Ageing is a major risk factor for chronic morbidities such as cancers, cardiovascular disorders, and neurodegeneration. These diseases contribute to a rising burden on families, communities, and healthcare across the world. Environmental and genetic factors contribute to ageing by influencing a complex network of pathways and processes that drive cellular dysfunction 1. These include the loss of protein homeostasis (proteostasis), which is characterised by the appearance and aggregation of misfolded and mislocalised proteins within cells and tissues 2.
Cells possess an array of protein quality control mechanisms collectively referred to as the proteostasis network (PN), which act to preserve proteome integrity. The PN coordinates protein synthesis, folding, disaggregation and degradation and integrates components of the translational machinery, molecular chaperones and co-chaperones and the proteolytic systems - the ubiquitin–proteasome system (UPS), and autophagy-lysosomal system - to ensure cell viability 3.
The cytosolic/nuclear arm of the PN is subject to regulation by heat shock transcription factor 1 (HSF-1), which protects the proteome by driving the expression of heat shock proteins (HSPs) that function as molecular chaperones 3,4. In line with its function, knock down of HSF-1 leads to increased protein aggregation, tissue dysfunction and decreased survival, whereas overexpression of HSF-1 maintains proteome integrity, promotes tissue health, and extends lifespan. While it is apparent that increasing HSF-1 activity is beneficial for longevity, our understanding of the mechanisms that act downstream of HSF-1 to prolong healthy tissue function, remains limited.
It is widely believed that HSF-1 regulates ageing by upregulating the expression of HSPs. However, in addition to HSPs, HSF-1 also controls the expression of genes encoding cytoskeletal components, metabolic enzymes, ribosomal subunits, chromatin factors and components of the UPS 5,6. Moreover, recent work has demonstrated roles for autophagy 7, maintenance of the cytoskeleton and lipid regulation 8,9 in HSF-1 mediated lifespan extension. These observations indicate that HSF-1 regulates longevity through mechanisms beyond HSP mediated chaperoning of the proteome.
Here, we employed an RNAi screen to identify the HSF-1 target genes that promote increased lifespan in C. elegans over expressing HSF-1 (hsf-1 OE). We have found that the sole worm ubiquilin, ubiquilin-1 (ubql-1), is necessary for hsf-1 OE to increase lifespan. Ubiquilins are multifaceted, conserved shuttle proteins that localise to the cytoplasm and nucleus 10 where they function as chaperones that aid in the degradation of substrates through the ubiquitin-proteasome system and autophagy. This is facilitated by their N-terminal Ubiquitin (Ub)-like (UBL) domain and C-terminal Ub-associated (UBA) domain, which enable binding to the proteasome and polyubiquitinated chains 11.
Despite its central role in protein degradation, we find that ubiquilin-1 does not promote longevity by altering general proteostasis capacity. Instead, ubiquilin-1 increases lifespan upon overexpression of HSF-1 by promoting a reduction in organellar protein degradation, mitochondrial network remodelling and metabolic rewiring.
Ubiquilin-1 is required for HSF-1 mediated lifespan extension
Overexpression of HSF-1 (hsf-1 OE) leads to extension of lifespan in C. elegans 9,12,13. To better understand the mechanisms that act downstream of HSF-1 to promote longevity, an RNAi screen was performed to determine which HSF-1 target genes are required for the increased lifespan of hsf-1 OE worms. Our RNAi screen consisted of 96 C. elegans genes shown to be directly regulated by HSF-1 under basal or stress conditions (Supplementary Fig. 1a and Supplementary Table 1) 14. We identified the gene ubiquilin-1 (ubql-1) as the strongest modifier of hsf-1 OE lifespan, without comparable effects on the lifespan of wildtype worms (Fig. 1a). Consistent with previous reports that ubql-1 is bound by HSF-1 (Supplementary Fig. 1b), hsf-1 OE worms exhibited increased ubql-1 expression in early adulthood (Fig. 1b).
