at 37 °C, induced with 1 mM IPTG for 4 h, collected and concentrated tenfold by centrifugation. Concentrated HT115 bacteria carrying the empty vector plasmid were mixed at a 4:1 ratio with concentrated HT115 bacteria expressing mCherry from the pDP151 plasmid. NGM agar plates were seeded with the bacterial mixture and allowed to dry at room temperature for 24 h. L4 larvae synchronized by an egg-laying protocol were randomly transferred onto the seeded plates and maintained under the indicated experimental conditions. Worms were analysed by fluorescence microscopy at day 1 and day 10 of adulthood. Before imaging, worms were gently collected from plates, washed three times with M9 buffer and transferred to OP50 E. coli plates for 2 h to eliminate residual fluorescent bacteria from the intestinal lumen. Worms were then immobilized in 0.1% sodium azide in M9 buffer on 2% agarose pads. Fluorescence images were acquired using the Zeiss Axio Zoom.V16 fluorescence microscope. For quantification of fluorescence signal, worms were outlined and quantified using ImageJ software (v.1.51s).

Chemotaxis assay

After development at 20 °C, synchronized adult worms were either kept at 20 °C or transferred to 15 °C until the ages specified in the corresponding figures. Chemotaxis tests were performed on freshly poured agar plates (2% agar, 5 mM KPO₄ at pH 6.0, 1 mM CaCl₂, 1 mM MgSO₄). The plates were sectioned into four quadrants, along with an inner circle measuring approximately 1 cm across diagonally. A benzaldehyde attractant (0.5% benzaldehyde in ethanol supplemented with 0.25 M sodium azide) was applied to one pair of diagonal quadrants, while the opposite quadrants received a control mixture (ethanol + 0.25 M sodium azide). On the specified adult days (as indicated in figure legends), worms were collected in S-Basal medium, washed three times to remove residual bacteria and deposited at the centre of the plate. After sealing with Parafilm, the plates were incubated at the indicated temperatures for 90 min. Worms were then counted in each quadrant, excluding those remaining within the central circle. The chemotaxis index was calculated using the following formula: (number of animal in attractant quadrants − number of animals in control quadrants)/total number of animals (ref. ⁶⁷).

Nose-touch assay

Age-synchronized worms were assessed for nose-touch response according to established protocols^28,50,68,69. In brief, the worms were placed onto a thin lawn of bacteria and allowed to move forward. A fine eyelash tool was gently positioned in the path of each moving worm. Animals that failed to reverse and instead continued crawling over or under the eyelash were scored as non-responsive. For each experimental group, 40 worms were tested, and their responses to 10 gentle eyelash touches were recorded.

BH2 and tyrosine supplementation

For BH2 supplementation, synchronized day 1 adult worms were collected and randomly distributed onto OP50 E. coli-seeded plates containing either BH2 (MedChemExpress, HY-W008646) or vehicle control (DMSO). For tyrosine supplementation, synchronized animals were placed on plates seeded with OP50 (or HT115 for RNAi experiments) and supplied with either tyrosine (Sigma-Aldrich, T8566) or the corresponding vehicle control (HCl). BH2 or tyrosine treatments were applied starting on day 1 of adulthood and continued until sample collection or maintained throughout adulthood for lifespan assays. Human cells were treated for 3 days with BH2 or tyrosine at the concentrations specified in the corresponding figures.

Human cell lines

HEK293T/17 cells (American Type Culture Collection (ATCC), CRL-11268) were plated onto 0.1% gelatin-coated plates and maintained in DMEM (Thermo Fisher Scientific, 11966025), supplemented with 10% FBS (Thermo Fisher Scientific, 10500064), 1% GlutaMAX (Life Technologies, 35050038) and 1% MEM non-essential amino acids (Thermo Fisher Scientific, 11140035) under standard conditions (37 °C, 5% CO₂). Isogenic control iPS cells (FUS^WT/WT) and ALS iPS cells (FUS^P525L/P525L) were provided by I. Bozzoni and A. Rosa. Both iPS cell lines were established and characterized for pluripotency in ref. ⁷⁰. In brief, control iPS cells were derived from a healthy donor and verified not to contain mutations in FUS. ALS iPS cells were generated from control iPS cells using transcription activator-like effector nuclease (TALEN)-directed mutagenesis and carry the homozygous FUS^P525L mutation associated with severe ALS⁷⁰. The identity of both iPS cell lines was authenticated in our laboratory by short-tandem-repeat profiling in a previous publication⁷¹. The iPS cell lines used in the present study were thawed directly from frozen stocks of these authenticated lines. iPS cells were cultured on Geltrex (Thermo Fisher Scientific, A1413302) in mTeSR1 medium (StemCell Technologies, 85850) at 37 °C with 5% CO₂. Undifferentiated iPS cell colonies were passaged using 2 mg ml⁻¹ dispase (StemCell Technologies, 07913), followed by mechanical dissociation with a glass pipette. All of the cell lines were routinely tested for mycoplasma contamination, and no contamination was detected.

