Rewiring of the ubiquitinated proteome determines ageing in C. elegans

Ageing is driven by a loss of cellular integrity1. Given the major role of ubiquitin modifications in cell function2, here we assess the link between ubiquitination and ageing by quantifying whole-proteome ubiquitin signatures in Caenorhabditis elegans. We find a remodelling of the ubiquitinated proteome during ageing, which is ameliorated by longevity paradigms such as dietary restriction and reduced insulin signalling. Notably, ageing causes a global loss of ubiquitination that is triggered by increased deubiquitinase activity. Because ubiquitination can tag proteins for recognition by the proteasome3, a fundamental question is whether deficits in targeted degradation influence longevity. By integrating data from worms with a defective proteasome, we identify proteasomal targets that accumulate with age owing to decreased ubiquitination and subsequent degradation. Lowering the levels of age-dysregulated proteasome targets prolongs longevity, whereas preventing their degradation shortens lifespan. Among the proteasomal targets, we find the IFB-2 intermediate filament4 and the EPS-8 modulator of RAC signalling5. While increased levels of IFB-2 promote the loss of intestinal integrity and bacterial colonization, upregulation of EPS-8 hyperactivates RAC in muscle and neurons, and leads to alterations in the actin cytoskeleton and protein kinase JNK. In summary, age-related changes in targeted degradation of structural and regulatory proteins across tissues determine longevity.

other tissues. Can the authors attempt to image actin in other cell types, such as the hypodermis? Inclusion of the intestine may also be a great control since they see no lifespan effect with intestinespecific knockdown. This case is made even more important considering the SDS-resistant actin aggregates found in the EPS-8 ubiquitin-less mutants. According to the muscle actin images, no aggregates are visible (just a deformed cytoskeletal network). Where are these aggregates coming from? Are there aggregates in other cell types? 5) RAC components traditionally are positive regulators of actin homeostasis, yet the authors show that knockdown of RAC components can phenocopy eps-8 knockdown, a condition that promotes cytoskeletal integrity. This needs to be addressed. It would also be useful to see cytoskeletal imaging with at least one of the RAC component knockdowns. Minor comments: 1) The presentation of the volcano plots does not seem very helpful for Figure 1. While I appreciate the authors trying to give an overview of the data, it does not really highlight anything important or useful for the reader. Is it possible to make the graphs more meaningful or simply present the data in a different way? Perhaps highlighting some critical targets that were identified -especially the ones that were discussed in the text (e.g. similar to Fig. 3e-f).
2) Combine extended data figures into 2-3 figures to make space for more important things (see major comment 2).

