Introduction

Aging is the progressive decline in an organism’s physiological functions that reduces its ability to maintain homeostasis, respond to stress and survive, thereby leading to increased probability of disease and death. Both stochastic and programmed events contribute to aging, and the lifespan of an organism is influenced by both its genetics and the environment. At present, the molecular mechanisms that contribute to longevity are incompletely understood.

To better understand the aging process, researchers have tried to identify genes that can increase lifespan and elucidate how these genes act to promote longevity. This research has greatly benefited from the use of genetic model organisms, primarily yeast, C. elegans, Drosophila and mice. Starting first with C. elegans, it has been shown in each of these organisms that mutations in a single gene can extend lifespan ^{1, 2, 3, 4, 5}. These long-lived genetic mutants have been divided into different pathways of lifespan extension based on which cellular functions are affected by the mutation.

Among the established pathways of longevity are: decreased insulin/IGF-1 signaling (daf-2 mutants) ¹; dietary restriction (eat-2 mutants) ⁶; decreased translation (ife-2 mutants) ^{7, 8}, germline ablation (glp-1 mutants) ⁹; mild impairment of mitochondrial function (clk-1, isp-1, nuo-6 mutants) ^{10, 11, 12}; decreased chemosensation (osm-5 mutants) ¹³; and increased mitochondrial superoxide (sod-2 mutants) ¹⁴.

In order to gain insight into the molecular mechanisms by which these genetic mutations increase lifespan, we and others have examined gene expression using either microarrays or RNA sequencing (RNA-seq) ^{15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25}. These previous studies have looked at gene expression in individual mutants or small groups of mutants and have used a variety of different experimental paradigms. As a result, it has not been possible to directly compare gene expression across a large panel of long-lived mutants, as it is unclear the extent to which differences observed would be due to differences in the experimental details or genuine differences between the strains.

In exploring the mechanisms underlying lifespan extension in long-lived mutants, it has become clear that at least some mechanisms of lifespan extension are shared across multiple long-lived mutants. For example, the p38-mediated innate immune signaling pathways is required for the extended longevity of daf-2, clk-1, isp-1, nuo-6, ife-2, eat-2 and osm-5 mutants ^{25, 26, 27}. The mitochondrial unfolded protein response is needed for the long life of clk-1, isp-1, nuo-6 and glp-1 mutants ^{17, 28}. An increase in reactive oxygen species contributes to the long lifespan of sod-2, clk-1, isp-1, nuo-6, daf-2, and glp-1 mutants ^{14, 29, 30, 31}. Disruption of the AMPK gene aak-2 decreases the long lifespan of daf-2, clk-1, isp-1 and glp-1 mutants ^{32, 33}. Decreasing the levels of the FOXO transcription factor DAF-16 reduces the long lifespan of several long-lived mutants ^{1, 6, 8, 13, 16, 34, 35}. However, it should be noted that the interpretation of these results is dependent on whether pathway disruption also decreases lifespan in wild-type worms and whether there is evidence of pathway activation ³⁶.

In this work, we sought to identify unique and shared genetic pathways contributing to lifespan extension in a panel of nine long-lived mutants. We found that there is a highly significant degree of overlap between differentially expressed genes in most pairs of long-lived mutants. Examining gene expression across all nine long-lived mutants reveals three distinct longevity groups. Interestingly, there are specific groups of genes that are significantly modulated in opposite directions in different longevity groups indicating that there are multiple effective strategies to achieve long life. In screening genes that are differentially expressed in multiple long-lived mutants, we find that individual differentially expressed genes can significantly affect both lifespan and resistance to stress.

Results

Magnitude of lifespan extension differs between long-lived mutants

In order to gain insight into shared and unique genetic pathways contributing to longevity in a panel of long-lived genetic mutants, we selected nine different genetic mutants to perform transcriptomic analysis using RNA-seq. Before proceeding with RNA-seq, we first wanted to confirm that all of these strains are long-lived and to compare the relative magnitudes of lifespan extension. The lifespan was performed in absence of FUdR as the RNA-seq samples would not be exposed to FUdR. Under these conditions, we confirmed that all nine strains lived significantly longer than wild-type worms (Figure 1). Further, we found that daf-2 worms had the greatest magnitude of lifespan extension followed by osm-5, nuo-6, isp-1, ife-2, eat-2, sod-2 and clk-1 worms (Figure 1)(glp-1 worms were not included in this list as the lifespan was performed under different conditions in which worms are developed at 25°C until adulthood). When we previously quantified the lifespan of these strains using 25 µM FUdR, we observed similar increases in lifespan where daf-2 worms lived the longest and osm-5, isp-1 and nuo-6 worms lived longer than eat-2, sod-2, clk-1 and ife-2 worms ²⁷.

Figure 1.