Motor neuron differentiation

Motor neurons were derived from iPS cell lines according to a monolayer-based differentiation protocol⁷². iPS cells were seeded on Geltrex-coated plates and cultured in mTeSR1 medium until confluent. We then initiated the differentiation induction with neuron differentiation medium composed of DMEM/F12 and Neurobasal (1:1; Thermo Fisher Scientific, 11330057 and 21103049), supplemented with GlutaMAX (Thermo Fisher Scientific, 35050038), non-essential amino acids, B27 (Thermo Fisher Scientific, 12587010) and N2 (Thermo Fisher Scientific, 17502048). From day 0 to day 6, the medium was further supplemented with 1 μM retinoic acid (Sigma-Aldrich, R2625), 1 μM smoothened agonist (SAG; Sigma-Aldrich, 566661), 10 μM SB-431542 (Miltenyi Biotec, 130-105-336) and 0.1 μM LDN-193189 (Miltenyi Biotec, 130-103-925). From day 7 to day 14, neuron differentiation media were supplemented with 1 μM retinoic acid, 1 μM SAG, 4 μM SU-5402 (Sigma-Aldrich, SML0443) and 5 μM DAPT (Sigma-Aldrich, D5942). After day 14, differentiated motor neurons were dissociated and replated on poly-l-ornithine (Sigma-Aldrich, P3655) and laminin-coated (Thermo Fisher Scientific, 23017015) plates in Neurobasal medium, supplemented with non-essential amino acids, GlutaMAX, N2, B27 and neurotrophic factors (10 ng ml⁻¹ BDNF (BIOZOL, 450-02) and 10 ng ml⁻¹ GDNF (BIOZOL, 450-10)).

Transfection of HEK293 human cells

HEK293T/17 cells (ATCC, CRL-11268) were seeded onto plates coated with 0.1% gelatin. Once cultures reached approximately 40% confluence, the cells were transfected with 1 μg of one of the following plasmids using FuGENE HD (Promega), according to the manufacturer’s protocol: CMV:mRFP-Q74^33,34, pLVX-Puro-TDP-43-WT, pLVX-Puro-TDP-43-A382T, pcDNA3.1-FUS-HA-WT, or pcDNA3.1-FUS-HA-P525L. The cells were then treated with 10 µM BH2 or vehicle control (DMSO) and lysed after 72 h of incubation in standard growth medium. The FUS-HA plasmids were provided by D. Dormann⁷³, and the pLVX-Puro-TDP-43 vectors were provided by S. Ferguson (Addgene, 133753 and 133756)⁷⁴.

Lentiviral infection of HEK293 human cells

Lentivirus (LV) non-targeting short hairpin RNA (shRNA) and LV CCT2 shRNA (TRCN0000029499)³⁰ in the pLKO.1-puro backbone were obtained from Mission shRNA (Sigma-Aldrich). The target sequence of the CCT2 shRNA construct, previously validated in ref. ³⁰, is provided in Supplementary Table 8. To generate stable shRNA-expressing HEK293 cell lines, cells were transduced with 5 µl of concentrated lentivirus and selected with 2 µg ml⁻¹ puromycin (Thermo Fisher Scientific, A1113803).

Immunocytochemistry

Cells were fixed with 4% paraformaldehyde in PBS for 20 min, followed by permeabilization with 0.2% Triton X-100 in PBS for 10 min and blocking with 3% BSA in 0.2% Triton X-100 in PBS for 10 min. Cells were then incubated with anti-MAP2 (Sigma-Aldrich, M1406, 1:300) and rabbit anti-cleaved caspase-3 (Cell Signaling Technology, 9661S, 1:300) for 2 h at room temperature. After washing with PBS, the cells were incubated with secondary antibodies (Alexa Fluor 488 goat anti-mouse (Thermo Fisher Scientific, A-11029, 1:500) and Alexa Fluor 568 F(ab′)2 fragment of goat anti-rabbit IgG (H+L) (Thermo Fisher Scientific, A-21069, 1:500)) together with Hoechst 33342 (Life Technologies, 1656104) for 1 h at room temperature. The coverslips were then rinsed in PBS, followed by a distilled water wash, and mounted onto microscope slides using FluorSave Reagent (Merck, 345789).