To verify our RNAi screen, we grew animals on empty vector control (L4440) or ubql-1(RNAi) and measured survival in two independent hsf-1 OE lines. Knockdown of ubql-1 suppressed the increased lifespan of both hsf-1 OE lines tested (Fig. 1c and Supplementary Fig. 1d). In addition, we also assessed lifespan in ubql-1(tm1574) mutants that harbour a 755bp deletion spanning the whole of exons 1 and 2, and express reduced levels of a truncated ubql-1 mRNA (Fig. 1b, Supplementary Fig. 1c and Supplementary Fig. 1e). Lifespan assays revealed that ubql-1(tm1574) mutants are shorter lived than wildtype controls and that the presence of the ubql-1(tm1574) mutation reduces the lifespan of hsf-1 OE worms to that of wildtype animals (Fig. 1d). Together our data establish UBQL-1 as a mediator of HSF-1 mediated longevity extension in C. elegans. Therefore, we sought to investigate the processes through which UBQL-1 impacts longevity.
Maintenance of proteostasis capacity does not require ubql-1 function
To better understand the mechanisms by which ubql-1 functions to supress ageing downstream of HSF-1, we first investigated whether ubiquilin-1 influences HSF-1 activity. Loss of ubql-1 function did not suppress the expression of canonical HSF-1 target genes (hsp-16.11 or hsp-70) basally, or in response to heat shock, in wildtype or hsf-1 OE worms (Fig. 2a-b and Supplementary Fig. 2a-b), demonstrating that HSF-1 activity is unaffected by the loss of ubql-1 activity.
Next, we asked whether ubql-1 is necessary for worms to manage heat induced protein folding stress. As expected, hsf-1 OE worms survived for longer than wildtype worms following heat stress as young adults (Fig. 2c). Despite being shorter lived, ubql-1 mutants were more stress resistant than wildtype worms; however, stress resistance was not further increased in ubql-1 mutants upon overexpression of HSF-1 (Fig. 2c). Given that ubql-1 mutants do not exhibit increased HSF-1 activity, these data suggest that ubql-1 is necessary for increased stress resistance in hsf-1 OE worms.
Many factors and pathways can contribute to stress resistance. Therefore, to more precisely examine the effects of ubql-1 on proteostasis capacity, we took advantage of well-described polyglutamine::YFP-based (PolyQ::YFP) proteostasis sensors 15,16 expressed exclusively in the intestine (Q44) or body wall muscles (Q35). PolyQ aggregation increased with age in both intestinal and muscle tissues (Fig. 2d-f and Supplementary Fig. 2c-e) of wildtype worms. PolyQ aggregation was strongly suppressed on day 3 and day 5 of adulthood in worms overexpressing HSF-1 (Fig. 2d-f and Supplementary Fig. 2c-e). Surprisingly, ubql-1(RNAi) did not alter the aggregation of polyQ proteins in the intestine or muscles (Fig. 2d-f and Supplementary Fig. 2c-e) and did not prevent hsf-1 OE from suppressing polyQ aggregation in these tissues compared to wildtype counterparts (Fig. 2d-f and Supplementary Fig. 2c-e). Taken together, these data indicate that UBQL-1 does not influence longevity in hsf-1 OE worms by broadly promoting proteostasis capacity.
Ubiquilin-1 promotes metabolic remodelling in hsf-1 OE animals
To ascertain how ubql-1 promotes longevity, we employed transcriptomics and proteomics to identify genes, proteins and pathways that are altered in hsf-1 OE worms in a UBQL-1 dependent manner. Increased HSF-1 activity generated substantial changes across both the proteome and transcriptome with a total of 1,564 proteins (948 increased and 616 decreased) and 3,262 transcripts (2,343 increased and 919 decreased) altered compared to wildtype worms (Fig. 3a, Supplementary Fig. 3a, Supplementary Table 3 and Supplementary Table 4). While we did not observe a strong overlap in the specific identities of altered proteins and genes across our proteomic and transcriptomic datasets (Supplementary Fig. 3b), there was a good functional correlation between the two datasets, with KEGG analysis 17 revealing that up-regulated genes and proteins were enriched for pathways regulating metabolism and protein processing in the endoplasmic reticulum (ER), while down-regulated genes and proteins were enriched for pathways that included mismatch repair, DNA replication and signalling pathways (Fig. 3b and Supplementary Fig. 3c).