Referee #2 (Remarks to the Author):
This manuscript from Vilchez and colleagues describes the finding that steady state levels of ubiquitinated substrates are altered in C elegans as a function of aging. Moreover, these ubiquitinated substrates can be rescued through dietary restriction or in genetic models of lifespan extension. The authors came to this observation through diGLY proteomics, a mass spectrometry method which allowed them to survey thousands of site specific ubiquitination events. Differentially ubiquitinated substrates were subsequently confirmed using linkage specific antibody pulldowns and western blot analysis of total protein levels. Functionally, the ubiquitin substrates IFB-2 and EPS-8 were both tested by genetic knockdown to demonstrate that suppressing their accumulation extended lifespan. Interestingly, the effects of each substrate manifests in a tissue dependent manner, within intestine and muscle/neuronal cells respectively. A key challenge for this paper is that the mechanism broadly decreasing protein ubiquitination remains to be resolved, and may itself be tissue dependent. These findings come on the heels of previous work from a subset of these authors involving the proteasome assembly factor Rpn6, suggesting that proteasome activity plays at least some role. Steady state levels of ubiquitinated substrates may be influenced by many other activities and there is likely a complex interplay involving transcription, translation, ubiquitination, degradation. All of these processes are dynamic, meaning that demonstration of the proposed mechanism requires an understanding of what process becomes deficient in aging cells to tilt the equilibrium in the way described: -It seems surprising that IFB-2 and EPS-8 would not have been previously identified as factors whose knockdown would extend lifespan, or whose protein levels increase with age. Please explain how or why these have been overlooked in previous screening work by the field.
-Given the complex temporal dynamics observed in Fig 1e-g, a more granular time course should be performed to understand when declines occur in WT worms and how Ub levels change as a function of time, particularly in the daf-2 model.
-As written, the text strongly emphasizes the ubiquitinated proteome, in spite of having performed global proteome analysis in parallel with some or all of the diGLY studies. From the figures and text, it was difficult to ascertain how protein and diGLY levels compare for individual proteins and how different lysine modification sites respond within affected proteins. While reflected in tabular format, this information is more important than many of the summary panels presented in figures 1-2.
-In terms of system dynamics, what does mRNA expression look like for each of the different ubiquitin genes? Are either the ribosomal fusion or stress response ubiquitin genes differentially regulated with age in the WT or mutant strains? -In terms of system dynamics, what is the t1/2 of ubiquitin itself in worms of different ages and genetic backgrounds tested here? This is particularly important for the claims made in lines 368-374.
-Can the authors differentiate between the possibilities that the decrease in conjugated ubiquitin observed in aged WT worms results from elevated DUB activity, rather than a deficiency in the ubiquitin conjugation system? -The statement on lines 125-127 does not accurately reflect what the western blots show.
-The assertions in lines 144-145 seem to be assumptions, since the proteome and diGLY proteome of muscles and neurons were not directly measured.
-Do the authors believe the underlying mechanism proposed to alter steady state protein ubiquitination levels for proteins such as IFB-2 and EPS-8 is conserved across different cell types? Or is it possible that this is tissue specific? In a similar vein, are these effects observed in a cell autonomous manner or only when cells are living in the context of an intact organism? -The assertion on lines 172-175 is overly simplistic and inappropriate given the endpoint experiment carried out. The ubiquitin system functions on the order of minutes to degrade individual substrates and is deeply intertwined with transcription, translation, protein localization and turnover processes. This paper would be improved if it approached modelling of this equilibrium in a more sophisticated manner.
-It would help if the text explained in more detail how Rpn6 knockdown alters the baseline proteome.
For example, what is the proposed impact of USP5 expression changes on phenomena being studied here?
-The concluding statement in lines 208-209 oversimplifies a complex and dynamic process.
-The text indicates that the data will be made available in PRIDE, although this should have been done prior to initial submission and data access provided to reviewers through in a PW protected manner as is now standard. More concerning is the comment that upon publication, additional data will be available upon request. For a paper such as this, this approach is no longer sufficient. The authors should assemble an interactive tool (e.g. Shiny Web app) so that readers of the paper can interact with the global proteome and diGLY proteome data and draw their own conclusions.
-Extending from the concern above, the Supplemental Tables have substantive deficits that make it impossible for an expert in mass spectrometry to judge the quality of these data. For example: o Supplemental Tables 1 & 3 do not provide the peptide sequences that were observed in the diGLY data. This is arguably the most critical piece of information to have in this table for purposes of assessing data integrity.
o Supplemental Table 2 does not provide any indication of the number of peptides used in assembling the quantification data per protein.
Referee #3 (Remarks to the Author): Review Koyuncu et al Nature The authors report the results of a systematic proteomic analysis in which they quantified changes in the ubiquitin-modified signatures during C. elegans ageing. They find that ubiquitination gets downregulated or upregulated for hundreds of proteins between day-5 and day-15 of adulthood. They focus on 10 specific proteins that become under-ubiquitinated during ageing, or hyper-ubiquitinated when the proteasome is inhibited, and show that knocking down the corresponding genes extends lifespan. For two of them, the intermediate filament IFB-2 enriched in the intestine and the actinbinding protein EPS-8, they further identify the lysine residues mediating ubiquitination and show that CRISPR/Cas9 changes of these lysines into arginines reduces lifespan. Last but not least, they examine why changes in the ubiquitination pattern of both proteins alters lifespan. They find that overexpression of IFB-2 cause the aggregation of other intestinal intermediate filaments and reduce animal resistance to intestinal colonization by E. coli. Likewise, they find that EPS-8 knockdown upregulates Rac and JNK/KGB-1 signaling in neurons and muscles, and reduces the disorganization of muscle myofilaments.
Hence the authors went from a proteome-wide study of protein ubiquitination during ageing to the cellular and molecular characterization of two proteins, showing why their ubiquitination matters during normal or genetically modified ageing. This is a remarkable achievement, which will be of interest to a very wide audience as a central quest in the ageing field is to define how to prolong healthy ageing and to identify pathways that could mediate so. Furthermore, experiments are overall very well done and the manuscript is well written. I only have rather minor comments. Of note, C. elegans intermediate filaments in the intestine, epidermis and excretory cell do not resemble any classical vertebrate intermediate filaments; in fact, they are related to lamins, such that the significance of their findings on IFB-2 will await confirmation of a related study in mouse (for instance) with keratins, vimentin, desmin, nestin, GFAP and the like. On the other hand, the results obtained with EPS-8 should be transposable in vertebrates.
Comments: 1. The authors underline that muscles and intestine exhibited a higher number of proteins with altered ubiquitination, with neurons and germline cells coming next. I find it surprising that the skin would not be represented in their list (if anything given the billions involved in the cosmetic industry to make our skin looking forever young!). For instance, EPS-8 plays also an important role in the epidermis and remains expressed in adults. 2. A question concerning the method used to synchronize animals for proteomic studies. The authors used FUDR to prevent the development of the progeny, which is indeed a classical method in the field. I have a potential concern with this approach, inasmuch FUDR inhibits mitochondria and makes animals sterile, both of which have impact on ageing. The issue is whether the treatment could affect the proteins identified in their screens. The ten proteins on which they focus are ok, but what about the hundred others? It would be important to have an assessment of it by probing a small subset of their dataset taking proteins for which antibodies are available. 3. The procedure used for immunoblots involves a centrifugation of worm extracts at 8000g for 5 min to collect the supernatant. The antibody MH33, which they used to probe the abundance of IFB-2, was generated against a highly insoluble worm fraction (Francis and Waterston, JCB 1991). The results shown in Fig. 4b,d are compelling but I was wondering whether the bulk of IFB-2 changes and whether they could correspond to squiggles. 4. Some lifespan assays have been done with the non-integrated lines DVG9, DVG197, and DVG198. The legends do not specify which panels report assays using these strains. 5. Related to this issue, the kcIs6 transgene was generated starting from an extrachromosomal array and is thus present in multiple copies. Could this induce aggregation on its own, and would IFB-2::CFP expressed as a single copy be prone to aggregation as well? 6. I randomly checked the sequences used for qPCR assays taking eps-8 as a test and could not find the position of the forward primer AGAAGAAAAGAAGTGGATTCCGAACT. Please check all sequences mentioned in Table S11.