Filter trap and western blotting

For filter trap and corresponding western blot assays, age-synchronized adult C. elegans were collected, washed with M9 buffer and snap-frozen in liquid nitrogen. The frozen pellets were thawed on ice and lysed in a non-denaturing buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 2 mM sodium orthovanadate, 1 mM PMSF, protease inhibitor cocktail (Roche)) using a Precellys 24 homogenizer. Lysates were cleared of worm debris by centrifugation (8,000 g, 5 min, 4 °C) and protein concentrations were measured using the BCA assay (Thermo Fisher Scientific). For western blot analysis, 30 μg of total protein was resolved by SDS–PAGE and transferred onto PVDF membranes (Millipore). For filter trap analysis, 100 μg of protein was supplemented with SDS to a final concentration of 0.5% SDS and loaded to a cellulose acetate membrane mounted in a slot-blot apparatus (Bio-Rad). After washing with 0.2% SDS, SDS-resistant aggregates were detected by immunoblotting. For immunoblotting analysis in C. elegans, including both filter trap and western blot assays, we used anti-IFB-2 (Developmental Studies Hybridoma Bank, MH33, 1:1,000) and anti-GFP (AMSBIO, TP401, 1:5,000) antibodies. Western blots were additionally probed using antibody against α-tubulin loading control (Sigma-Aldrich, T6199, 1:5,000).

For analysis of CAT-4 levels in AID strains, synchronized populations of C. elegans grown on E. coli OP50 were collected at late-L4/young adult stage after auxin treatment, washed three times with M9 buffer and snap-frozen in 1× SDS sample buffer supplemented with DTT. The samples were then thawed, heated at 90 °C for 5 min, dissociated using a Precellys homogenizer and centrifuged at 15,000 rpm for 15 min at 22 °C. The supernatants were collected and proteins were separated on 5–10% Bis-Tris gels before transfer to a nitrocellulose membrane (Amersham Protran, Cytiva). Membranes were blocked for 1 h at room temperature with 1× ROTI Block in TBS, and incubated for 1 h at room temperature with anti-Flag M2-HRP mouse (Sigma-Aldrich, A8592, 1:10,000) or anti-α-tubulin mouse (Sigma-Aldrich, T6074, 1:5,000). Membranes probed with anti-α-tubulin were incubated with goat anti-mouse IRDye 800CW secondary antibody (Li-Cor, 926-322101:10,000) for 1 h at room temperature. After three washes with TBS-T, membranes were imaged using a ChemiDoc Imaging System (Model MP, Bio-Rad).

For filter trap and western blot analyses of HEK293 cells, the cells were lysed in buffer containing 50 mM HEPES (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 2 mM sodium orthovanadate, 1 mM PMSF and protease inhibitors, followed by homogenization through a 27-gauge needle. Lysates from cells expressing mRFP-Q74 were cleared at 8,000g for 5 min at 4 °C. Lysates from cells expressing FUS-HA-WT, FUS-HA-P525L, pLVX-Puro-TDP-43-WT or pLVX-Puro-TDP-43-A382T were centrifuged at 1,000g for 5 min at 4 °C. Protein concentrations were determined by BCA assay. For western blotting, 30 μg of protein was analysed. For filter trap analyses, 100 μg of total protein was brought to 0.5% SDS and loaded onto cellulose acetate membranes as described above, followed by washing in 0.2% SDS and immunoblot detection of SDS-resistant aggregates. For immunoblotting in filter trap and western blot experiments, HEK293 lysates were probed using antibodies against mCherry (Abcam, ab167453, 1:5,000), FUS (Abcam, ab154141, CL0190, 1:1,000), TDP-43 (Abcam, ab225710, 1:1,000), anti-LC3B (Sigma-Aldrich, L7543, 1:1,000), anti-CCT2 (Abcam, ab92746, 1:1,000) and β-actin (Abcam, 8226, 1:5,000).

For necroptosis analysis, iPS-cell-derived motor neurons were lysed in RIPA buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate, 1 mM PMSF, 1 mM EDTA and protease inhibitor cocktail). Immunoblotting was performed with anti-phospho-RIP (Ser166) (Cell Signaling Technology, 65746, D1L3S, 1:1,000), anti-RIP (Cell Signaling Technology, 3493, D94C12, 1:1,000) and β-actin (Abcam, 8226, 1:5,000) antibodies.

Densitometry of filter trap and western blot experiments was performed using ImageJ (v.1.51s). For quantification of filter trap and western blot data, we present the results as relative changes compared with the corresponding control conditions. To average and statistically analyse independent experiments for these assays, we normalized test conditions to their corresponding control groups measured concurrently in each replicate experiment. Statistical analysis was performed using GraphPad Prism (v.10.6.0). Gel source data are provided in Supplementary Fig. 1.

MitoSOX staining

HEK293 cells were incubated with 1 μM MitoSOX Red superoxide indicator (Invitrogen, M36008) and Hoechst 33342 (Life Technologies, 1656104) in HBSS containing calcium and magnesium for 30 min at 37 °C in the dark. After incubation, cells were washed three times with HBSS containing calcium and magnesium, mounted onto microscope slides and imaged for subsequent analysis.