Having shown that all nine genetic mutants are long-lived, both with and without FUdR, we proceeded to analyze their gene expression by performing RNA-seq on pre-fertile young adult worms. We used a minimum of six biological replicates per strain. Principal component analysis (PCA) for each long-lived mutant demonstrated that all of the long-lived mutants could be clearly distinguished from wild-type worms based on gene expression (Figure S1-S9). We found that there are hundreds to thousands of differentially expressed genes in each of the long-lived mutant strains (Figures S1-9). In all strains, we observed both significantly upregulated and significantly downregulated genes (Figures S1-S9). Lists of differentially expressed genes from each long-lived mutant (Table S1) were used for enrichment analysis to identify pathways and functions that are overrepresented in each of the long-lived mutants individually (Figure S10-18).

We have compared the differentially expressed genes identified in this study to previous gene expression studies involving these long-lived mutant strains. In each case, we observed a highly significant degree of overlap (Figure S19-S25). We have provided a list of the overlapping genes between our study and previous studies in Table S2.

Expression of specific single genes at young adulthood is highly correlated with longevity

Having identified thousands of genes that are differentially expressed in long-lived mutants, we next sought to determine the extent to which expression of individual differentially expressed genes is associated with longevity. To do this, we compared the expression levels of each individual gene to their lifespan with and without FUdR. Using this analysis, we found that there are 853 genes whose expression levels are positively correlated with longevity measured in absence of FUdR, and 1459 genes whose expression levels are positively correlated with longevity when FUdR is present (Figure 2A; Table S3). Of these, there were 753 genes in common. In contrast, there were very few genes whose expression levels are negatively correlated with longevity (45 in absence of FUdR, 149 with FUdR; Figure 2B; Table S3).

Figure 2.

Individual genes exhibited R-squared values as high as 0.901 (p=0.0004) for positively correlated genes (Figure 2C) or 0.833 (p=0.0028) for negatively correlated genes (Figure 2D). Combined, this suggests that the expression level of individual genes at young adulthood can be predictive of lifespan.

Highly significant degree of overlap in differentially expressed genes between pairs of long-lived mutants

We next compared differentially expressed genes between pairs of long-lived mutants to determine the extent to which each mutant exhibited overlap with the other mutants and assess which long-lived mutants are most similar in gene expression. To determine the significance of each overlap, we compared the observed number of overlapping genes to the number of overlapping genes expected if two similarly sized gene sets were selected completely at random. We found that all of the long-lived mutants show a statistically significant overlap with multiple other long-lived mutants for both upregulated (Figure 3; Figure S26) and downregulated genes (Figure S26,S27). While genes differentially expressed in daf-2 and ife-2 worms exhibited a significant overlap with genes differentially expressed in all other long-lived mutants, genes differentially expressed in eat-2 worms showed relatively little overlap with other long-lived mutants (Figure 3; Figure S26,S27). Combined, these results suggest that there are common genetic pathways contributing to longevity shared across multiple long-lived mutants, but that some long-lived mutants have relatively unique changes in gene expression. This may represent the existence of different strategies to achieve long life.

Figure 3.

Long-lived genetic mutants segregate into three distinct longevity groups based on gene expression

Having observed a significant degree of overlap between differentially expressed genes between pairs of long-lived mutants, we next looked for overlaps across differentially expressed genes from multiple long-lived mutants. Heatmaps comparing gene expression across all of the long-lived mutants reveal that the long-lived mutants cluster into three distinct longevity groups (Figure 4A,B; Figure S28). The largest group, which we will refer to as group 1, contains daf-2, nuo-6, isp-1, clk-1, glp-1 and sod-2 worms. The second group, which we will refer to as group 2, contains osm-5 and eat-2 mutants. Finally, ife-2 worms exhibit large numbers of both upregulated and downregulated genes which show partial overlap with both group 1 and group 2 genes.

Figure 4.

Interestingly, some groups of genes upregulated in group 1 were found to be downregulated in group 2, while some groups of genes downregulated in group 1 were found to be upregulated in group 2 (Figure 4C,D). For example, of the 507 genes commonly upregulated in the group 1 mutants daf-2, isp-1, nuo-6 and glp-1, 188 of these genes are downregulated in group 2 mutants osm-5 and eat-2, while only 12 are also upregulated in group 2. Of the 498 genes that are commonly downregulated in the group 1 mutants daf-2, isp-1, nuo-6 and glp-1, 258 of these genes are upregulated in group 2 mutants osm-5 and eat-2, while only 2 are also downregulated in group 2. This suggests that group 1 and group 2 longevity mutants may be utilizing different strategies to live long, and that genetic pathways that affect longevity can be modulated in different ways to achieve long life.

To gain further insight into the genetic pathways that are modulated in different directions between group 1 and group 2 mutants, we examined the expression of individual genes from specific pathways known to affect longevity. We and others have demonstrated that the FOXO transcription factor DAF-16 contributes to lifespan extension in daf-2 worms, as well as other group 1 mutants including nuo-6, isp-1, clk-1 and glp-1 ^{1, 16, 37}. In examining the expression of high confidence DAF-16 target genes ³⁸, we found that genes that are activated by DAF-16 (Class I target genes) are upregulated in most of the group 1 longevity mutants, but either unchanged or decreased in group 2 mutants (Figure 5). Similarly, genes that are repressed by DAF-16 (Class II DAF-16 target genes) were found to be downregulated in group 1 longevity mutants but not in group 2 or group 3 mutants (Figure S29).

Figure 5.

We and others have also shown that the mitochondrial unfolded protein response (mitoUPR) is required for the extended longevity of group 1 longevity mutants nuo-6, isp-1, clk-1 and glp-1 ^{17, 28}. As with the DAF-16 target genes, we found that mitoUPR target genes are upregulated in most of the group 1 longevity mutants, but downregulated or unchanged in group 2 longevity mutants (Figure 6). At the same time, there were several genes that are significantly upregulated in group 2 longevity mutants but downregulated in group 1 longevity mutants (Figure S30). This included the PQM-1 target genes F55G11.2 and fat-7 ³⁸. Combined, this indicates that genetic pathways that affect longevity can be modulated in opposite directions to achieve lifespan extension (Table S4 provides a list of genes that are significantly upregulated in group 1 worms (daf-2, glp-1, isp-1, nuo-6, clk-1, and sod-2 worms) but downregulated in group 2 worms (osm-5 and eat-2 worms), and vice versa).

Figure 6.

To gain further insight into the pathways contributing to longevity in the long-lived mutants and how these differ between the three longevity groups, we performed transcription factor analysis on the RNA-seq data to determine which transcription factors might be driving the longevity-associated transcriptional changes. To do this we used two complementary approaches: (1) transcription factor inference, which is based on the coordinated expression changes of known transcription factors; and (2) motif enrichment analysis, which is based on identifying transcription factor binding motifs in the promoters of differentially expressed genes. After identifying which transcription factors were identified for each individual mutant (Table S5), we then compared the identified transcription factors across all nine mutants.

Interestingly, while 33 of the same transcription factors were implicated in group 1 and group 2 longevity mutants, 25 are modulated in different directions (activated in group 1, repressed in group 2 or vice versa) while only 5 are modulated in the same direction (Figure S31). This indicates that although group 1 and group 2 longevity mutants may modulate overlapping pathways to achieve long lifespan, in most cases these pathways are modulated in opposite directions.

Previous research has shown that aging is associated with widespread changes in gene expression³⁹. Accordingly, we compared gene expression in the long-lived mutants to gene expression changes that take place during aging. We found that all of the long-lived mutants exhibit changes in gene expression that were both the same direction as changes that take place during aging (e.g. upregulated in long-lived mutant and upregulated during aging or downregulated in long-lived mutant and downregulated during aging) and genes that were modulated in opposite direction as changes that take place during aging (upregulated in long-lived mutant and downregulated during aging or downregulated in long-lived mutant and upregulated during aging (Table S6). Interestingly, group 1 longevity mutants exhibited a greater number of transcriptional changes that were in the same direction as genes that are modulated during aging, while group 2 longevity mutants exhibit a greater number of changes that are in the opposite direction as genes that are modulated during aging (Figure S32). ife-2 mutants (group 3) exhibited a similar number of differentially expressed genes that were in the same direction and opposite direction as genes that are modulated during aging.

Genes that are commonly upregulated in the majority of long-lived mutants are involved in innate immunity, defense and metabolism

We next sought to determine the extent to which there are genetic pathways contributing to longevity that are shared across all or many of the long-lived mutants. We found that there were little or no genes that were differentially expressed in the same directions across seven or more long-lived mutants (Figure 7A, Table S7). There were 196 genes that were found to be significantly upregulated in six or more long-lived mutants. There were 362 genes that were found to be downregulated in five or more long-lived mutants. We performed Gene Ontology (GO) term and KEGG pathway enrichment on these sets of overlapping genes. Among the genes upregulated in seven or more long-lived mutants, there was an overrepresentation of genes involved in innate immunity, defense and metabolism (Figure 7B, Figure S33). This is consistent with our previous work highlighting the contribution of innate immune signaling and pathways of cellular resilience to longevity ^{16, 17, 26, 27, 36}. Genes downregulated in five or more long-lived mutants included genes involved in translation, gene expression, ribosome function and metabolism (Figure 7C, Figure S34). This is consistent with existing work showing that decreasing translation can extend lifespan ^{7, 8, 40}.

Figure 7.

Individual genes that are modulated in the majority of long-lived mutants are required for lifespan extension and resistance to stress

Having successfully identified genetic pathways contributing to longevity across multiple long-lived mutants, we next determined the extent to which individual differentially expressed genes contributed to lifespan extension in the long-lived mutants. To do this, we focused on the 196 genes that are significantly upregulated in six or more long-lived mutants. We rationalized that the upregulation of these genes could either be contributing to longevity, associated with longevity or could be acting to limit longevity. To distinguish between these possibilities, we planned to decrease the expression of these upregulated genes using RNA interference (RNAi) and then measure lifespan. Of these 196 genes, we tested 116 RNAi clones which were (1) in the Ahringer RNAi library; and (2) grew after one or two attempts. We decided to knock down these genes in either daf-2 or nuo-6 mutants, depending on which strain exhibited the greatest magnitude of upregulation, as these two strains provide a large window to see the effects of knockdown. We included daf-16 RNAi as a positive control in each experiment.

Among the genes tested in daf-2 mutants, we found that the majority of RNAi clones did not affect daf-2 lifespan but some RNAi clones were found that modestly increased or decreased longevity (Figure S35). For those RNAi clones that decreased lifespan, the magnitude of decrease was always much less than daf-16 RNAi. Similarly, we found that most of the RNAi clones targeting commonly upregulated genes in nuo-6 mutants had no effect on lifespan, but that there were clones that both increased and decreased nuo-6 longevity (Figure S36). In order to compare clones across the different lifespan trials, we calculated a weighted mortality score for each gene. This score was calculated as the mortality for the RNAi clone minus the mortality of empty vector all divided by the mortality of daf-16 RNAi minus the mortality of empty vector (Figure S37). Based on these mortality scores, we selected seven genes for confirmation in nuo-6 mutants and testing in wild-type: C08F11.7, ugt-62, K05C4.9, DC2.5, C05B5.5, T07C4.5 and W03B1.7. We found that RNAi targeting all of these genes decreased lifespan in both nuo-6 and wild-type worms (Figure 8B,C; Figure S38). This suggests that these genes are generally important for longevity and not specifically required for the extended longevity of nuo-6 worms.

Figure 8.

For further characterization, we decided to focus on three genes. These genes were selected based on the strength and reproducibility of the increase in weighted mortality in order to identify genes with a clear, consistent impact. C05B5.5 and T07C4.5 were ruled out because they had an inconsistent impact on weighted mortality. W03B1.7 was ruled out because it did not have a strong enough effect on weighted mortality. Finally, C08F11.7, ugt-62, and K05C4.9 were chosen as they had a greater impact on weighted mortality than DC2.5.

First, we examined the expression of these three genes across the panel of the long-lived mutants. We found that all three of these genes are upregulated in group 1 longevity mutants but not in group 2 or group 3 (Figure 9A). Based on this, we wondered if these genes are more important for the lifespan of group 1 longevity mutants than group 2 or group 3. Accordingly, we knocked down these genes in daf-2 (group 1), eat-2 (group 2) and ife-2 (group 3) worms and measured lifespan. Surprisingly, we found that RNAi targeting each of these genes did not decrease daf-2 lifespan (Figure 9B). In contrast, RNAi targeting C08F11.7 or ugt-62 decreased the lifespan of both eat-2 worms and ife-2 worms, while RNAi targeting K05C4.9 significantly decreased ife-2 lifespan and resulted in a trend towards decreased lifespan in eat-2 worms (Figure 9C,D). Combined, this suggests that C08F11.7, ugt-62, and K05C4.9 are generally affecting longevity and not specifically important for group 1 mutants.

Figure 9.

Overexpression of individual upregulated genes can increase stress resistance and lifespan

Having shown that knockdown of C08F11.7, ugt-62, and K05C4.9 can decrease lifespan, we next sought to determine the extent to which increasing expression of these genes could increase longevity. Accordingly, we generated novel strains that ubiquitously overexpress C08F11.7, ugt-62, and K05C4.9 under the rpl-28, rpl-2 and eft-3 promoters, respectively (these strains will be referred to as C08F11.7 OE, ugt-62 OE, and K05C4.9 OE worms). We used different ubiquitous promoters for each strain so that it would be possible to combine the strains together without transcription factor dilution. In measuring the lifespan of these strains, we found that overexpression of C08F11.7 significantly increased lifespan (Figure 10A), while ugt-62 overexpression caused a small decrease in lifespan (Figure 10B), and K05C4.9 overexpression decreased lifespan (Figure 10C). Interestingly, worms overexpressing both C08F11.7 and ugt-62 together exhibited a larger increase in lifespan than C08F11.7 overexpression worms alone (Figure 10D).

Figure 10.

Next, we decided to determine the extent to which overexpression of these three genes could enhance resistance to stress ⁴¹. We found that both C08F11.7 OE and ugt-62 OE worms have increased resistance to heat stress (35°C) compared to wild-type worms, while heat stress resistance was unaffected by overexpression of K05C4.9 (Figure 10E). In examining bacterial pathogen resistance, we found that C08F11.7 OE worms exhibit increased survival when exposed to P. aeruginosa strain PA14, while ugt-62 OE worms survive as long as wild-type worms (Figure 10F). Overexpression of C08F11.7 or ugt-62 both resulted in increased resistance to oxidative stress (4 mM paraquat; Figure 10G). In contrast, none of the overexpression strains exhibited increased survival when exposed to osmotic stress (450 mM NaCl, 500 mM NaCl; Figure 10H) or anoxia (72 hours; Figure 10I). Combined, this indicates that overexpression of C08F11.7 increases both stress resistance and lifespan while overexpression of ugt-62 increases resistance to stress.

Discussion

In this work, we compared gene expression changes across a panel of nine long-lived mutants. We found that these mutants exhibit significant overlap among their differentially expressed genes, but that there appear to be both common and opposing strategies to achieve long life. Genes involved in metabolism, immunity, translation and stress resilience were found to be central regulators of longevity. Individual genes were found to be predictive of lifespan and to be able to affect both lifespan and stress resistance individually.

This research extends previous single mutant transcriptomic studies by analyzing a panel of long-lived mutants under the same experimental conditions in order to facilitate direct comparison by reducing technical variation. We found that there are hundreds of genes whose expression at young adulthood exhibits a significant positive correlation with lifespan. The expression of these genes can be used as a predictor of longevity. Previous studies have reported genes that can act as predictors of lifespan ^{42, 43, 44, 45, 46}. This is consistent with research showing that changes during development or early adulthood are sufficient to increase lifespan ^{47, 48, 49, 50} and suggest that an organism’s aging trajectory can be set early in life.

Comparing gene expression across pairs of long-lived mutants demonstrated that most long-lived mutants share a significant number of differentially expressed genes with multiple other long-lived mutants. The number of overlapping genes that we observed may be an underestimate of the actual number of overlapping genes. When we previously examined gene expression by RNA-seq in the same strain with the same experimental paradigm but at different times, we observed that there were significant differences in gene expression ⁵¹. Thus, at least some of the differences between the long-lived mutants may be due to normal variations in gene expression measurement. Nonetheless, the fact that there are large numbers of genes that are differentially expressed in the same direction in multiple long-lived mutants suggests that at least some of the genetic pathways contributing to lifespan extension in these mutants are shared.

At the same time, the long-lived mutants also exhibited significant differences in gene expression. An important finding of this work is that the long-lived mutants examined cluster into three distinct longevity groups based on gene expression. This clustering suggests that extended lifespan can arise from different molecular strategies. Moreover, the observation that many genes that are upregulated in group 1 longevity mutants are downregulated in group 2 longevity mutants, and vice versa, indicates that the same pathways can be modulated in opposite directions to increase lifespan. The group 1 longevity mutants (daf-2, glp-1, clk-1, isp-1, nuo-6, sod-2) exhibit activation of DAF-16 and ATFS-1 target genes and the loss of these transcription factors has been shown to decrease their longevity ^{1, 16, 35}. The group 2 longevity mutants (eat-2, osm-5) exhibit upregulation of other genes, not upregulated in group 1 mutants, including PQM-1 target genes. ife-2 (group 3) mutants exhibit the greatest number of transcriptional changes, which exhibit overlap with changes in both group 1 and group 2 mutants.

While our data identify several genes that are regulated in opposite directions in group 1 and group 2 longevity mutants, we do not yet know the extent to which each of these genes contribute to the longevity of group 1 and group 2 mutants. The three genes that we focused on for further characterization (C08F11.7, ugt-62 and K05C4.9) are upregulated in group 1 longevity mutants but not group 2 mutants. Contrary to what might be expected, RNAi knockdown of these genes does not decrease the lifespan of the group 1 longevity mutant daf-2 but does decrease the lifespan of the group 2 longevity mutant eat-2. We recently reviewed the effect of different resilience pathways on the lifespan of long-lived genetic mutants.

Disruption of daf-16, sek-1, skn-1, hsf-1, ire-1 and trx-1 can decrease lifespan in both group 1 and group 2 longevity mutants, but also decreases lifespan in wild-type worms suggesting that at least in some mutants the effect on longevity may be non-specific ³⁶. Disruption of hif-1 does not affect the longevity of group 2 mutants, but does affect the lifespan of some group 1 mutants (clk-1, isp-1, nuo-6) but not others (daf-2, glp-1).

It is also unclear whether the longevity strategies utilized by group 1 and group 2 longevity mutants can operate simultaneously or whether they are mutually exclusive. To more definitively answer this question, it would be interesting to cross different combinations of group 1 and group 2 longevity mutants to see the extent to which different longevity groups synergize or potentially inhibit lifespan extension.

Among the genes commonly upregulated in multiple long-lived mutants, there was an enrichment for genes involved in innate immunity, defense and metabolism. Our group and others have previously demonstrated an important role for innate immune signaling in the extended longevity of long-lived mutants ^{25, 26}. We and others have also shown that various pathways of cellular resilience promote longevity in long-lived mutants ^{16, 17, 27, 36, 48, 52, 53}.

Genes that were found to be downregulated in multiple long-lived mutants were involved in translation and gene expression, anabolic processes that promote growth. Previous work has demonstrated that decreasing translation is sufficient to extend lifespan ^{7, 8, 40}. Together, these patterns support a model in which long-lived animals alter their metabolism to shift the balance from growth and biosynthesis toward defense and repair.

To move beyond correlation and test causation, we performed a targeted RNAi screen to examine the effect of knocking down genes that are upregulated in multiple long-lived mutants on lifespan. The majority of RNAi clones targeting commonly upregulated genes had no significant effect on longevity. This could be due to multiple reasons: (1) increased expression of the gene in long-lived mutants does not contribute to longevity; (2) increased expression of the gene contributes to longevity but the effect size is too small to detect when acting individually; (3) the level of knockdown achieved by the RNAi clone was not sufficient to inhibit gene function thereby resulting in no effect on longevity (we did not measure the level of knockdown achieved as we do not know what level of knockdown is need to affect function and it was unfeasible for this screen); or (4) the RNAi clone did not target the correct gene (the Ahringer RNAi library is known to have errors and we only confirmed the sequence of RNAi clones for positive hits).

Among the RNAi clones that significantly affected the longevity of daf-2 or nuo-6 worms, some clones increased lifespan and some decreased lifespan. We focused on the RNAi clones that decreased lifespan as this suggests that these genes are required for the longevity of these long-lived mutants. We selected seven genes to re-confirm in nuo-6 mutants and test in wild-type worms. We found that RNAi clones targeting each of the seven genes (C08F11.7, ugt-62, K05C4.9, DC2.5, C05B5.5, T07C4.5 and W03B1.7) significantly decreased both nuo-6 and wild-type lifespan. The fact that these genes also shorten wild-type lifespan suggests that these genes are acting as general pro-longevity factors and not specifically contributing to the extended lifespan of nuo-6 worms. Based on this conclusion, we examined the effect of overexpressing three of these genes in a wild-type background. We found that overexpression of C08F11.7 results in increased lifespan, thermotolerance, and resistance to oxidative stress and bacterial pathogens. In contrast, overexpression of K05C4.9 significantly decreased lifespan. This indicates that not all genes that are upregulated in long-lived mutants are acting to promote longevity. In the case of K05C4.9, increasing or decreasing its levels reduces longevity suggesting that its levels need to be tightly regulated to achieve a normal lifespan. In future work it would be interesting to examine additional healthspan parameters in these strains ⁵⁴.

Having quantified gene expression levels in nine different long-lived mutants, we developed a tool for the research community for examining expression of genes of interest across the panel of long-lived mutants in this study. This tool can be accessed at the website: https://vanraamsdonk.shinyapps.io/mutant_comparison_viewer/. This tool can be used to look up the expression of a gene or genes of interest in one or more of the long-lived mutants that we studied. The tool is online for everyone to use and can generate graphs comparing gene expression across long-lived mutants, indicate whether changes observed are statistically significant and provide access to the raw data. This tool can be used by researchers studying aging to test hypotheses about specific genes or genetic pathways that may be contributing to longevity. In addition to the online tool we generated for viewing our RNA-seq data, there are a number of excellent online tools that can be used to further analyse the data we generated in order to gain further insights into the aging process ⁵⁵.

Conclusions

In this work, we compared gene expression across a panel of long-lived genetic mutants. We found that these mutants share many gene expression changes suggesting that they employ a number of common genetic pathways to promote longevity. This includes pathways involved in cellular resilience and innate immunity. At the same time, the long-lived mutants also show unique patterns of gene expression that separates the mutants into three distinct longevity groups that in some cases modulate the same genetic pathways in opposite directions to achieve long life. While most individual differentially expressed genes do not exhibit a detectable effect on longevity, at least some of the differentially expressed genes can affect both lifespan and stress resistance individually. Overall, this work demonstrates that there are both shared and unique pathways contributing to lifespan extension in genetic long-lived mutants.

Materials and Methods

Strains

The strains used in this study included: N2 (wild-type), ife-2 (ok306), clk-1(qm30), sod-2(ok1030), eat-2(ad1116), osm-5(p813), nuo-6(qm200), isp-1(qm150), daf-2(e1370), glp-1(e2141), PHX6925 sybIs6925[eft-3p::K05C4.9, ges-1p::mCherry], PHX6945 sybIs6945[rpl-2p::ugt-62, myo-3p::mCherry] and PHX6868 sybIs6868[rpl-28p::C08F11.7, myo-2p::mCherry]. PHX6925, PHX6945 and PHX 6868 were generated by SunyBiotech. All strains were grown and maintained in nematode grown medium (NGM) plates at 20°C except glp-1(e2141), which was maintained at 20°C, but developed at 25°C for experiments. Plates were seeded with OP50 E. coli as a food source.

Lifespan

Lifespan analyses were performed on NGM plates seeded with OP50 bacteria. For most lifespan experiments no 5-fluoro-2’-deoxyuridine (FUdR) was used. In cases where FUdR was utilized, we used a low concentration of 25 µM, which we have previously shown not to affect wild-type lifespan and have minimal effects on other genotypes ⁵⁶. When FUdR was used, worms were transferred to the FUdR plates as prefertile young adults. Only adult lifespan was measured so that differences in development time would not affect the lifespan. During the lifespan study, worms were checked every 2-3 days until death. Worms with internal hatching or externalization of internal organs were removed from the study.

RNA isolation

RNA was isolated using TRIZOL as described previously ⁵⁷. Briefly, pre-fertile young adult worms were washed three times in M9 buffer and frozen in TRIZOL. After freeze-thawing three times, RNA was separated using chloroform, precipitated with isopropanol, and washed with 75% ethanol before being dissolved in RNAse free water. Isolated RNA was then frozen until library preparation and sequencing.

RNA sequencing and Bioinformatic Analysis

Library preparation and RNA sequencing was performed as previously described ¹⁵. RNA-seq data is available on NCBI GEO: GSE179825 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE179825 ^{58, 59}, GSE93724 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE93724 ¹⁶, GSE110984 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE110984 ¹⁷ and was analyzed by the Harvard School of Public Health Bioinformatics core for this paper as previously described ^{26, 58}.

Read mapping and expression level estimation

All samples were analyzed with an RNA-seq workflow provided by the bcbio-nextgen framework (https://bcbio-nextgen.readthedocs.org/en/latest/). Initial read quality was evaluated with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to verify that library preparation and sequencing produced data suitable for downstream processing. When required, adapter sequences, polyA contamination, and low-quality bases were removed using cutadapt (http://code.google.com/p/cutadapt/). Trimmed reads were aligned to the Ensembl build WBcel235 (release 90) of the C.elegans genome using STAR ⁶⁰. Alignment quality was reviewed by examining metrics such as coverage uniformity, rRNA levels, genomic distribution of mapped reads (including matches to annotated transcripts and introns), and overall library complexity. Transcript abundances were quantified with Salmon ⁶¹ to identify transcript-level abundance estimates and then collapsed down to the gene-level using the R Bioconductor package tximport ⁶². Principal component analysis (PCA) and hierarchical clustering verified that samples grouped appropriately by batch and mutant background.

Differential gene expression and enrichment analysis

Gene-level differential expression was assessed with the DESeq2 package in R ⁶³. For each comparison between mutant and wild-type samples, genes meeting a false discovery rate (FDR) cutoff of 0.01 were considered significant. When experiments spanned two sequencing batches, the model included a batch term to account for this source of variation.

Overlap analysis

Differentially expressed genes were divided according to whether they were up- or down-regulated and then compared across lists. The statistical significance of the overlaps was evaluated with a hypergeometric test.

Heatmap generation

For each dataset involving a long-lived mutant versus wild-type, raw counts were transformed using DESeq2’s regularized log (rlog) method ⁶³, which stabilizes variance across the expression range and enhances unbiased clustering.

Weighted Venn diagrams

Weighted Venn diagrams were prepared using BioVenn https://www.biovenn.nl/.

UpSetR plots

UpsetR plots were prepared using R. https://pmc.ncbi.nlm.nih.gov/articles/PMC5870712/

Gene Ontology

Gene ontology enrichment plots were prepared using ShinyGO 0.85.1 https://bioinformatics.sdstate.edu/go/.

Overlap between gene sets

The significance of overlap between lists of differentially expressed genes was determined by calculating the Jaccard Index (number of overlapping genes/total number of genes) and significance was assessed using Fisher’s exact test.

Targeted RNAi lifespan screen

For the targeted RNAi lifespan screen, there were RNAi clones for 154 of the 196 genes upregulated in 6 or more long-lived mutants. We found that several of these clones did not grow after two attempts leaving 116 clones. 52 of these clones were tested in daf-2 mutants and 63 were tested in nuo-6 mutants. For each lifespan trial, daf-2 or nuo-6 worms were synchronized by bleaching. The resulting eggs were placed on RNAi plates to start the lifespan study. Worms were transferred to new plates as necessary prior to offspring developing to adulthood. Survival of the worms was assessed approximately every 5 days.

Heat stress assay

Prefertile young adult animals were transferred to a 35°C incubator on NGM plates seeded with OP50 bacteria. Survival was measured every hour beginning at 4 hours. Three biological replicates with 20 worms per replicate were performed.

Bacterial pathogen stress assay

Resistance to bacterial pathogens was assessed by exposure to P. aeruginosa strain PA14 using a modified slow kill assay as we have done previously ^{26, 59}. An overnight culture of P. aeruginosa strain PA14 was seeded onto NGM plates containing 20 mg/L FUdR. After growth at 37°C overnight and then room temperature overnight, day 3 adult worms, which had been grown on plates containing FUdR from adulthood, were transferred to the plates seeded with PA14. Survival was monitored daily with plates incubated at 20°C. Four biological replicates with 30-60 worms per replicate were performed.

Osmotic stress assay

Worms at day 1 of adulthood were placed in 450 mM and 500 mM NaCl NGM plates seeded with 200 µL 5x OP50 bacteria⁶⁴. Survival was scored after 48 hours of incubation at 20°C. Three biological replicates with 20 worms per replicate were performed.

Paraquat assay

To assess chronic oxidative stress resistance, day 1 adult worms were placed in NGM plates with 4 mM Paraquat and 100 µL FUdR, prepared 3 days before the experiment⁶⁵. The plates were seeded with 200 µl of 10X OP50 bacteria one day before the start of the experiment, protected from the light as they dried overnight. 20 worms were picked onto each plate and scoring was done every 2-3 days until death. Three biological replicates were performed.

Anoxia assay

Day 1 adult animals were placed in NGM plates with 50 µM FUdR, seeded with 2x OP50. After picking the worms into the plates, the plates were placed in a BD GasPak™ EZ Anaerobe pouch system, which was tightly closed after the addition of the gas pack. Animals were placed on incubation at 20°C for 72 hours, and then the pouch was opened to allow reoxygenation. Survival was scored after 24 hours of reoxygenation at 20°C. Three biological replicates with 20 worms each were performed.

Statistical Analysis

Statistical analysis was performed using Graphpad PRISM Version 9.4.1. For lifespan and survival analyses, a log-rank test was used. For comparisons of a single variable across multiple groups, a one-way ANOVA with Dunnett’s multiple comparisons test was used. Error bars indicate standard error of the mean. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Supplemental Figures

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Figure S7.

Figure S8.

Figure S9.

Figure S10.

Figure S11.

Figure S12.

Figure S13.

Figure S14.

Figure S15.

Figure S16.

Figure S17.

Figure S18.

Figure S19.

Figure S20.

Figure S21.

Figure S22.

Figure S23.

Figure S24.

Figure S25.

Figure S26.

Figure S27.

Figure S28.

Figure S29.

Figure S30.

Figure S31.

Figure S32.

Figure S33.

Figure S34.

Figure S35.

Figure S36.

Figure S37.

Figure S38.

Data availability

RNA-seq data has been deposited on GEO: GSE179825, GSE93724, GSE110984. All other data and strains generated in the current study are included with the manuscript or available from the corresponding author on request. Material requests should be addressed to Jeremy Van Raamsdonk.

Acknowledgements

Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P30 OD010440). We would also like to acknowledge the C. elegans knockout consortium and the National Bioresource Project of Japan for providing strains used in this research.

Additional information

Author Contributions

Conceptualization: JVR. Methodology: ZDR, JG, ATG, GB, SKS, UA, MM, MMS, JVR. Investigation: ZDR, JG, ATG, GB, SKS, UA, MM, MMS, JVR. Visualization: ZDR, JG, ATG, GB, SKS, MM, JVR. Writing – original draft: JVR. Writing – review and editing: ZDR, JG, ATG, GB, SKS, UA, MM, MMS, JVR. Supervision: JVR.

Funding

This work was supported by the Canadian Institutes of Health Research (CIHR; http://www.cihr-irsc.gc.ca/; JVR) and the Natural Sciences and Engineering Research Council of Canada (NSERC; https://www.nserc-crsng.gc.ca/index_eng.asp; JVR). JVR was the recipient of a Research Scholar career award from the Fonds de Recherche du Québec Santé (FRQS) and Parkinson Quebec. SKS and ZDR were supported by training awards from the FRQS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Funding

Canadian Institutes of Health Research (CIHR)

• Jeremy Van Raamsdonk

Natural Sciences and Engineering Research Council of Canada (NSERC)

• Jeremy Van Raamsdonk

Additional files

Table S1.

Table S2.

Table S3.

Table S4.

Table S5.

Table S6.

Table S7.