In this study, Koyunku et al. performed multiple proteomics screens to characterize the dysregulation of ubiquitination during aging. The immense amount of data in the manuscript is commendable, starting first with analyzing ubiquitin status of the proteome during four time points of aging, and then extending these studies to two additional paradigms of longevity: calorie restriction (eat-2) and IIS (daf-2). To further add weight to their findings, they performed general proteomics analysis during aging to complement their ubiquitin screens to directly determine whether changes in ubiquitin status affected protein levels during aging. They followed up these beautifully crafted datasets with a tremendous amount of analysis and description, which is truly spectacular, although on the very rare occasion, there is some over-analysis of data and premature conclusions made (see specific points below). As if this curation of data were not enough, the authors then followed up on two specific candidate proteins, and performed a very thorough analysis of the genetic mechanism whereby IFB-2 and EPS-8 affect the aging process. This manuscript is very well put together with a very rich and valuable dataset for the field of protein homeostasis and aging and is further strengthened by the characterization of two novel mechanistic pathways that alter aging through cytoskeletal remodeling. It is sure to be a major contribution to the fields of aging, protein homeostasis, and cytoskeletal biology and will make a very strong publication for Nature. Below are just some points that can be easily addressed during revision to strengthen the manuscript even further.
Major comments: 1) I am not convinced with the analysis of tissue-specific changes to the ubiquitin proteome. The authors performed this analysis by using previously published datasets to determine which proteins are expressed in a tissue-specific manner, then analyzing the changes that happened for these proteins. The correct way to perform this would be to use tissue-specific protocols and performing proteomics in a tissue-specific manner. This is of course too large an ask for a revision, but minimally, the authors need to address the limitations of their analysis. Ideally, it would be best to remove this analysis and data completely, as the conclusions drawn from this analysis are not strong -showing all tissues displayed changes in ubiquitin status with age with perhaps muscle and intestine showing slightly more changes than neurons or germline -is not a novel or interesting conclusion, especially from an analysis method with numerous caveats.
Reviewer #1 is absolutely right and this concern was also raised by Reviewer #2. We have now made clear in the text that Extended Data Fig. 6a Fig. 6b), supporting that this process occurs through the organism".
2) The manuscript is so data-heavy in the beginning, and a lot of the data is presented in a manner that is not an easy read for a broad audience. For example, it requires the downloading of large excel spreadsheets that we must comb through to get an understanding of the data. I recommend that the authors present some heat maps or other ways to translate some of the spreadsheet data into an easily digestible format. The authors have plenty of room in the Extended Data figures to try and present the data in an easier format, especially because most of the current Extended Data figures are very small (some only on panel), and can be combined into 2-3 figures.
We really appreciate this suggestion from Reviewer #1. We have now replaced volcano plots by heat maps allowing us to present more visibly the global changes in the ubiquitinated proteome discovered in our proteomics experiments. We have also included additional heat maps to visualize other relevant comparisons at the proteome level from all the proteomics analysis included in the Supplementary Tables: • We have now replaced the 9 individual volcano plots presented in Figure 1b of our first submission by heat maps (please see Fig. 1b). These heat maps show in an easier format the widespread changes identified by ubiquitin proteomics: 1) The steady-state amounts of numerous Ub-peptides significantly changed with age in wild-type and long-lived mutants when compared with their respective genetic background at day 1 of adulthood, 2) In wild-type worms, the total number of differentially abundant Ub-peptides increased after day 5 of adulthood and most of these changes were linked with a downregulation in the steady-state levels of Ub-peptides, 3) In contrast, long-lived mutants had fewer number of downregulated Ub-peptides during aging. In fact, daf-2 had an increase in the number of upregulated Ub-peptides with age. Together with the bar graphs presented in Fig. 1c (Fig. 1f). This heat map shows that DR and reduced IIS rescues age-related changes in the Ub-proteome. Detailed information about the specific number of rescued changes by each pro-longevity pathway are presented in Venn diagrams (Extended Data Fig. 1d). Taken together, these graphs summarize Supplementary Table 6. • In addition, we have also included heat maps representing significant agerelated changes in the protein levels of E3 ubiquitin ligases and DUBs (Extended Data Fig. 4a- Fig. 5a). Moreover, the Venn diagram presented in Fig. 2b summarizes our integrated analysis of proteomics data from aging and proteasome-less worms to define proteasome targets that become less ubiquitinated and degraded with age. The Venn diagram also highlights the 10 age-dysregulated proteasome targets identified in our analysis. Thus, these graphs summarize Supplementary Table 8.
3) One concern with the IFB-2 and EPS-8 ubiqutinless mutants is that it may impact the function of the molecules beyond just preventing its degradation. The authors should at least textually comment on why standard overexpression constructs were not synthesized. Ideally, if they can perform validations that these mutations don't affect canonical function of the proteins (e.g. rescue experiments), that would be helpful.
Although ubiquitin-less mutants are more relevant to determine whether there is a direct link between loss of ubiquitination in IFB-2 and EPS-8 with increased protein levels of these factors and regulation of longevity, we agree with Reviewer #1 that standard overexpression constructs could further support our conclusions.
We found that ubiquitin-less ifb-2 mutations increased IFB-2 protein levels in young adult worms and shortened lifespan (Fig. 2g-h and Extended Data Fig. 7a), indicating that upregulation of IFB-2 levels decreases lifespan. In support of this conclusion, we have observed that IFB-2 overexpression is sufficient to shorten lifespan (Extended Data Fig. 7b).
Likewise, ubiquitin-less eps-8 mutant animals exhibited dysregulated high levels of EPS-8 protein at young adult stages and a short-lived phenotype ( Fig. 2i- Fig. 7e). These data further support that the upregulation of EPS-8 protein levels underlies the short lifespan phenotype of ubiquitin-less eps-8 mutant animals.
4) The authors perform visualization of actin filaments solely in the muscle, despite seeing that eps-8 knockdown in muscle and neurons are sufficient to cause a functional phenotype. While neuronal actin integrity is challenging due to the small size of neurons, this phenotype begs to interrogate actin in other tissues. Can the authors attempt to image actin in other cell types, such as the hypodermis? Inclusion of the intestine may also be a great control since they see no lifespan effect with intestine-specific knockdown. This case is made even more important considering the SDS-resistant actin aggregates found in the EPS-8 ubiquitin-less mutants. According to the muscle actin images, no aggregates are visible (just a deformed cytoskeletal network). Where are these aggregates coming from? Are there aggregates in other cell types?
To visualize actin filaments in the intestine and the epidermis, we have now used transgenic C. elegans strains that express the actin-binding construct LifeAct::mRuby specifically in these tissues (Higuchi-Sanabria et al Mol. Biol. Cell 2018). In contrast to the robust beneficial effects in muscle cells (Fig. 5d), knockdown of eps-8 did not prevent or only slightly ameliorated age-associated changes in actin organization within intestinal and epidermal cells, respectively (Extended Data Fig. 10b-c). In this regard, it is important to note that similar to knockdown of eps-8 in the intestine alone, we have now observed that tissue-specific knockdown of eps-8 in the epidermis does not extend lifespan (Extended Data Fig. 10a).
We agree with Reviewer #1 that the most prominent age-related phenotype in the images of muscle cells is a deformed actin cytoskeletal network. However, we also find areas with higher accumulation of actin compared with the rest of the muscle actin cytoskeleton that could partially correspond to the aggregating actin detected by filter trap, whereas the actin staining is more uniform in young worms ( Fig. 5d and 5h). The areas of concentrated actin observed in aged worms are rescued by knockdown of either eps-8 and RAC orthologues ( Fig. 5d and 5h). Moreover, ubiquitin-less EPS-8 accelerated the appearance of concentrated actin areas (Fig. 5i). Please see also below the aforementioned figures, where we have highlighted and showed higher magnification of the concentrated actin areas for the Reviewer (Fig. R1a-c). Thus, we cannot discard that aggregation of actin protein occurs in muscle cells. To determine in which tissues the age-dysregulated levels of EPS-8 promote actin aggregation, we have now performed filter trap assay upon tissue-specific RNAi of eps-8. We found that knockdown of eps-8 in the muscle reduces the amounts of actin aggregates during aging (Extended Data Fig. 10d). Likewise, downregulation of eps-8 in neurons also resulted in decreased actin aggregates (Extended Data Fig. 10e). However, knockdown of eps-8 in the epidermis or intestine did not reduce age-related actin aggregation (Extended Data Fig. 10f-g). Taken together, these results indicate a link between upregulated eps-8 and aggregation of actin in both muscle cells and neurons during aging. 5) RAC components traditionally are positive regulators of actin homeostasis, yet the authors show that knockdown of RAC components can phenocopy eps-8 knockdown, a condition that promotes cytoskeletal integrity. This needs to be addressed. It would also be useful to see cytoskeletal imaging with at least one of the RAC component knockdowns. Minor comments:

As
1) The presentation of the volcano plots does not seem very helpful for Figure 1. While I appreciate the authors trying to give an overview of the data, it does not really highlight anything important or useful for the reader. Is it possible to make the graphs more meaningful or simply present the data in a different way? Perhaps highlighting some critical targets that were identified -especially the ones that were discussed in the text (e.g. similar to Fig. 3e-f).
As indicated in our response to major comment #2, we have now replaced all the Volcano plots previously presented in Fig. 1-2 by heat maps. These heat maps show in an easier format the global changes in the Ub-proteome and how they compare with total protein levels. While Figure 1 and related Extended Figure 1 summarize global changes across the proteome, we have now focused Figure 2 on the specific agedysregulated proteasome targets identified in our proteomics analysis. As such, the different graphs presented in Figure 2 already highlight these critical targets early in the manuscript. In

Referee #2 (Remarks to the Author):
This manuscript from Vilchez and colleagues describes the finding that steady state levels of ubiquitinated substrates are altered in C elegans as a function of aging. Moreover, these ubiquitinated substrates can be rescued through dietary restriction or in genetic models of lifespan extension. The authors came to this observation through diGLY proteomics, a mass spectrometry method which allowed them to survey thousands of site specific ubiquitination events. Differentially ubiquitinated substrates were subsequently confirmed using linkage specific antibody pulldowns and western blot analysis of total protein levels. Functionally, the ubiquitin substrates IFB-2 and EPS-8 were both tested by genetic knockdown to demonstrate that suppressing their accumulation extended lifespan. Interestingly, the effects of each substrate manifests in a tissue dependent manner, within intestine and muscle/neuronal cells respectively. A key challenge for this paper is that the mechanism broadly decreasing protein ubiquitination remains to be resolved, and may itself be tissue dependent. These findings come on the heels of previous work from a subset of these authors involving the proteasome assembly factor Rpn6, suggesting that proteasome activity plays at least some role. Steady state levels of ubiquitinated substrates may be influenced by many other activities and there is likely a complex interplay involving transcription, translation, ubiquitination, degradation. All of these processes are dynamic, meaning that demonstration of the proposed mechanism requires an understanding of what process becomes deficient in aging cells to tilt the equilibrium in the way described: -It seems surprising that IFB-2 and EPS-8 would not have been previously identified as factors whose knockdown would extend lifespan, or whose protein levels increase with age. Please explain how or why these have been overlooked in previous screening work by the field.
* Regarding IFB-2 and EPS-8 as factors whose knockdown extends lifespan: There are four large-scale RNAi longevity screens in C. elegans (Lee et al, Nat. Genet. 2003;Hamilton et al, Genes Dev. 2005;Hansen et al, PLoS Genet. 2005;Curran et al, PLoS Genet. 2007 (Hansen et al, PLoS Genet. 2005) or FUdR treatment (Lee et al, Nat. Genet. 2003;Hamilton et al, Genes Dev. 2005;Curran et al, PLoS Genet. 2007), which are common methods to inhibit progeny and facilitate longevity screens. However, sterility and FUdR treatment can affect distinct biological processes (Feldman et al, PLoS One 2014;Scott el al, Cell 2017;Lee et al, Nat. Metabolism 2019). Since our targeted RNAi screen against age-dysregulated proteasome targets was focused on 10 genes, we performed the lifespan experiments in fertile wild-type worms without FUdR treatment to avoid potential effects of sterility and FUdR treatment. b) Since we defined changes in ubiquitination and protein levels during the aging of adult worms, we started the RNAi treatment for lifespan assays after development (i.e. day 1 of adulthood). This is the main difference with previous largescale RNAi longevity screens, as they started the RNAi treatment at embryonic (Hansen et al, PLoS Genet. 2005) or early larval stages (Lee et al, Nat. Genet. 2003;Hamilton et al, Genes Dev. 2005). As such, these large-scale RNAi screens overlook regulators of adult lifespan if these genes are essential for embryo and/or larval development. A fourth screen started RNAi treatment during later larval stages, but they used a different genetic background (eri-1 mutant worms) and FUdR treatment (Curran et al, PLoS Genet. 2007) making difficult to compare it with our lifespan experiments. In summary, essential genes for development are underrepresented in genome-wide RNAi screens for post-developmental phenotypes such as aging whereas our study circumvents this limitation. This is particularly important for proteins such as EPS-8 and IFB-2, which have a role in normal development. For instance, loss of eps-8 during development triggers embryonic lethality and defects in the morphogenesis of distinct tissues in C. elegans (Croce et al, Nat. Cell Biol. 2004;Ding et al, Plos One 2008). Likewise, IFB-2 is necessary for intestinal morphogenesis during C. elegans development (Geisler et al, Sci. Rep. 2020, Karabinos et al, Eur. J. Cell Biol. 2004. We have now made more clear in the main text that several age-dysregulated proteasome targets are essential for development and we started our targeted RNAi screen at day 1 of adulthood. For example, the text now says: " Several age-dysregulated proteasome targets have an essential role during development. For instance, loss of 30 , IFB-2 31 , RPL-4 32 or F54D1.6 33 before reaching adulthood leads to embryogenic and developmental defects. Whereas these factors endow benefits early in life, we asked whether the age-associated increase in their levels have detrimental effects for adult lifespan. We hypothesized that if age-dysregulated proteasome targets are not essential for adult viability, genetic interventions that specifically diminish their levels only during adulthood can prolong longevity. To this end, we performed single RNAi treatment against the age-dysregulated proteasome targets after development into adult worms". We have also specified in the Methods section and Figure Legends that we started RNAi treatment after development.
* Regarding IFB-2 and EPS-8 as factors whose protein levels increase with age: Our proteomics analysis of global protein levels indicates that both IFB-2 and EPS-8 amounts increase with age, as we validated by western blot experiments. A previous study from Prof. Kenyon's laboratory carried out one of the most comprehensive proteomics analysis of protein levels in C. elegans (Narayan et al, Cell Systems 2016). In this work, the authors performed a SILAC-based deep analysis of protein levels comparing young (day 1) and older (day 10) adult worms. Although the authors mostly focused on enriched gene ontology (GO) term analysis to identify age-related processes, other proteins that did not belong to enriched GO terms were also changed with age. -Given the complex temporal dynamics observed in Fig 1e-g, a more granular time course should be performed to understand when declines occur in WT worms and how Ub levels change as a function of time, particularly in the daf-2 model.
-As written, the text strongly emphasizes the ubiquitinated proteome, in spite of having performed global proteome analysis in parallel with some or all of the diGLY studies. From the figures and text, it was difficult to ascertain how protein and diGLY levels compare for individual proteins and how different lysine modification sites respond within affected proteins. While reflected in tabular format, this information is more important than many of the summary panels presented in figures 1-2.
We have now re-written the text and included new figures to emphasize the comparison between diGLY and total protein levels for individual proteins: •  (Fig. 1e) Fig. 1b and Supplementary Table 3). Since we detected a higher number of changes in the levels of Ub-peptides after day 5 of adulthood (Fig. 1c), we directly compared young (day 5) with aged (day 15) wild-type worms for further analysis of age-related differences. Indeed, old worms underwent a deep remodeling of their Ub-modified proteome when compared with day 5-adults ( Fig. 1e and  Supplementary Table 4). In aged wild-type worms, the steady-state levels of 1813 Ubpeptides were downregulated whereas 350 Ub-peptides were upregulated (Fig. 1e and  Supplementary Table 4), further supporting that aging is particularly associated with a decline of ubiquitinated peptides. Among these age-related differences, only 582 downregulated and 123 upregulated Ub-modified peptides correlated with a change in the total levels of the protein in the same direction ( Fig. 1e and Supplementary Table  5). However, 905 downregulated and 189 upregulated Ub-peptides corresponded to proteins that did not change in abundance with age ( Fig. 1e and Fig. 1c). The rest of differentially abundant Ub-peptides were inversely correlated with the corresponding protein levels ( Fig. 1e and Fig. 1c). Thus, our integrative analysis demonstrated that not all the differences in Ub-peptides can be simply ascribed to a similar change in the total protein levels".
-In terms of system dynamics, what does mRNA expression look like for each of the different ubiquitin genes? Are either the ribosomal fusion or stress response ubiquitin genes differentially regulated with age in the WT or mutant strains?
We have now assessed the mRNA levels for the ubiquitin-encoding gene ubq-1, the ubiquitin-ribosomal fusion genes ubq-2 and ubl-1 as well as the ubiquitin-stress response gene usp-14 in wild-type and long-lived mutant strains at day 1 and day 15 of adulthood (please see Fig. 1j and Extended Data Fig. 3a-c). Data Fig.  3a). We also observed a significant upregulation of ubq-1 transcript levels in long-lived daf-2 mutant animals with age (Extended Data Fig. 3b-c). Nevertheless, aged wildtype worms expressed higher or similar levels of ubq-1, ubq-2 and ubl-1 when compared with age-matched eat-2 and daf-2 mutant animals (Fig. 1j). Moreover, aging and prolongevity pathways did not change the expression of usp-14 ( Fig. 1j and Extended Data Fig. 3a-c), a ubiquitin-stress response gene induced by ubiquitin deficiency 21 ".

The text now says: "With the global decline of Ub-proteins in aged wild-type worms, we asked whether this process directly correlates with changes in the expression of ubiquitin itself. Ubiquitin is encoded by three different genes in C. elegans, i.e. ubq-1 and the ubiquitin-ribosomal fusion genes ubq-2 and ubl-1. In contrast to the global downregulation of Ub-peptides, aged wild-type worms had higher transcript levels of both ubq-1 and ubq-2 when compared with young worms, whereas ubl-1 mRNA levels remained similar (Extended
-In terms of system dynamics, what is the t1/2 of ubiquitin itself in worms of different ages and genetic backgrounds tested here? This is particularly important for the claims made in lines 368-374. We have now treated the worms with cycloheximide to block ubiquitin synthesis (Hanna et al, Mol. Cell Biol. 2003) and assessed t1/2 of protein ubiquitin itself in wild-type, eat-2 and daf-2 animals at day 1 and 10 of adulthood (Extended Data Fig. 3e-f). We found that the half-life of ubiquitin is approximately 2 hours in young wild-type worms and approximately 30 minutes longer in aged wild-type worms (Extended Data Fig. 3e-f). Data Fig. 3ef). Taken together, these data indicate that the downregulated amounts of ubiquitinated proteins characteristic of aged wild-type worms is not associated with a decrease in the half-life of ubiquitin itself.

Likewise, long-lived eat-2 and daf-2 mutants exhibited similar half-lives for protein ubiquitin when compared with age-matched wild-type worms (Extended
-Can the authors differentiate between the possibilities that the decrease in conjugated ubiquitin observed in aged WT worms results from elevated DUB activity, rather than a deficiency in the ubiquitin conjugation system?  Fig. 4b). (Fig. 1k). We found that single knockdown of csn-6/COPS6, H34C03.2/USP4, F07A11.4/USP19, math-33/USP7, usp-5/USP5, usp-48/USP48 or otub-3/OTUD6A ameliorates the decline of Ub-protein levels in aged wild-type worms, whereas downregulation of other DUBs did not prevent this phenotype (Extended Data Fig. 4c-g). Altogether, these results suggest that elevated DUB activity could tilt the equilibrium towards the global decrease in ubiquitination levels characteristic of aged worms.