Proteasome activity

C. elegans were lysed in proteasome activity assay buffer (50 mM Tris-HCl, pH 7.5, 10% glycerol, 0.5 mM EDTA, 5 mM MgCl₂, 2 mM ATP and 1 mM dithiothreitol) using a Precellys 24 homogenizer (Bertin Technologies). Human cells were collected in the same buffer and lysed by passage through a 27-gauge needle attached to a 1 ml syringe 10 times. Lysates were centrifuged at 10,000g for 10 min at 4 °C. For each sample, 25 μg of total protein was transferred to a 96-well microplate (BD Falcon) and incubated with the fluorogenic trypsin-like proteasome substrate Ac-Arg-Leu-Arg-AMC (Enzo, BWL-AW9785-0005). Fluorescence generated by substrate cleavage (excitation, 380 nm; emission, 460 nm) was monitored using a microplate fluorometer (EnSpire, PerkinElmer) every 5 min for 2 h (worm lysates) or 1 h (human cell lysates). Proteasome activity was quantified as the slope of fluorescence accumulation over time. For comparisons across independent experiments, slope values were normalized to the corresponding control condition within each experiment before statistical analysis.

Reporting summary

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

1. Conti, B. & de Cabo, R. Promoting health and survival through lowered body temperature. Nat. Aging 5, 740–749 https://doi.org/10.1038/s43587-025-00850-0 (2025).

   Article  PubMed  PubMed Central  Google Scholar 

2. Lee, H. J. et al. Cold temperature extends longevity and prevents disease-related protein aggregation through PA28γ-induced proteasomes. Nat. Aging 3, 546–566 https://doi.org/10.1038/s43587-023-00383-4 (2023).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

3. Zhao, Z. J. et al. Body temperature is a more important modulator of lifespan than metabolic rate in two small mammals. Nat. Metab. 4, 320–326 https://doi.org/10.1038/s42255-022-00545-5 (2022).

   Article  CAS  PubMed  Google Scholar 

4. Conti, B. et al. Transgenic mice with a reduced core body temperature have an increased life span. Science 314, 825–828 https://doi.org/10.1126/science.1132191 (2006).

   Article  ADS  CAS  PubMed  Google Scholar 

5. Hosono, R., Mitsui, Y., Sato, Y., Aizawa, S. & Miwa, J. Life span of the wild and mutant nematode Caenorhabditis elegans. Effects of sex, sterilization, and temperature. Exp. Gerontol. 17, 163–172 (1982).

   Article  CAS  PubMed  Google Scholar 

6. Lamb, M. J. Temperature and lifespan in Drosophila. Nature 220, 808–809 (1968).

   Article  ADS  CAS  PubMed  Google Scholar 

7. Liu, R. K. & Walford, R. L. Increased growth and life-span with lowered ambient temperature in annual fish Cynolebias adloffi. Nature 212, 1277–1278 https://doi.org/10.1038/2121277a0 (1966).

   Article  ADS  Google Scholar 

8. Valenzano, D. R., Terzibasi, E., Cattaneo, A., Domenici, L. & Cellerino, A. Temperature affects longevity and age-related locomotor and cognitive decay in the short-lived fish Nothobranchius furzeri. Aging Cell 5, 275–278 https://doi.org/10.1111/j.1474-9726.2006.00212.x (2006).

   Article  CAS  PubMed  Google Scholar 

9. Lee, D. et al. MDT-15/MED15 permits longevity at low temperature via enhancing lipidostasis and proteostasis. PLoS Biol. 17, e3000415 https://doi.org/10.1371/journal.pbio.3000415 (2019).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

10. Lee, H. J. et al. Prostaglandin signals from adult germline stem cells delay somatic ageing of Caenorhabditis elegans. Nat. Metab. 1, 790–810 https://doi.org/10.1038/s42255-019-0097-9 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

11. Xiao, R. et al. A genetic program promotes C. elegans longevity at cold temperatures via a thermosensitive TRP channel. Cell 152, 806–817 https://doi.org/10.1016/j.cell.2013.01.020 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

12. Zhang, B. et al. Environmental temperature differentially modulates C. elegans longevity through a thermosensitive TRP channel. Cell Rep. 11, 1414–1424 https://doi.org/10.1016/j.celrep.2015.04.066 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

13. Protsiv, M., Ley, C., Lankester, J., Hastie, T. & Parsonnet, J. Decreasing human body temperature in the United States since the industrial revolution. eLife 9, e49555 https://doi.org/10.7554/eLife.49555 (2020).

14. Waalen, J. & Buxbaum, J. N. Is older colder or colder older? The association of age with body temperature in 18,630 individuals. J. Gerontol. A 66, 487–492 https://doi.org/10.1093/gerona/glr001 (2011).