Loss of ubql-1 function in wildtype worms resulted in altered abundance (FDR p < 0.05) of 68 proteins, of which, 43 were increased, and 25 were decreased (Supplementary Fig. 3d, Supplementary Table 3). Similarly, RNA-seq analysis revealed that ubql-1(tm1574) mutants exhibited 402 up-regulated and 16 down-regulated (Log 2-FC, FDR p < 0.05) genes compared to wildtype worms (Supplementary Fig. 3e, Supplementary Table 4). Of the proteomic and transcriptomic changes observed in hsf-1 OE worms, 12 proteins and 61 genes were increased in a ubql-1 dependent manner, and 4 proteins and 17 genes were down-regulated in a ubql-1 dependent manner (Fig. 3c-f and Supplementary Fig. 3f). Interestingly, many ubql-1 dependent genes had roles in lipid metabolism (Fig. 3f) and UBQL-1 has been shown to interact strongly with the central metabolic regulator, NHR-49 18.
Among the proteins whose levels were reduced in hsf-1 OE worms lacking a fully functional UBQL-1, was the acyl-CoA synthetase-2 (ACS-2), a key enzyme in mitochondrial beta-fatty acid oxidation that is known to localise to mitochondria 19–21 (Fig. 3d). ACS-2 levels are controlled by NHR-49, which has been shown to be necessary for lipid homeostasis and increased lifespan in hsf-1 OE worms 8,19–23. Furthermore, five of the other 11 proteins regulated in hsf-1 OE animals by UBQL-1 (UMPS-1, PGP-6, IRG-1, SEU-1, and CRML-1) have been shown to associate with NHR-49 (Fig. 3d and Supplementary Fig. 3g), with only irg-1 and pgp-6 exhibiting concomitant changes in transcription (Fig. 3f). Moreover, exposure to nhr-49(RNAi) shortened lifespan across all backgrounds tested, with the loss of either NHR-49 or UBQL-1 suppressing hsf-1 OE lifespan to a similar extent, with no additive effect observed (Supplementary Fig. 3h and i). Taken together, our data suggest that the increased lifespan of hsf-1 OE worms is mediated by UBQL-1 dependent metabolic remodelling, at least in part, through NHR-49 activity, possibly by altering the stability of NHR-49 complexes.
Ubiquilin-1 promotes down-tuning of endoplasmic reticulum and mitochondrial associated degradation components in hsf-1 OE animals
Ubiquilins have a central role in protein degradation pathways associated with the cytosol/nucleus, endoplasmic reticulum (ER) and mitochondria 11,24–26. Therefore, we reasoned that UBQL-1 promotes metabolic remodelling and lifespan extension in hsf-1 OE animals by promoting the degradation of key target proteins. Therefore, we focused our attention on proteins whose levels were elevated in hsf-1 OE animals upon reduction of UBQL-1 function.
Among the proteins whose levels are decreased upon hsf-1 OE, we identified 4 whose reduction was dependent on UBQL-1. Among these, the protein NPL-4.1 displayed the strongest increase in abundance in ubql-1 mutant animals (Fig. 3d). NPL-4.1 is a central component of the CDC-48-NPL-4-UFD complex, which is at the core of both ER associated protein degradation (ERAD) and mitochondria associated protein degradation (MAD)27, suggesting that the reduced activity of these pathways may be linked to lifespan extension in hsf-1 OE worms. Consistent with this, we also observed a strong reduction in npl-4.1, npl-4.2, ufd-1, cdc-48.1 and cdc-48.2 mRNA levels in hsf-1 OE worms compared to wild type animals, although transcript levels were not restored in ubql-1 mutants (Supplementary Fig. 4a). Consistent with a reduction in ERAD, we observed that “protein processing in the ER” was the strongest enriched term among our set of increased proteins in hsf-1 OE worms (Fig. 3b) and that hsf-1 OE worms were highly sensitive to ER stress induced by tunicamycin (Supplementary Fig. 4b). While we did not observe evidence for activation of the UPRmt (Supplementary Fig. 4d-e), we did find that, consistent with a reduction in MAD, mitochondrial network organisation was dramatically altered upon hsf-1 OE 28 (Supplementary Fig. 4c). Together, these observations suggest that a mild impairment of organellar protein degradation is associated with the increased lifespan of hsf-1 OE worms.