-The statement on lines 125-127 does not accurately reflect what the western blots show.
We have now rewritten these lines. The text now says: "Among the numerous ubiquitination changes in wild-type worms during aging, the proteomics data indicated that the equilibrium is tilted towards a decline in ubiquitination levels. To further assess these changes, we performed western blot analysis at the same ages investigated in our proteomics assay. Indeed, day 10 and day 15 wild-type worms exhibited a dramatic decrease in the global levels of Ub-proteins compared with day 5 and day 1 wild-type worms (Fig. 1g). On the other hand, day 10 and 15 eat-2 mutant worms did not have a strong downregulation in Ub-protein levels compared with younger ages (Fig. 1h). In daf-2 mutant worms, the levels of Ub-proteins were upregulated compared with day 1 adults of the same genetic background, a phenotype that was already acute at day 5 of adulthood ( Fig. 1i) Fig. 2a-c)".

. A more granular time-course analysis indicated that the robust decline in Ub-protein levels of wild-type worms occurs after day 8 of adulthood, whereas the increase in Ub-proteins of daf-2 mutant animals starts from day 2 (Extended Data
-The assertions in lines 144-145 seem to be assumptions, since the proteome and diGLY proteome of muscles and neurons were not directly measured. Reviewer #2 is absolutely right and this concern was also raised by Reviewer #1. As indicated in our response to Reviewer #1, we have now made clear in the text that Extended Data Fig. 6a presents a bioinformatic analysis combining our proteomics data with previously published datasets of tissue-specific expression and toned down the conclusions from this analysis. For instance, we have removed the statement regarding that the muscle and intestine have slightly more changes than neurons or germline. We agree that this conclusion requires ubiquitin proteomics experiments in a tissue-specific manner, which are not feasible due to the lack of protocols to isolate most of the tissues (e.g., muscle, epidermis) or obtain sufficient protein amounts of other tissues such as the germline and intestine for diGly enrichment. Data Fig. 6b). The text now says: "Using previously published datasets of tissue-specific expression 37 , we classified proteins that exhibit differences in their Ubpeptides according to the tissues where these proteins are expressed. The bioinformatic analysis indicated that distinct tissues such as the germline, muscle, intestine, epidermis or neurons express multiple proteins that contain downregulated Ub-peptides with age (Extended Data Fig. 6a Data Fig. 6b), supporting that this process occurs through the organism".

and Supplementary Table 10). Similar to whole organism lysates, we observed a decrease in the global amounts of Ub-proteins in isolated germlines, intestines and heads during aging (Extended
-Do the authors believe the underlying mechanism proposed to alter steady state protein ubiquitination levels for proteins such as IFB-2 and EPS-8 is conserved across different cell types? Or is it possible that this is tissue specific? In a similar vein, are these effects observed in a cell autonomous manner or only when cells are living in the context of an intact organism? As mentioned above, we have now extruded different tissues and body parts of C. elegans to assess whether the age-associated decline in the steady-state levels of ubiquitination is conserved across tissues. Similar to lysates from the entire animal, there was a pronounced decrease in the global amounts of Ub-proteins in isolated germlines, intestines and heads (Extended Data Fig. 6b). These data support that ageassociated downregulation in global ubiquitination levels occurs in different cell types. Data Fig. 6c), correlating with the decline in ubiquitination levels across tissues. Since the ifb-2 gene is specifically expressed in intestinal cells (Bossinger et al, Dev. Biol. 2004), its protein levels were upregulated in isolated intestines of old worms (Extended Data Fig. 6d). However, we did not detect IFB-2 in a distinct somatic tissue even during the aging process (Extended Data Fig.  6d).

Moreover, we have now examined age-related changes in the EPS-8 and IFB-2 protein levels of isolated tissues. EPS-8 is ubiquitously expressed in the soma (Ding et al, Plos One 2008; Stetak et al, EMBO J. 2006) and its protein amounts were increased in different somatic tissues with age (Extended
To further assess the impact of the ubiquitin-proteasome system in the intracellular regulation of IFB-2 and EPS-8 protein levels, we have now performed tissue-specific knockdown of rpn-6. Loss of rpn-6 in the intestine, epidermis, neurons or muscle increased EPS-8 levels in young worms, indicating that the ubiquitin-proteasome system modulates EPS-8 in all these tissues (Extended Data Fig. 6e-h). However, only intestinal loss of rpn-6 was sufficient to upregulate IFB-2 levels in young worms (Extended Data Fig. 6e-h), according to the specific expression of IFB-2 in the intestine. Taken together, these experiments support that the ubiquitin-proteasome system acts in a cell autonomous manner to regulate IFB-2 and EPS-8 levels.
Since we performed our proteomics experiments in the context of an intact organism, it is possible that cell non-autonomous mechanisms also regulate steady-state levels of ubiquitination in distal tissues during aging. Indeed, growing evidence demonstrates that interorgan communication is also an important determinant of organismal aging (Taylor et al, Nat. Rev. Mol. Cell Biol. 2014). For instance, the nervous system elicits signals that modulate aging of distal tissues (Taylor et al, Nat. Rev. Mol. Cell Biol. 2014). To assess whether cell non-autonomous events influence the age-associated decline in ubiquitination levels, we have now examined unc-13 mutant worms which are deficient in the release of neurotransmitters from small clear vesicles (Richmond et al, Nature Neuroscience 1999). Importantly, blocking neurotransmitter release did not affect the steady-state levels of ubiquitination in young worms, but it exacerbated the age-associated decline in older worms (Extended Data  Fig. 6i). Concomitantly, the protein levels of ubiquitously expressed EPS-8 and intestinal-specific IFB-2 were further upregulated in aged unc-13 mutants when compared with age-matched wild-type worms (Extended Data Fig. 6j-k), suggesting a role of cell non-autonomous mechanisms in this process.
-The assertion on lines 172-175 is overly simplistic and inappropriate given the endpoint experiment carried out. The ubiquitin system functions on the order of minutes to degrade individual substrates and is deeply intertwined with transcription, translation, protein localization and turnover processes. This paper would be improved if it approached modelling of this equilibrium in a more sophisticated manner.
Reviewer #2 is absolutely right. We have now re-written the text and added new experimental data, performing a more sophisticated approach for the modelling of changes in the steady-state of ubiquitination and protein levels: • We have now removed the specific sentences indicated by Reviewer #2: "Thus, differences in the amounts of these Ub-modified peptides could ensue from agerelated changes in the transcriptional or translational regulation of protein levels.
In addition, 905 downregulated and 189 upregulated Ub-modified peptides did not result in alterations of protein abundance with age ( Fig. 3a and (Fig. 1f, Extended Data Fig.  1d and Supplementary Table 6). Among the 350 upregulated Ub-peptides in aged wild-type worms, eat-2 and daf-2 animals had decreased ubiquitination for 234 and 251 peptides, respectively (Fig. 1f, Extended Data Fig. 1d and  Supplementary Table 6). Thus, pro-longevity signaling pathways rescued age-related changes in the Ub-proteome, further supporting that alterations in regulated mechanisms contribute to this process".
• We have now performed a series of experiments that indicate that elevated DUB activity is an important factor to tilt the equilibrium towards the decrease in ubiquitination levels during aging. For instance, we have now treated the worms with a broad-spectrum DUB inhibitor (PR-619) and found that this treatment rescues the age-associated decline in ubiquitination levels (Fig. 1k). Using our proteomics data of global protein levels, we observed that 14 DUBs were significantly upregulated in aged wild-type worms (Extended Data Fig. 4b).
Among them, single knockdown of csn-6/COPS6, H34C03.2/USP4, F07A11.4/USP19, or otub-3/OTUD6A ameliorates the age-associated decline of Ub-protein levels in wildtype worms (Extended Data Fig. 4c-g). • Among the numerous age-related changes in the ubiquitinated proteome, our aim was to define proteasome targets that become dysregulated with age. Therefore, we have kept the focus on the ubiquitin-proteasome system. By integrating ubiquitin and total protein proteomics of both aging and proteasome-less worms with immunoprecipitation experiments of Lys48-linked polyUb chains, we identified age-dysregulated proteasome targets such as IFB-2 and EPS-8. As we have now discussed in the text, we cannot discard that other activities could also influence the levels of these proteins. The Discussion section says: "Although our findings support at least a partial role of the ubiquitin-proteasome system in regulating the amounts of specific longevity modulators, it is important to note that changes in other processes such as translation or protein localization could also contribute to regulating the ubiquitination, protein levels and activity of these factors". • Nevertheless, we believe that the upregulation of proteins such as IFB-2 and EPS-8 upon rpn-6 RNAi supports at least a partial a role of the proteasome in regulating their levels. We further strengthened this conclusion with supporting data from other experiments: 1) the mRNA levels of EPS-8 and IFB-2 did not increase upon aging or rpn-6 RNAi, 2) blocking ubiquitination of EPS-8 and IFB-2 by gene editing resulted in increased levels of these proteins and 3) knockdown of rpn-6 in the intestine, epidermis, muscle and neurons increased the levels of the ubiquitously expressed EPS-8 whereas only knockdown of rpn-6 in the intestine upregulated the amounts of the intestinal protein IFB-2. • We have now explained in more detail our criteria to identify potential proteasome targets, while indicating that we cannot discard that other ubiquitinated proteins could also be modulated by the proteasome. The text now says: "Given the high number of downregulated Ub-peptides during aging, we hypothesized that a subset of these events could reduce selective degradation of specific proteins by the proteasome. Our integrated analysis revealed age-related changes in multiple Ub-peptides that directly correlated with similar differences in the protein levels or corresponded to proteins with unchanged abundance, making difficult to interpret whether these Ub-sites modulate proteasomal degradation of the protein (Fig. 2a and Supplementary Table 5). To identify downregulated ubiquitin marks involved in protein degradation, we focused on changes in Ub-peptide levels that inversely correlated with protein amounts. Notably, we found that 192 proteins exhibited less ubiquitination in at least one of their lysine sites during aging, while the total levels of the protein increased ( Fig. 2a and Supplementary Table 5). Thus, these events could indicate a loss of ubiquitin marks in target proteins, resulting in diminished degradation of the protein with age. If Ub-proteins are indeed proteasomal targets, defects in proteasome activity could reduce their degradation in young animals, leading to increased amounts of both the protein and the corresponding Ub-peptides".
-It would help if the text explained in more detail how Rpn6 knockdown alters the baseline proteome. For example, what is the proposed impact of USP5 expression changes on phenomena being studied here?
We have now added a heat map representing differentially abundant proteins upon rpn-6 RNAi treatment (please see Extended Data Fig. 5a) and explained in more detail how rpn-6 knockdown alters the baseline proteome. As suggested by Reviewer #2, we have also discussed about the potential impact of usp-5 expression changes upon rpn-6 knockdown. The text now says: "To decrease proteasome function in young adult worms, we knocked down rpn-6, a specific activator of 26S proteasome assembly and activity 10,27 . Given the role of the proteasome in multiple biological pathways, rpn-6 RNAi resulted in widespread changes in the proteome baseline of young adults (day 5), including 297 downregulated proteins and 509 upregulated proteins (Extended Data Fig. 5a and Supplementary Table 8). Besides potential indirect effects on protein levels caused by proteasome dysfunction, we hypothesized that these upregulated proteins also include direct proteasome targets which are less degraded upon rpn-6 RNAi, particularly if they also have increased Ub-peptides". (…) "Likewise, loss of rpn-6 did not upregulate the transcript levels of these targets, with the only exception of the deubiquitinase usp-5 (Extended Data Fig. 5c). Given the role of USP-5 in protein ubiquitin homeostasis 28 , a process challenged by deficits in proteasome activity, the increased expression of usp-5 gene could indicate a compensatory mechanism to protect from the deleterious effects induced by rpn-6 RNAi".
-The concluding statement in lines 208-209 oversimplifies a complex and dynamic process.
We apologize for this oversimplification. We have now replaced the concluding statement "Concomitantly, these proteins cannot be recognized and degraded by the proteasome, resulting in upregulated levels of the protein in aged animals" by "Altogether, our data indicate that the profound rewiring of the Ub-proteome during aging could also decrease the steady-state ubiquitination levels of distinct proteasome targets. In turn, lower ubiquitination marks could reduce the proportion of these targets for recognition and subsequent degradation by the proteasome, eventually contributing to the upregulation in their total protein amounts with age".
-The text indicates that the data will be made available in PRIDE, although this should have been done prior to initial submission and data access provided to reviewers through in a PW protected manner as is now standard. More concerning is the comment that upon publication, additional data will be available upon request. For a paper such as this, this approach is no longer sufficient. The authors should assemble an interactive tool (e.g. Shiny Web app) so that readers of the paper can interact with the global proteome and diGLY proteome data and draw their own conclusions.
We apologize for not providing access to the raw proteomics data in our initial submission. We have now deposited all the proteomics data in PRIDE with access for the reviewers using the following reviewer account details: • -Extending from the concern above, the Supplemental Tables have substantive deficits that make it impossible for an expert in mass spectrometry to judge the quality of these data. For example: o Supplemental Tables 1 & 3 do not provide the peptide sequences that were observed in the diGLY data. This is arguably the most critical piece of information to have in this table for purposes of assessing data integrity.
We apologize for not providing peptide sequences for the diGLY proteomics data in our first submission. We have now included the peptide sequences in the corresponding Supplementary Tables for all the ubiquitin-proteomics experiments: Supplementary  Tables 1, 4 and 8.
o Supplemental Table 2 does not provide any indication of the number of peptides used in assembling the quantification data per protein.
We are very sorry for not including this relevant information in our first submission. We have now included the total number of identified peptides per protein as well as the number of peptides used for quantification of individual protein levels in Supplementary  Tables 2, 8 and 9.

Referee #3 (Remarks to the Author):
Review Koyuncu et al Nature The authors report the results of a systematic proteomic analysis in which they quantified changes in the ubiquitin-modified signatures during C. elegans ageing. They find that ubiquitination gets downregulated or upregulated for hundreds of proteins between day-5 and day-15 of adulthood. They focus on 10 specific proteins that become under-ubiquitinated during ageing, or hyper-ubiquitinated when the proteasome is inhibited, and show that knocking down the corresponding genes extends lifespan. For two of them, the intermediate filament IFB-2 enriched in the intestine and the actinbinding protein EPS-8, they further identify the lysine residues mediating ubiquitination and show that CRISPR/Cas9 changes of these lysines into arginines reduces lifespan. Last but not least, they examine why changes in the ubiquitination pattern of both proteins alters lifespan. They find that overexpression of IFB-2 cause the aggregation of other intestinal intermediate filaments and reduce animal resistance to intestinal colonization by E. coli. Likewise, they find that EPS-8 knockdown upregulates Rac and JNK/KGB-1 signaling in neurons and muscles, and reduces the disorganization of muscle myofilaments.
Hence the authors went from a proteome-wide study of protein ubiquitination during ageing to the cellular and molecular characterization of two proteins, showing why their ubiquitination matters during normal or genetically modified ageing. This is a remarkable achievement, which will be of interest to a very wide audience as a central quest in the ageing field is to define how to prolong healthy ageing and to identify pathways that could mediate so. Comments: 1. The authors underline that muscles and intestine exhibited a higher number of proteins with altered ubiquitination, with neurons and germline cells coming next. I find it surprising that the skin would not be represented in their list (if anything given the billions involved in the cosmetic industry to make our skin looking forever young!). For instance, EPS-8 plays also an important role in the epidermis and remains expressed in adults.
Reviewer #3 is absolutely right and we apologize for not including the epidermis in our bioinformatic analysis of proteins with altered ubiquitination levels in our first submission. We have now included the epidermis in the bioinformatic analysis (please see Extended Data Fig. 6d). Similar to other tissues, the epidermis also expressed a high number of proteins that contain downregulated Ub-peptides with age (Extended Data Fig. 6a).