    Article  Google Scholar 

15. Loeb, J. & Northrop, J. H. Is there a temperature coefficient for the duration of life? Proc. Natl Acad. Sci. USA 2, 456–457 (1916).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

16. Fanet, H., Capuron, L., Castanon, N., Calon, F. & Vancassel, S. Tetrahydrobioterin (BH4) pathway: from metabolism to neuropsychiatry. Curr. Neuropharmacol. 19, 591–609 https://doi.org/10.2174/1570159X18666200729103529 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

17. Werner, E. R., Blau, N. & Thony, B. Tetrahydrobiopterin: biochemistry and pathophysiology. Biochem. J. 438, 397–414 https://doi.org/10.1042/BJ20110293 (2011).

    Article  CAS  PubMed  Google Scholar 

18. Loer, C. M. et al. Cuticle integrity and biogenic amine synthesis in Caenorhabditis elegans require the cofactor tetrahydrobiopterin (BH4). Genetics 200, 237–253 https://doi.org/10.1534/genetics.114.174110 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

19. Eichwald, T. et al. Tetrahydrobiopterin: beyond its traditional role as a cofactor. Antioxidants 12, 1037 https://doi.org/10.3390/antiox12051037 (2023).

20. Ohashi, A. et al. Membrane transport of sepiapterin and dihydrobiopterin by equilibrative nucleoside transporters: a plausible gateway for the salvage pathway of tetrahydrobiopterin biosynthesis. Mol. Genet. Metab. 102, 18–28 https://doi.org/10.1016/j.ymgme.2010.09.005 (2011).

    Article  CAS  PubMed  Google Scholar 

21. Egge, N. et al. Age-onset phosphorylation of a minor actin variant promotes intestinal barrier dysfunction. Dev. Cell 51, 587–601 https://doi.org/10.1016/j.devcel.2019.11.001 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

22. Koyuncu, S. et al. Rewiring of the ubiquitinated proteome determines ageing in C. elegans. Nature 596, 285–290 https://doi.org/10.1038/s41586-021-03781-z (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

23. Dillin, A. et al. Rates of behavior and aging specified by mitochondrial function during development. Science 298, 2398–2401 https://doi.org/10.1126/science.1077780 (2002).

    Article  ADS  CAS  PubMed  Google Scholar 

24. Durieux, J., Wolff, S. & Dillin, A. The cell-non-autonomous nature of electron transport chain-mediated longevity. Cell 144, 79–91 https://doi.org/10.1016/j.cell.2010.12.016 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

25. Moran, G. R. 4-Hydroxyphenylpyruvate dioxygenase. Arch. Biochem. Biophys. 433, 117–128 https://doi.org/10.1016/j.abb.2004.08.015 (2005).

    Article  CAS  PubMed  Google Scholar 

26. Hommen, F., Bilican, S. & Vilchez, D. Protein clearance strategies for disease intervention. J. Neural Transm. https://doi.org/10.1007/s00702-021-02431-y (2021).

27. Brignull, H. R., Moore, F. E., Tang, S. J. & Morimoto, R. I. Polyglutamine proteins at the pathogenic threshold display neuron-specific aggregation in a pan-neuronal Caenorhabditis elegans model. J. Neurosci. 26, 7597–7606 https://doi.org/10.1523/JNEUROSCI.0990-06.2006 (2006).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

28. Koyuncu, S. et al. The aging factor EPS8 induces disease-related protein aggregation through RAC signaling hyperactivation. Nat. Aging https://doi.org/10.1038/s43587-025-00943-w (2025).

29. Kitamura, A. et al. Cytosolic chaperonin prevents polyglutamine toxicity with altering the aggregation state. Nat. Cell Biol. 8, 1163–1170 https://doi.org/10.1038/ncb1478 (2006).

    Article  CAS  PubMed  Google Scholar 

30. Noormohammadi, A. et al. Somatic increase of CCT8 mimics proteostasis of human pluripotent stem cells and extends C. elegans lifespan. Nat. Commun. 7, 13649 https://doi.org/10.1038/ncomms13649 (2016).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

31. Tam, S., Geller, R., Spiess, C. & Frydman, J. The chaperonin TRiC controls polyglutamine aggregation and toxicity through subunit-specific interactions. Nat. Cell Biol. 8, 1155–1162 https://doi.org/10.1038/ncb1477 (2006).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

32. Morimoto, R. I. The heat shock response: systems biology of proteotoxic stress in aging and disease. Cold Spring Harb. Symp. Quant. Biol. 76, 91–99 https://doi.org/10.1101/sqb.2012.76.010637 (2011).

    Article  CAS  PubMed  Google Scholar 

33. Balaji, V. et al. A dimer-monomer switch controls CHIP-dependent substrate ubiquitylation and processing. Mol. Cell 82, 3239–3254 https://doi.org/10.1016/j.molcel.2022.08.003 (2022).