To determine how gross defects in ERAD or MAD impact lifespan in hsf-1 OE or wild type worms, we used RNAi to knock-down the E3 ligase, SEL-11/HRD-1 (which is necessary for ERAD), or individual components of the CDC-48-NPL-4-UFD-1 complex. Knockdown of sel-11 reduced lifespan to a comparable extent in both wildtype and hsf-1 OE worms, suggesting that ERAD is required for normal lifespan but does not interact with hsf-1 OE to influence lifespan (Fig. 4a). In contrast, while knockdown of npl-4.1/4.2, ufd-1 or cdc-48.1 was detrimental to all genotypes tested, knockdown of any of these factors inverted the lifespans of wildtype and hsf-1 OE worms (Fig. 4b-d). While these effects were not suppressed by loss of UBQL-1 function, this is likely because UBQL-1 activity becomes critical for the removal of toxic proteins that build up upon severe loss of the CDC-48-UFD-1-NPL-4 complex.
Together, these data suggest that increased HSF-1 expression compromises organellar protein degradation pathways, leading to a mitochondrial network adaptation that results in a more fused mitochondrial network and altered metabolic homeostasis.
HSF-1 overexpression promotes lifespan by altering mitochondrial dynamics
To shed light on whether UBQL-1 is necessary for changes in mitochondrial function and network dynamics upon hsf-1 OE, we first assessed respiration rates, total ATP-levels and lipid stores (measured by Oil-Red-O staining 29) across our four genotypes. Overexpression of hsf-1 reduced oxygen consumption rates (OCR) and increased mitochondrial fusion in a ubql-1 dependent manner but did not alter ATP levels or total trigylceride levels (Fig. 5a-c and Supplementary Fig. 5a-f). However, we did observe a comparable reduction in lipid stores upon loss of UBQL-1 function in wildtype and hsf-1 OE worms (Supplementary Fig. 5b and c), suggesting a previously undiscovered role for ubiquilins in lipid homeostasis.
Mitochondrial metabolic capacity is linked to network morphology 30 and NHR-49 is known to be important for mitochondrial dynamics 23. Maintenance of mitochondrial networks is controlled via fission and fusion 31,32, therefore, to test whether mitochondrial network dynamics are important for the extended lifespan of hsf-1 OE worms, we measured survival in worms in which core fusion/fission mediators were targeted by RNAi, specifically, the cytosolic dynamin-like GTPase DRP-1 33 and the membrane-anchored dynamin-like GTPases FZO-1/mitofusin 34 and EAT-3/OPA1 35.
RNAi against fzo-1 or eat-3 extended lifespan in wildtype worms to a level comparable to that seen upon hsf-1 OE (Fig. 5d and e). However, no additive effect on lifespan was observed in hsf-1 OE; fzo-1(RNAi) or eat-3(RNAi) groups (Fig. 5d and e). In contrast, drp-1(RNAi) only modestly increased lifespan in wildtype animals and strongly suppressed lifespan extension upon hsf-1 OE (Fig. 5f). Given that mitochondrial fission contributes to mitochondrial homeostasis by segregating damaged mitochondria for degradation via mitophagy 36, we investigated whether mitophagy is also necessary for the elevated lifespan of hsf-1 OE worms. Knockdown of the kinase PINK-1 (a core mitophagy inducer) did not suppress lifespan extension in hsf-1 OE worms (Supplementary Fig. 5g) and mitochondrial copy number was unaltered by hsf-1 OE or ubql-1 mutation (Supplementary Fig. 5h). Together, our data show that in response to hsf-1 overexpression, UBQL-1-mediated changes in mitochondrial network dynamics promote longevity independently of mitophagy or UPRmt activation. Furthermore, our findings reveal that the longevity promoting effects conferred by hsf-1 OE stem from mitochondrial network adaptations in response to reduced organellar protein degradation.