As Reviewer #3 indicates, EPS-8 plays an important role in the epidermis during development and remains expressed in this tissue during adulthood.
We have now performed tissue-specific knockdown of rpn-6 after development to further assess the impact of the ubiquitin-proteasome system in the intracellular regulation of EPS-8 levels in different tissues, including the epidermis. Notably, loss of rpn-6 in the intestine, neurons, muscle or epidermis increased EPS-8 levels in young worms, indicating that the ubiquitin-proteasome system modulates EPS-8 in all these tissues during adulthood (Extended Data Fig. 6e-h).
Although the ubiquitin-proteasome system modulates EPS-8 levels in the epidermis, we found that epidermal-specific knockdown of eps-8 does not extend longevity (Extended Data Fig. 10a). The text now says: "In contrast to ifb-2, tissue-specific knockdown of eps-8 in the intestine alone did not affect lifespan (Fig. 4a). Likewise, tissue-specific knockdown of eps-8 in the epidermis did not modulate lifespan (Extended Data Fig.  10a). However, tissue-specific knockdown in either the muscle or neurons extended adult lifespan (Fig. 4b-c), indicating that EPS-8 activity in these cells regulates organismal longevity".
In addition, we have now assessed whether knockdown of eps-8 ameliorates ageassociated alterations in actin organization within epidermal cells (Extended Data Fig.  10b). The text now says: "In contrast to the robust beneficial effects in muscle cells, knockdown of eps-8 did not prevent or only slightly ameliorated age-associated changes in actin organization within intestinal and epidermal cells, respectively (Extended Data Fig. 10b-c)". Moreover, we have now observed by filter trap assay that knockdown of eps-8 in the epidermis alone does not reduce age-related actin aggregation (Extended Data Fig. 10f).

2.
A question concerning the method used to synchronize animals for proteomic studies. The authors used FUDR to prevent the development of the progeny, which is indeed a classical method in the field. I have a potential concern with this approach, inasmuch FUDR inhibits mitochondria and makes animals sterile, both of which have impact on ageing. The issue is whether the treatment could affect the proteins identified in their screens. The ten proteins on which they focus are ok, but what about the hundred others? It would be important to have an assessment of it by probing a small subset of their dataset taking proteins for which antibodies are available.
This is a very important point raised by Reviewer #3. Given the large amounts of total protein required for the proteomic experiments, the only way to obtain sufficient populations of synchronized worms was to inhibit progeny. As Reviewer #3 indicates, FUdR is extensively used in the aging field to inhibit progeny. Since FUdR does not affect longevity at the standard conditions used in our manuscript (e.g., Mitchell et al, J. Gerontol. 1979;Hosono et al, Exp. Gerontol. 1982;Gandhi et al, Mech. Ageing Dev. 1980 Figure  R2a below). Importantly, we obtained similar results by western blot in wild-type worms without FUdR treatment (Figure R2b). 3. The procedure used for immunoblots involves a centrifugation of worm extracts at 8000g for 5 min to collect the supernatant. The antibody MH33, which they used to probe the abundance of IFB-2, was generated against a highly insoluble worm fraction (Francis and Waterston, JCB 1991). The results shown in Fig. 4b,d are compelling but I was wondering whether the bulk of IFB-2 changes and whether they could correspond to squiggles.

Figure R2. Assessment of a subset of proteins by western blot in worms without
As Reviewer #3 indicates, MH33 antibody was generated against a highly insoluble worm fraction and we completely agree that further controls are needed to strengthen our conclusions. We have now performed western blot with antibody against IFB-2 (MH33) of whole C. elegans extracts without the 8000xg centrifugation step. Similar to supernatant samples (Fig. 2e), we observed an increase in the total levels of IFB-2 in aged worms when the whole C. elegans extract was analyzed (please see Extended Data Fig. 5h). In addition, analysis of whole worm extracts also supports that ubiquitinless IFB-2 double mutation hastens the upregulation of IFB-2 protein levels (Extended Data Fig. 7a). We have indicated in the corresponding

Figures when we used whole worm lysates and also made this point clear in the
Methods section: "When indicated in the corresponding figure, we also performed analysis of total IFB-2 levels with antibody MH33 in whole C. elegans lysates without the centrifugation step".
To circumvent any interference on the results ensued from a potential higher affinity of MH33 antibody towards the IFB-2 of the highly insoluble fraction, we have now performed western blot using anti-GFP antibody in worms expressing IFB-2 tagged with either CFB or GFP. To this end, we analyzed both the transgenic strain :CFP]IV) from Prof. Leube's lab and also a new strain that expresses endogenous IFB-2 fused to GFP generated by CRISPR/Cas-9. Importantly, the western blots using GFP antibody further validated that IFB-2 protein levels increase with age (please see Extended Data Fig. 5f-g).
4. Some lifespan assays have been done with the non-integrated lines DVG9, DVG197, and DVG198. The legends do not specify which panels report assays using these strains.
We  Fig. 7b)". We have also specified the used of these specific strains in the corresponding figure legend: "Non-integrated transgenic DVG197 and DVG198 strains overexpressing ifb-2 under sur-5 promoter exhibit a short lifespan phenotype compared to the control strain (P< 0.0001)".
5. Related to this issue, the kcIs6 transgene was generated starting from an extrachromosomal array and is thus present in multiple copies. Could this induce aggregation on its own, and would IFB-2::CFP expressed as a single copy be prone to aggregation as well?
It is important to note that the filter trap assay with antibody to IFB-2 (MH33) presented in Fig. 3e was performed in wild-type worms. Thus, these data support that endogenous IFB-2 protein aggregates during aging. We have now made clear that these experiments were performed using wild-type worms in the main text, Fig. 3e panel and corresponding Figure Legend.
Moreover, we have now tagged endogenous IFB-2 using CRISPR-Cas9 methodology and performed GFP imaging of these worms at different ages. Similar to multi-copy transgenic IFB-2::CFP, endogenous IFB-2 tagged with GFP also showed mislocalization and accumulation into aggregates during aging (please see Extended Data Fig. 8a). Moreover, we have performed filter trap assays with antibody to GFP of worms expressing endogenous IFB-2::GFP and confirmed aggregation of IFB-2 protein (Extended Data Fig. 8b).
6. I randomly checked the sequences used for qPCR assays taking eps-8 as a test and could not find the position of the forward primer AGAAGAAAAGAAGTGGATTCCGAACT. Please check all sequences mentioned in Table S11.
We have double checked all the primer sequences indicated in Table S11 (now  Supplementary Table 12) and confirmed that all of them are correct. It is important to note that EPS-8 has many different isoforms and we designed the qPCR primers for the transcript Y57G11C.24g.1 (please see Figure R3 below), which is the most abundant isoform in C. elegans according to our proteomics data. We apologize for not specifying the specific transcript for EPS-8 in our first submission. We have now included this relevant information in Supplementary Table 12.   tttcagccagtcaaatgccatttgcttaaaaagcgaggaaatttatcatcagtgaagaagtgtaaaacatctatt  aaaaatgATGCGTCGAGGTGGATCGATGGGTCCACCGAGCGGGGATCCATATCAGAGTCGCCCATCCCCCGGAGG  CTACTACTACAACCGTTCAACGCCTGGTGGTCAGCCAGCTCCATCACCCTCACACAGTCAACAATCCGCATCTTC  ACATCATCCAAGAGGAGTGCCAATGTCACAACCAATCGCCCGCCGATCGGACTACCGTACTGGAAGTGAACAAAT  GACTCCACGATCCGATCATCGTGGCCCATCGATGGGTTACGGTAATGGGGGTTCTGTGGATCAACGTGTCGATGA  CGTCACGCCGTCCTACTACGTGGAACATCTTGCCACGTTTGCAGTTGGAAGACAGTTTGGACTCACTTTTCCAGC  CGATGGCATCAGAAAGTTGAAACAAATGGAAAAGAATTCAGCCATTTGGGCGCAACCTCTAATTCTTCGGTTTCG  ACACAACGCAGTGACGGTAGAAGACGACAACGGGGAGCTCGTCGAGCAATTTCCGTTGGAATTAATCGAGCAGCC  GACGGCTCACGTGAGCAATGATTCTCGTGAGACTTACAACAATGTGCTCCTTTTTGTTGTCCGGGAGGACCGGAA  GAGGATGAGCACGCCCACCGAGATGCACATTTTCCAGTGTATCCGTGTATCCGCTACCGATGTGGCCGAGGACCT  CAAAAACTACGTACAAGGTCAATTCCGTCGTGTGCGTAATGGCCGACGTACCGCCGCGCCGACGCATCTTCAAGC  TCAACAACAACAAATGCCATTCTACCCACCGGACGACGCTTCAATCAGCAGTGAAACATCTGAAATGTTCGAACG  AGATGTGAATACACTGAATCGTTGTTTCGATGATATCGAACGATTTGTGGCGAGAATTCAATCGGCCGCACTGGC  TCAGCGAGAAATTGAGCAACAGAATCATAGATATCGAACCGCGAATCGTCGGGACAAGAAGAACCAGCAGCCACC  AGATCCGAATGGCATCCTATTTATGAGAGCTCAACTTCCATTGGAGTCTGAATTTGTCGATATTTTGAAGAAGTT  CAAGCTCTCCTTTAATCTACTGGCCAAACTCAAAAACCACATCCACGAGCCAAATGCTCCGGAACTTTTGCACTT  TCTCTTCACGCCACTAGCTGTGATCCTTGAAGCATGCCACTGGGGGCTCGGAAGAAATGTTGCTCCAACTGTTGC  CTCGCCGCTGCTCTCGTTGGAAGCTCGTGAGCTCATGCAAAACTGTTTGACAAGTCATGAGAGTGACATTTGGAT  GAGCCTCGGAGAAGCCTGGAGAACTCCACCAGAGGACTGGACAAAACCACTTCCACCACCATATCGTCCAATCTT  CAATGATGGGTTTGCCCCTTACGGGGTCGCCGACAGAGCCATGGCAACTCCAAATCAAATCCATCGTGGACATTC  TGCTCCGCCAGAGCACTTCCGTCAGCCACCGCCCAGAGAGAGGAATATGGTGGATACGCTCGAATTCGATCGATT  AACTCTGGAACGCGAGCGACTCGAATTTGAAAAGGCAAAGATTATGGAAAGGGAGAGTCGTTTGAGGCATGAGGA  GAAGCAGATTGAAGATGAAAAGCGACGAATGCACGCCGAGAAGGATCTCATCACAAAAGAGACAACACAACCAGT  TCCACCACCAGCTGCAGTCGTTACACATCAACCAATCACCAAGCGATATGATCCACCAATTTCCATCTCTCCACC  ACCACAACGTAACAACTATTCACACGTGAAGGTTACTGTAGATTCTGACACATCGCCACGTCAGCAGGCATTCAT  TGATGACATCGTGGCAAAAGGTGGCAAGTTGGCAGTGGTCACCTATGATAGAGGAGGTCAGAATACGAAGGAGCT  GACTGTTCACAAGGGGGAGTATTTGGAGGTTATCTTCGACGAGCGCAACTGGTGGGAGTGCAAGAATATGCATCA  AAGAGTCGGATACGTTCCACACACAATTTTGTCAATGGTGCCATTTGAGCAACAACAATATGCGTCACCAAACCA  TCACAACAATTCCTCATCAACCGGTGGCTACAACAACGGACATCATCAAGGCCCAGCGCAAAACGTCTATCGTCC  ACCCCCACCACCACTAGTATCCGATGCAGGTGTCCAGGTCGAAATCCGCCGGGAACATGTTGCTCCGCCACCACC  TCCAGTTGTAATTCCACCACCACCACCGCCAGTGCGAACTCCAACTATGGAAGAGTTGCTCCGCATGCAACAACA  ACAACAGCAGAAGCAGAGAAAACCACCAGTTGAAGAACCAGTCTACCAACCGCAACCCGCTCAACGTGCCGGATT  GGAAATCTGGAGACGGAATCAGAATCCTCAATTGATTCAAGAAGTGACTGAAACTGTGGGACAACACAGTGGAGA  CATTTTAAGACCAGCAATGCAAAGGGCCACTCGAGTGGCAATCAACGAAAAGTCTTCTCCGGAAGACGTGACTCG  TTGGTTACAGGAGAAGGGATTCTCACCAAGAGTAATTGACCTATTGGATGGTCAAGATGGTGCCAATCTATTCTC  CTTGTCAAAATTGCATCTGCAACAAGCTTGTGGAAGAGATGAGGGTGGATATTTGTACAGTCAATTGTTGGTTCA GAAGAAAAGAAGTGGATTCCGAACTCACACTGGCGATGAATTGAAAGCAATTCTCAATCACAGACGAACCCACGT GGAATTATCAAATGAAGCTGCAGCTGATGAACCTGTATTCACTATTAACCCAATTCTCTAGgtttttcaaatttc tgtctaacttttccaacgacaaaaatccaaatttcacaaaaaaaatattgatttaaatattattttcgctaattc ttttggctagctcaacaataagaaatcccccgattttccagatgaatactcattccaaacttactttcaatatgt tcaatttactgcctccgttaatccatttttgtttcagattttgaattttcaacatttttaattgaattatgtttc aaataaatgcttctgcgaata the lack of obvious actin aggregates in micrographs of old muscle tissue was questioned by Reviewer #1, while whole-worm filter trap analysis suggested these aggregates should occur. The authors now show that eps-8 knockdown specifically in muscle or neurons clears those aggregates, while in epidermis and intestine it does not. However, they also point to micrographs of old worms in which stained F-actin shows undulations and other deformities. In their rebuttal, the authors enlarge sections of these micrographs and point to the higher action accumulation in certain regions "that could partially correspond to the aggregated action detected by filter trap" analysis. I am not convinced by this argument and likely these images are being over-interpreted. All I can observe are undulating actin filaments that are not optically resolved due to close proximity.
We agree with Reviewer #4. We have now made clear in the text that the aggregation of actin was detected by filter trap: "Filter trap experiments indicated that actin protein also aggregates during aging, whereas knockdown of eps-8 rescued this phenotype (Extended Data Fig. 10e)".
The comment of Reviewer #1 on the role of RAC components as regulators of actin homeostasis, as well as the subsequent minor comments, are well-addressed by the authors.
Finally, Reviewer #2 made a short but important remark on the relative roles of a decreased ubiquitin conjugation system versus elevated DUB activity to explain the decrease in ubiquitination in aging worms. This led the authors to conduct a new interesting series of experiments that brings this manuscript to (an even) higher level. Their most important finding was that elevated DUB activity may be a major cause of the global decrease in protein ubiquitination in aged worms. Using a broad-spectrum DUB inhibitor, the authors showed that the age-related decrease of ubiquitination can be reversed. The central theme in this manuscript is the link between decreased protein ubiquitination and, in some cases, the resulting protein accumulation, the subsequent functional deterioration, and its final effect on lifespan. Therefore it is difficult to understand why the authors did not include a simple, low-tech, but crucially important lifespan experiment that compares controls with animals treated (from adulthood) with the DUB inhibitor PR-619. One would expect that the PR-619 treated animals would be long-lived because many proteins (such as EPS-8 and IFB-2) would not tend accumulate with age, or at least accumulate at a slower pace. Likely, the authors have run this obvious experiment by now and I encourage them to include it in the manuscript.
We have now assessed whether DUB inhibitor PR-619 prolongs lifespan (please see Extended Data Fig. 3h). Notably, we observed that the treatment with PR-619 at day 9 of adulthood for 24 h or from day 9 of adulthood until the end of the lifespan extends lifespan of wild-type worms.