    Article  CAS  PubMed  Google Scholar 

34. Llamas, E. et al. In planta expression of human polyQ-expanded huntingtin fragment reveals mechanisms to prevent disease-related protein aggregation. Nat. Aging https://doi.org/10.1038/s43587-023-00502-1 (2023).

35. Alirzayeva, H. et al. ALS-FUS mutations cause abnormal PARylation and histone H1.2 interaction, leading to pathological changes. Cell Rep. 43, 114626 https://doi.org/10.1016/j.celrep.2024.114626 (2024).

    Article  CAS  PubMed  Google Scholar 

36. Crabtree, M. J., Hale, A. B. & Channon, K. M. Dihydrofolate reductase protects endothelial nitric oxide synthase from uncoupling in tetrahydrobiopterin deficiency. Free Radic. Biol. Med. 50, 1639–1646 https://doi.org/10.1016/j.freeradbiomed.2011.03.010 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

37. Crabtree, M. J., Smith, C. L., Lam, G., Goligorsky, M. S. & Gross, S. S. Ratio of 5,6,7,8-tetrahydrobiopterin to 7,8-dihydrobiopterin in endothelial cells determines glucose-elicited changes in NO vs. superoxide production by eNOS. Am. J. Physiol. Heart Circ. Physiol. 294, H1530–H1540 https://doi.org/10.1152/ajpheart.00823.2007 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

38. Janaszak-Jasiecka, A., Ploska, A., Wieronska, J. M., Dobrucki, L. W. & Kalinowski, L. Endothelial dysfunction due to eNOS uncoupling: molecular mechanisms as potential therapeutic targets. Cell. Mol. Biol. Lett. 28, 21 https://doi.org/10.1186/s11658-023-00423-2 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

39. Brenner, S. The genetics of Caenorhabditis elegans. Genetics 77, 71–94 (1974).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

40. Husken, K. et al. Maintenance of the intestinal tube in Caenorhabditis elegans: the role of the intermediate filament protein IFC-2. Differentiation 76, 881–896 https://doi.org/10.1111/j.1432-0436.2008.00264.x (2008).

    Article  CAS  PubMed  Google Scholar 

41. Kumsta, C. et al. The autophagy receptor p62/SQST-1 promotes proteostasis and longevity in C. elegans by inducing autophagy. Nat. Commun. 10, 5648 https://doi.org/10.1038/s41467-019-13540-4 (2019).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

42. Yigit, E. et al. Analysis of the C. elegans Argonaute family reveals that distinct Argonautes act sequentially during RNAi. Cell 127, 747–757 https://doi.org/10.1016/j.cell.2006.09.033 (2006).

    Article  ADS  CAS  PubMed  Google Scholar 

43. Watts, J. S. et al. New strains for tissue-specific RNAi studies in Caenorhabditis elegans. G3 10, 4167–4176 https://doi.org/10.1534/g3.120.401749 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

44. Qadota, H. et al. Establishment of a tissue-specific RNAi system in C. elegans. Gene 400, 166–173 https://doi.org/10.1016/j.gene.2007.06.020 (2007).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

45. Yang, Y. et al. Autophagy protein ATG-16.2 and its WD40 domain mediate the beneficial effects of inhibiting early-acting autophagy genes in C. elegans neurons. Nat. Aging 4, 198–212 https://doi.org/10.1038/s43587-023-00548-1 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

46. Espelt, M. V., Estevez, A. Y., Yin, X. & Strange, K. Oscillatory Ca²⁺ signaling in the isolated Caenorhabditis elegans intestine: role of the inositol-1,4,5-trisphosphate receptor and phospholipases C β and γ. J. Gen. Physiol. 126, 379–392 https://doi.org/10.1085/jgp.200509355 (2005).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

47. Noormohammadi, A. et al. Mechanisms of protein homeostasis (proteostasis) maintain stem cell identity in mammalian pluripotent stem cells. Cell. Mol. Life Sci. 75, 275–290 https://doi.org/10.1007/s00018-017-2602-1 (2018).

    Article  CAS  PubMed  Google Scholar 

48. Ghanta, K. S. & Mello, C. C. Melting dsDNA donor molecules greatly improves precision genome editing in Caenorhabditis elegans. Genetics 216, 643–650 https://doi.org/10.1534/genetics.120.303564 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

49. Amrit, F. R., Ratnappan, R., Keith, S. A. & Ghazi, A. The C. elegans lifespan assay toolkit. Methods 68, 465–475 https://doi.org/10.1016/j.ymeth.2014.04.002 (2014).

    Article  CAS  PubMed  Google Scholar 

50. Fernandez-Abascal, J. et al. A glial ClC Cl⁻ channel mediates nose touch responses in C. elegans. Neuron 110, 470–485 https://doi.org/10.1016/j.neuron.2021.11.010 (2022).

