Wide-Scale Comparative Analysis of Longevity Genes and Genetic Interventions

Hundreds of genes have been identified as being involved in the control of lifespan in the four common model organisms (yeast, worm, fruit fly and mouse). A major challenge is to determine if longevity-associated genes (LAGs) are model-specific or may play a universal role as longevity regulators across diverse taxa. A wide-scale comparative analysis of the 1,805 known LAGs across 205 species revealed that (i) LAG orthologs are substantially over-represented, from bacteria to mammals, especially noted for essential LAGs; (ii) the effects on lifespan, when manipulating orthologous LAGs in different model organisms, were mostly concordant, despite of a high evolutionary distance between them; (iii) the most conserved LAGs were enriched in translational processes, energy metabolism, development, and DNA repair. The least conserved LAGs were enriched in autophagy (Fungi), G-proteins (Nematodes), and neuroactive ligand-receptor interactions (Chordata). The results also suggest that antagonistic pleiotropy is a conserved principle of aging.

was also supported by studies on specific LAGs or pathways such as Foxo (Martins et  As seen in Fig. 1, for the vast majority of InParanoid species, the fraction of 13 conserved genes was significantly higher for LAGs than for the entire proteome of the 14 same model organism. The few exceptions were fringe cases where the baseline 15 orthology is either very high (phylogenetically very close species, for example, C. 16 elegans and C. briggsae), or very low (phylogenetically very distant species, for 17 example, M. musculus and K. cryptofilum) (Suppl. Table 1).

19
Remarkably, despite the high diversity of the species under analysis, the ratio between 20 the LAG orthologs and the orthologs of the entire proteome was relatively constant 21 along the evolutionary axis (Fig. 2). This could indicate that the high conservation of 22 LAGs is relatively independent of evolutionary distance.  Table 1). The   Table 1).

5
The species (X-axis) are ordered in descending order of orthology percentage for the LAGs.

11
It should be taken into account that genes that a priori have orthologs in humans and 12 are involved in basic biological processes or major diseases are more often tested for 13 their potential effect on lifespan. Despite this obvious bias, an important observation 14 is that among the model organisms examined, the highest conservation ratio was   Using post developmental interventions such as RNAi has allowed for discovering 1 worm longevity regulators that could not be discovered otherwise, because their pre-  One of the strong features of InParanoid is that it provides the best balance between 17 sensitivity and specificity (Chen et al. 2007). Yet, the proteomes found in the . Therefore, the BioGrid data could serve as an additional, high quality control 24 for a more rigorous testing of the evolutionary conservation of LAGs. For this 25 purpose, we used the interactomes of S. cerevisae, C. elegans, and D. melanogaster. 1 As seen in Suppl. Fig. 1, the same trend of over-conservation of LAGs was also 2 observed in comparison to the BioGrid control. Mouse was not included in the 3 analysis because its BioGrid gene list still contains a relatively small portion of the 4 entire genome and thus could not provide a reliable control.

6
Altogether, the results clearly show a high evolutionary conservation of LAGs across 7 distant species. With regard to this, a question arises as to whether this observation is 8 attributed to an enrichment of specific categories that are known to be strongly 9 preserved in the course of evolution. From the available data on gene and protein 10 annotations for the four model species we noted that LAGs are enriched in genes that 11 belong to categories known to be extraordinarily conserved in evolution, such as the 12 ribosomal or mitochondrial genes (Suppl . Tables 2 and 3). However, exclusion of 13 LAGs belonging to these categories from the analysis had almost no impact (Suppl.  LAG is highly evolutionary conserved, it does not automatically translate to its role in 24 a public mechanism of aging. In fact, in order to draw conclusions on public or 25 private mechanisms from the presence or lack of orthologs, one must (i) have a 1 context on the mode of operation of a given protein as its function could differ 2 between species; or (ii) compare groups of proteins belonging to a certain pathway or 3 category, so that generalized assumptions may be made. The data used in this study 4 only allows for the second approach. Thus, we comprised lists of proteins under 5 different conservation criteria, e.g. proteins that have orthologs across at least 12 6 phyla or have orthologs in a limited number of taxa only (for more details, see Suppl. 7 Table 4). As shown in Fig. 3, LAGs are generally conserved over more phyla than the    To get further insight into the universality of longevity-associated pathways, we 1 carried out an enrichment analysis for LAGs that are conserved across a large number 2 of phyla ("public") and those that are specific to certain taxonomic group(s) 3 ("private"). For the "private" analysis, we used the phylum of the corresponding 4 model organim (as depicted in the Suppl. Table 5), since smaller taxonomic groups 5 did not yield statistically conclusive results. Full detailed enrichment analysis is 6 available in Suppl. Table 6.  Table 6); the results from the yeast and worm 17 models were more significant and thus more reliable.    higher than the discordant ones (p < 0.003). That is, manipulation of LAGs has, more 23 than often, the same effect in different species (Fig. 4, Suppl. Table 7). A substantial 24 portion of the genetic interventions in yeast and worms could not be clearly defined as 25 concordant or discordant with other model organisms (Fig. 4a, green), mostly due to a 1 major difference in the methods of evaluation. In particular, in yeast studies, signs of 2 premature aging and a reduction of lifespan could actually be just the results of  When looking at pairwise comparisons (Fig. 4B), it is evident that the level of 6 concordancy is very high for some pairs of species (for example, M. musculus and D.  Fig. 3, the observed discordancy could not be explained by sequence dissimilarity.

12
One of the possible explanations for the observed discordancy is that in these cases 13 orthologous LAGs were discovered by interventions which greatly differ from one 14 another (e.g., knockout and overexpression). As such, if a given LAG is knocked-out 15 and as a result the animal ages more rapidly, that LAG is defined as a "pro-longevity" 16 gene; however, an overexpression of the same LAG will not necessarily increase 17 lifespan. For example, a knockout of G protein, alpha subunit (gpa-9) in C. elegans 18 increases maximum lifespan by up to 50%, but paradoxically, its overexpression also   . Table 7). Altogether, the results indicate a clear trend of concordancy in the 5 effects of LAG manipulations across model species despite a high evolutionary 6 distance between them, and the observed cases of discordancy could mostly be 7 attributed to technical rather than biological issues.  preserved role in longevity. Yet, we also observed LAGs that are highly conserved 17 only in a limited number of taxa, or that displayed discordant effects when tested in 18 more than one species, which could be attributed to "private" mechanisms of aging.

19
Definitely, more comparative studies are warranted to better discriminate between    The analysis was performed for 205 species (all species available excluding parasites; 24 for a full list, see Suppl. Table 1). The ortholog extraction was performed Suppl. Table 5.   Suppl. Fig. 3. Conservation index (CI) compared to concordancy of longevity effects. Each 2 dot represents a pairwise comparison between orthologs of LAGs that were tested for their 3 effect on longevity in more than one model species. The conservation index is the pairwise 4 alignment score normalized to the protein amino acid length. 5 6 Suppl. Fig. 4. Method similarity score compared to concordancy of longevity effects. Each 7 dot represents a pairwise comparison between orthologs of LAGs that were tested for their 8 effect on longevity in more than one model species. The method similarity score was 9 determined as: 0 = interventions of opposite directions (e.g. knockout and overexpression); 10 1 = intervention of the same direction but with varied methods (e.g. knockout and RNAi); 2 = 11 interventions that are identical or very close to identical (e.g. knockout and knockout). 12