    Article  CAS  PubMed  Google Scholar 

51. Koopman, M., Gungordu, L., Seinstra, R. I. & Nollen, E. A. A. Neuronal overexpression of hTDP-43 in Caenorhabditis elegans impairs different neuronally controlled behaviors and decreases fecundity. Micropubl. Biol. https://doi.org/10.17912/micropub.biology.000769 (2023).

52. Segref, A. et al. Thermosensation in Caenorhabditis elegans is linked to ubiquitin-dependent protein turnover via insulin and calcineurin signalling. Nat. Commun. 13, 5874 https://doi.org/10.1038/s41467-022-33467-7 (2022).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

53. Porta-de-la-Riva, M., Fontrodona, L., Villanueva, A. & Ceron, J. Basic Caenorhabditis elegans methods: synchronization and observation. J. Vis. Exp. 64, e4019 https://doi.org/10.3791/4019 (2012).

54. Dieterle, F., Ross, A., Schlotterbeck, G. & Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in H-1 NMR metabonomics. Anal. Chem. 78, 4281–4290 https://doi.org/10.1021/ac051632c (2006).

    Article  ADS  CAS  PubMed  Google Scholar 

55. Schmidt, C. et al. Integrated metabolomics data analysis to generate mechanistic hypotheses with MetaProViz. Mol. Syst. Biol. https://doi.org/10.1038/s44320-026-00231-8 (2026).

56. Pollard, K. S., Dudoit, S. & van der Laan, M. J. Multiple testing procedures: the multtest package and applications to genomics. In Bioinformatics and Computational Biology Solutions Using R and Bioconductor. Statistics for Biology and Health (eds Gentleman, R. et al.) 249–271 (Springer, 2005).

57. Searle, B. C. et al. Generating high quality libraries for DIA MS with empirically corrected peptide predictions. Nat. Commun. 11, 1548 https://doi.org/10.1038/s41467-020-15346-1 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

58. Chambers, M. C. et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat. Biotechnol. 30, 918–920 https://doi.org/10.1038/nbt.2377 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

59. Demichev, V., Messner, C. B., Vernardis, S. I., Lilley, K. S. & Ralser, M. DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. Nat. Methods 17, 41–44 https://doi.org/10.1038/s41592-019-0638-x (2020).

    Article  CAS  PubMed  Google Scholar 

60. Tyanova, S. et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat. Methods 13, 731–740 https://doi.org/10.1038/nmeth.3901 (2016).

    Article  CAS  PubMed  Google Scholar 

61. Hoogewijs, D., Houthoofd, K., Matthijssens, F., Vandesompele, J. & Vanfleteren, J. R. Selection and validation of a set of reliable reference genes for quantitative sod gene expression analysis in C. elegans. BMC Mol. Biol. 9, 9 https://doi.org/10.1186/1471-2199-9-9 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

62. Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using real-time quantitative PCR and the \({2}^{\varDelta \varDelta {C}_{{\rm{T}}}}\) method. Methods 25, 402–408 https://doi.org/10.1006/meth.2001.1262 (2001).

    Article  CAS  PubMed  Google Scholar 

63. Dillin, A., Crawford, D. K. & Kenyon, C. Timing requirements for insulin/IGF-1 signaling in C. elegans. Science 298, 830–834 https://doi.org/10.1126/science.1074240 (2002).

    Article  ADS  CAS  PubMed  Google Scholar 

64. Artan, M., Schoen, H. & de Bono, M. Proximity labeling of DAF-16 FOXO highlights aging regulatory proteins. Nat. Commun. 16, 11355 https://doi.org/10.1038/s41467-025-66409-0 (2025).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

65. Han, S. K. et al. OASIS 2: online application for survival analysis 2 with features for the analysis of maximal lifespan and healthspan in aging research. Oncotarget 7, 56147–56152 https://doi.org/10.18632/oncotarget.11269 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

66. Calculli, G. et al. Systemic regulation of mitochondria by germline proteostasis prevents protein aggregation in the soma of C. elegans. Sci. Adv. 7, 26 https://doi.org/10.1126/sciadv.abg3012 (2021).

67. Ward, S. Chemotaxis by the nematode Caenorhabditis elegans: identification of attractants and analysis of the response by use of mutants. Proc. Natl Acad. Sci. USA 70, 817–821 https://doi.org/10.1073/pnas.70.3.817 (1973).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

68. Chalfie, M., Hart, A. C., Rankin, C. H. & Goodman, M. B. Assaying mechanosensation. Wormbook https://doi.org/10.1895/wormbook.1.172.1 (2014).

69. Hart, A. C., Kass, J., Shapiro, J. E. & Kaplan, J. M. Distinct signaling pathways mediate touch and osmosensory responses in a polymodal sensory neuron. J. Neurosci. 19, 1952–1958 https://doi.org/10.1523/JNEUROSCI.19-06-01952.1999 (1999).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

70. Lenzi, J. et al. ALS mutant FUS proteins are recruited into stress granules in induced pluripotent stem cell-derived motoneurons. Dis. Model. Mech. 8, 755–766 https://doi.org/10.1242/dmm.020099 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

71. Koyuncu, S. et al. The ubiquitin ligase UBR5 suppresses proteostasis collapse in pluripotent stem cells from Huntington’s disease patients. Nat. Commun. 9, 2886 https://doi.org/10.1038/s41467-018-05320-3 (2018).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

72. Hill, S. J. et al. Two familial ALS proteins function in prevention/repair of transcription-associated DNA damage. Proc. Natl Acad. Sci. USA 113, E7701–E7709 https://doi.org/10.1073/pnas.1611673113 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

73. Dormann, D. et al. ALS-associated fused in sarcoma (FUS) mutations disrupt transportin-mediated nuclear import. EMBO J. 29, 2841–2857 https://doi.org/10.1038/emboj.2010.143 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

74. Roczniak-Ferguson, A. & Ferguson, S. M. Pleiotropic requirements for human TDP-43 in the regulation of cell and organelle homeostasis. Life Sci. Alliance 2, e201900358 https://doi.org/10.26508/lsa.201900358 (2019).

75. Sud, M. et al. Metabolomics Workbench: an international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools. Nucleic Acids Res. 44, D463–D470 https://doi.org/10.1093/nar/gkv1042 (2016).

    Article  CAS  PubMed  Google Scholar 

Download references

Authors and Affiliations

1. Institute for Integrated Stress Response Signaling, Faculty of Medicine, University Hospital Cologne, Cologne, Germany

   Hyun Ju Lee, Cansu Doğan, Seda Koyuncu, Markus Wehrmann, Dunja Petrovic, Rute Loureiro, Amit Kumar Sahu, Joana Schnubel, Pavel Grobushkin & David Vilchez

2. Cologne Excellence Cluster for Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany

   Hyun Ju Lee, Cansu Doğan, Seda Koyuncu, Markus Wehrmann, Murat Artan, Dunja Petrovic, Rute Loureiro, Amit Kumar Sahu, Joana Schnubel, Pavel Grobushkin, Dimitrios Prymidis, Theodoros Georgomanolis, Agnieszka Sokol, Ming Yang, Thorsten Hoppe, Christian Frezza & David Vilchez

3. Institute for Genetics, Faculty of Mathematics and Natural Sciences, University of Cologne, Cologne, Germany

   Murat Artan, Thorsten Hoppe, Christian Frezza & David Vilchez

4. Institute for Metabolomics in Ageing, Faculty of Medicine, University Hospital Cologne, Cologne, Germany

   Dimitrios Prymidis, Theodoros Georgomanolis, Agnieszka Sokol, Ming Yang & Christian Frezza

5. Center for Molecular Medicine Cologne (CMMC), University of Cologne, Cologne, Germany

   Christian Frezza & David Vilchez

Contributions

H.J.L. and D.V. supervised the study. D.V. conceived the study and designed the experiments. H.J.L. and C.D. performed most of the experiments in C. elegans. S.K. carried out most of the experiments in C. elegans and human cell disease models. H.J.L., C.D., S.K. and D.V. performed data analysis and interpretation. M.W. performed proteomics and one of the metabolomics experiments defining cold-induced metabolites in day 5 adult worms (experiment 1). M.A. generated the cat-4::SL2::GFP and AID strains. D. Petrovic contributed to lifespan and other experiments. R.L. carried out the metabolomics experiment in wild-type and daf-2-mutant worms at different ages. A.K.S. performed bioinformatic analyses, including comparisons of effect sizes and assessment of reproducibility across independent metabolomics datasets. J.S. performed some of the replicate lifespan experiments. P.G. contributed to lifespan experiments and plasmid construction. D. Prymidis, T.G., A.S. and M.Y. contributed to metabolomics experiments and analysis. T.H. provided expertise in proteostasis. C.F. provided critical expertise in metabolomics and advice on other aspects of the study. D.V. wrote the manuscript. All of the authors revised and approved the paper.

Corresponding authors

Correspondence to Hyun Ju Lee or David Vilchez.

Competing interests

H.J.L. and D.V. are listed as co-inventors on a pending patent application submitted by Ernesto Llamas Pamanes (application no. LU606057), covering the use of BH4, its precursors and downstream metabolites for the prevention and treatment of motor dysfunction. The other authors declare no competing interests.

Peer review information

Nature thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Reporting Summary (download PDF )

Peer Review File (download PDF )

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions