Advertisement

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Long-lived rodents reveal signatures of positive selection in genes associated with lifespan

Abstract

The genetics of lifespan determination is poorly understood. Most research has been done on short-lived animals and it is unclear if these insights can be transferred to long-lived mammals like humans. Some African mole-rats (Bathyergidae) have life expectancies that are multiple times higher than similar sized and phylogenetically closely related rodents. To gain new insights into genetic mechanisms determining mammalian lifespans, we obtained genomic and transcriptomic data from 17 rodent species and scanned eleven evolutionary branches associated with the evolution of enhanced longevity for positively selected genes (PSGs). Indicating relevance for aging, the set of 250 identified PSGs showed in liver of long-lived naked mole-rats and short-lived rats an expression pattern that fits the antagonistic pleiotropy theory of aging. Moreover, we found the PSGs to be enriched for genes known to be related to aging. Among these enrichments were “cellular respiration” and “metal ion homeostasis”, as well as functional terms associated with processes regulated by the mTOR pathway: translation, autophagy and inflammation. Remarkably, among PSGs are RHEB, a regulator of mTOR, and IGF1, both central components of aging-relevant pathways, as well as genes yet unknown to be aging-associated but representing convincing functional candidates, e.g. RHEBL1, AMHR2, PSMG1 and AGER. Exemplary protein homology modeling suggests functional consequences for amino acid changes under positive selection. Therefore, we conclude that our results provide a meaningful resource for follow-up studies to mechanistically link identified genes and amino acids under positive selection to aging and lifespan determination.

Author summary

As an adaption to different environments rodents have evolved a wide range of lifespans. While most rodents are short-lived, along several phylogenetic branches long-lived species evolved. This provided us a unique opportunity to search for genes that are associated with enhanced longevity in mammals. Towards this, we computationally compared gene sequences of exceptional long-lived rodent species (like the naked mole-rat and chinchilla) and short-lived rodents (like rat and mouse) and identified those which evolved exceptional fast. As natural selection acts in parallel on a multitude of phenotypes, only a subset of the identified genes is probably associated with enhanced longevity. Applying several tests, we ensured that the dataset is related to aging. We conclude that lifespan extension in rodents can be attributed to changes in their defense against free radicals, iron homeostasis as well as cellular respiration and translation as central parts of the growth program. This confirms aging theories assuming a tradeoff between fast growth and long lifespan. Moreover, our study offers a meaningful resource of targets, i.e. genes and specific positions therein, for functional follow-up studies on their potential roles in the determination of lifespan–regardless whether they are currently known to be aging-related or not.

Citation: Sahm A, Bens M, Szafranski K, Holtze S, Groth M, Görlach M, et al. (2018) Long-lived rodents reveal signatures of positive selection in genes associated with lifespan. PLoS Genet 14(3): e1007272. https://doi.org/10.1371/journal.pgen.1007272

Editor: Gregory S. Barsh, Stanford University School of Medicine, UNITED STATES

Received: January 9, 2018; Accepted: February 22, 2018; Published: March 23, 2018

Copyright: © 2018 Sahm et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: Read data that was used for assemblies, i.e. for Ansell’s mole-rat, giant mole-rat, silvery mole-rat and greater cane rat assemblies, was deposited as ENA (European Nucleotide Archive) study PRJEB20584. Read data that was used for expression analysis during aging in naked mole-rat and laboratory rat was deposited as GEO (Gene Expression Omnibus) series GSE98746. The results of the genome-wide analyses of positive selection and gene expression during aging are provided via the supplement tables. Visualizations of the underlying alignment data along with the amino acid sites under positive selection are available via our ftp server (ftp://genome.leibniz-fli.de/pub/mrps2017/supplement_data.tar.gz) and from the Dryad Digital Repository: https://doi.org/10.5061/dryad.75b406n.

Funding: This work was funded by the Deutsche Forschungsgemeinschaft (DFG, PL 173/8-1 and DA 992/3-1), the European Community’s Seventh Framework Programme (FP7-HEALTH-2012-279281), as well as the Leibniz association (SAW-2012-FLI-2). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Most of the available information about the genetic mechanisms that govern lifespan and aging were obtained by studying single-gene mutations in invertebrates or short-lived, highly inbred vertebrate species. However, it is not clear whether insights about aging relevant genes and pathways gained from these species can be applied to long-lived species like human [1]. In addition, lifespan extensions under artificial laboratory conditions resulting from single gene mutations or other genetic, pharmacologic and/or lifestyle interventions are far smaller than natural variation of lifespan among species shaped by natural selection. Maximum lifespan of vertebrates varies about two orders of magnitude and is positively correlated with body mass [2, 3]. Therefore, comparative evolutionary approaches that search for genetic differences between closely related species that are long- and short-lived with respect to their body mass may reveal novel candidate genes and pathways or open new perspectives on known ones, e.g. by identifying amino acid sites under positive selection that are of potential functional relevance.

Rodents are an ideal taxon for such an approach. While the majority of species is short-lived, such as mice, rats and hamsters, there are long-lived exceptions, such as chinchillas, blind mole rats (Spalax sp.) and several African mole-rat species including the naked mole-rat (Heterocephalus glaber) [4, 5]. Furthermore, genome and transcriptome sequences of long- and short-lived species are available and can be used for comparative analysis.

African mole-rats (family Bathyergidae) are subterranean rodents that feed from roots and tubers. The family comprises six genera; for five out of these, maximum lifespan records are available for at least one species. Notably, and in contrast to most other rodents, all of these species have a maximum lifespan of above ten years and exceeding the predictions of the power-law that describes body mass/lifespan relationships in mammals [5]. At the extreme of this distribution, Zambian mole-rats from the Fukomys micklemi clade [6] with its best studied representative Ansell´s mole-rat (F. anselli) and the giant mole-rat (F.mechowii), as well as naked mole-rat, have maximum lifespans of at least ca. 20, 22 and 31 years, respectively. These values are 212%, 194% and 368% with respect to the predicted lifespan based on their body mass ([4], giant mole-rat percentage calculated with own lifespan data and same formula). In contrast, the established biomedical model organisms mouse (Mus musculus) and rat (Rattus norvegicus) have a maximum lifespan of 3.8 and 4 years, respectively, which is 51% and 32% of the predicted value. Remarkably, the greater cane rat (Thryonomys swinderianus) that is closely related to the African mole-rats reaches only 28% of the predicted maximum lifespan (Fig 1).

Due to a number of unique phenotypes, the naked mole-rat became the focus of intensive research [7]. The naked mole-rat shows (i) the longest lifespan among rodents, (ii) minimal aging-related decline in reproductive and physiological parameters, as well as (iii) and an extremely low aging-related increase in mortality rate [8, 9]. Among thousands of examined animals only six recently discovered cases of spontaneous tumors have been described [10, 11]. Interestingly, cancer resistance is shared with blind mole rat, which is, despite its name, rather distantly related to African mole-rats (Fig 1). However, different mechanisms are proposed for cancer resistance in these two taxa.

thumbnail
Download:
Fig 1. Nucleotide-based phylogeny of the analyzed rodents.

Species or branches regarded in the present analyses as long-lived or leading to enhanced longevity, respectively, are depicted in bold. The branch numbers used in the text are shown in blue. The numbers of genes with signs of positive selection on the branches are colored in red. The first number after the species name shows the recorded maximum lifespan and the second number is the percentage of the observed vs. expected maximum lifespan based on the respective body mass. The maximum lifespans and ratios were taken from [4], except for silvery mole-rat (personal communication by R. Sumbera) and giant mole-rat (own data). For these two species, the expected maximum lifespans were calculated with the same mammalian allometric equation used by [4]. The scale bar represents 0.05 substitutions per site.

https://doi.org/10.1371/journal.pgen.1007272.g001

The search for signatures of positive selection represents a powerful approach to identify the genetic basis of phenotypes of interest. Positive selection is the fixation of an allele in a taxon driven by its positive effect on fitness. It has, however, to be taken into account that selection/adaptation–in particular over long evolutionary time scales–is driven by a multitude of environmental stimuli that affect a multitude of phenotypes. Therefore, a genotype-phenotype link cannot be provided by such analysis. It requires additional knowledge and, finally, experimental validation. Respectively, bioinformatic positive selection studies are primarily hypothesis-driven and hypothesis-generating research. In protein-coding sequences (CDSs), positive selection results in an increased rate of non-synonymous substitutions as compared to genetic drift. Statistical models based on the ratio of non-synonymous to synonymous substitution rates (dN/dS) are widely used in comparative genomics and allow the identification of specific amino acids within a given gene that changed due to positive selection [12–14].

Consequently, several studies performed genome-scale scans for positively selected genes (PSGs) in African mole-rats and blind mole rat. The first study [15] searched for PSGs on the very long naked mole-rat branch in a four-species comparison with human as an outgroup and the mouse and rat as further rodents. Among the 142 identified PSG candidates, three were members of a five-protein complex involved in alternative lengthening of the telomeres. The second study [16], used ten species with the guinea pig (Cavia porcellus) as most closely related species and scanned for PSGs along the branches leading to naked mole-rat, Damaraland mole-rat (Fukomys damarensis) and their last common ancestor (LCA), identifying 334, 179 and 82 candidates, respectively, including candidates associated with neurotransmission of pain in the naked mole-rat. A third study [17] used species from all six African mole-rat genera and searched the branch of the LCA of all African mole-rats that follows divergence from the guinea pig. Signs of positive selection were identified in 513 genes, including loci associated with tumorigenesis, aging, morphological development and sociality. All three studies suffer from a methodological limitation that is common in positive selection studies: in none of these, a closer related species than guinea pig was included. As guinea pig is not the closest relative of African mole-rats not expressing the phenotypes of interest, it cannot be excluded that fixation of detected signs of positive selection predates–and therefore could not contribute to–the evolution of these phenotypes [18]. A fourth study [19] examined the blind mole rat branch using the Chinese hamster (Cricetulus griseus) as the most closely related outgroup. Among the 48 PSG candidates, several were linked to necrosis, inflammation and cancer.

To better resolve the above-mentioned ambiguities and to achieve a higher resolution of positive selection in respect to the evolution of enhanced longevity along rodent phylogenetic branches, we analyzed genomic and transcriptomic data of 17 species–data from public sources and original data generated for this study. In particular, we generated genomic data for the greater cane rat as a key species absent from previous analysis and for the silvery mole-rat (Heliophobius argenteocinereus). We systematically scanned 11 evolutionary branches associated with the evolution of enhanced longevity (6 corresponding to extant species and 5 to ancestral branches). This approach enabled us to date more precisely the occurrence of signatures of positive selection on multiple evolutionary branches.

In addition, we recently observed that PSGs in short-lived and fast-growing killifish were significantly more often up- than down-regulated during aging [14]. This finding is consistent with the concept of antagonistic pleiotropy [20] suggesting that the same genes that are positively selected for fast growth and maturation at young age are drivers of aging at old age. The antagonistic pleiotropy hypothesis is well supported, e.g. by the fact that growth rate and lifespan are negatively correlated, both between species and within many species [5, 21]. Respectively, we generated RNA-seq data from young and old naked mole-rats and rats to analyze the regulation of PSGs during aging, as selection may act both on the sequences of proteins and on their pattern of expression [22].

We found the identified PSGs to be enriched for genes related to aging and to show an expression pattern fitting the antagonistic pleiotropy theory of aging. Moreover, PSGs are linked to functional terms relevant to aging, like cellular respiration, metal ion homeostasis, regulation by the mTOR pathway, inflammation and the antioxidant defense. We discuss the implications of our results on the current understanding and for future follow-up studies of the genetic basis of aging and lifespan.

Results and discussion

To gain new insights into the genetic mechanisms determining the lifespan in mammals, we performed a comparative genomic and transcriptomic study among long- and short-lived rodents. By searching for signs of positive selection on phylogenetic braches associated with the evolution of enhanced longevity, we aimed to provide a set of target genes/sites for future follow-up approaches to explore mechanistically their putative link to aging and lifespan determination. As natural selection acts in parallel on a multitude of phenotypes, we are aware that only a subset of the targets evolved under selection for enhanced longevity. To ensure that the dataset nevertheless represents a meaningful resource for aging research, we postulated four criteria for evaluation: (i) the PSGs show expression patterns during aging that are compatible with established theories of aging, (ii) the gene set is enriched for genes known to be aging-related, (iii) the gene set contains functional candidates for being relevant for aging but have not yet been associated therewith, and (iv) protein homology modeling of known aging-related genes suggests functional consequences for amino acid changes under positive selection.

As starting points for our analysis, we generated CDS libraries for five rodent species (four African mole-rat species and greater cane rat) based on transcriptomic and genomic data (S1/S2 Tables). Together with publicly available rodent CDS catalogs (S1 Table), we obtained data for 17 species, including several additional African mole-rats, chinchilla, blind mole rat and short-lived outgroups like the guinea pig, mouse and rat (Fig 1). From these sequences, we predicted orthologs and best matching isoforms between the species, calculated alignments and applied the branch-site test of positive selection to multiple branches [23].

Based on the lifespans of the extant species, we regarded six extant as well as five ancestral branches as leading to enhanced longevity and examined them for positive selection (Fig 1). We limited our analyses to those branches and did not consider a comparison to branches leading to short-lived rodents by two reasons. First, short/normal lifespan is widespread among rodents, and having a short lifespan is most likely the ancestral state in rodents. It is therefore much easier to identify phylogenetic branches on which lifespan was prolonged, e.g. within the mole-rat clade, than to identify branches on which lifespan was reduced. Second, there is an ongoing discussion that the same genes/pathways may be involved in the evolution of both short and long lifespan [24, 25], it is currently impossible to in silico predict whether positive selection in short- and long-lived species modulates gene functions in opposite directions.

In total, we detected 259 PSGs (false discovery rate (FDR) <0.1, branch-site test). Nine genes were found on multiple branches (S3 Table), resulting in a non-redundant set of 250 PSGs (S4 Table, S5–S15 Tables). Signs of positive selection for the same gene on multiple branches indicate possible parallel evolution. Among those, we found AMHR2 (anti-Mullerian hormone receptor type 2) to be positively selected both on branch 2 (naked mole-rat) and branch 11 (blind mole rat). While AMHR2 plays a role in male fetal development and in ovarian follicle development of the adult female [26], no function with regard to aging is described yet. However, the protein kinase domain of AMHR2 contains the greatest number of longevity-selected positions based on a regression analysis with 33 mammalian species [27]. This domain contains 3 of 8 and 2 of 3 positively selected sites on branch 2 (naked mole-rat) and branch 11 (blind mole rat), respectively.

As a number of sequence/alignment characteristics potentially leads to a higher sensitivity of the underlying test of positive selection and thus could bias subsequent enrichment analysis, we specifically tested whether the identified PSGs had higher taxon coverage (S1 Fig) or greater sequence lengths (S2 Fig) than all tested genes. Neither was the case. Furthermore, analyses of GC-composition showed no differences between PSGs and all tested genes (S3/S4 Fig).

Different studies on positive selection in mole-rats show minor overlaps

First, we compared our list of PSGs with the PSGs detected in previous studies of positive selection in mole-rats (S16 Table). As observed before, [17] PSGs from different studies show no or small overlaps. This is not surprising because the branches examined in previous studies differed from the branches examined in this study, even though some of them are named similarly. For example, Kim et al. examined a “naked mole-rat branch” using the house mouse as closest related species [15]. In our study, the sister taxon to naked mole-rat is represented by other African mole-rats and the house mouse is used only as an outgroup (Fig 1). In a similar way, the analysis of the African mole-rat ancestor by previous studies [17, 19] differs from ours as we incorporated the greater cane rat as closest related short-lived species and used guinea pig as an outgroup. We therefore analyzed evolutionary processes on a shorter phylogenetic distance that closely matches the appearance of the phenotypes under investigation. In addition, there are methodological differences between the studies, e.g. regarding ortholog prediction or alignment filtering. Unfortunately, the contribution of these technical variables to the discrepancies cannot be assessed as the alignments used for the previous studies are not available and cannot be compared with those generated and provided in our study (Supplement Data). Those five genes that were, despite the mentioned limitations, detected also by previous studies on the naked mole-rat branch (AMHR2, IMP4, MYBPHL, MPZL2, TACC2; S16 Table) can be considered as showing particular strong signals of positive selection.

Positive selection leading to enhanced longevity and age-related expression are linked

Next, we analyzed the regulation of PSGs during aging–as selection may act both on the sequences of proteins and on their pattern of expression [19]–to identify potential links between positive selection on the analyzed branches and genetic determinants of lifespan. In general, directionality analysis of gene regulation during aging is complicated by the fact that the direction itself is not informative, whether the respective gene function is either causing or counteracting aging. For example, up-regulation of a causative gene may accelerate aging and shorten lifespan while adaptive up-regulation to counteract aging phenotypes may extend longevity. Based on our findings that up-regulation of PSGs in short-lived species may cause aging [14], we hypothesized that selection for enhanced longevity is more compatible with attenuation of gene activity–either on the level of protein function or gene regulation–since avoiding damage is easier than improving repair.

Moreover, genetics of aging is highly complex in general, and we do not assume that enhanced longevity evolved along the analyzed branches in the same way. But a number of genes/pathways are shown to be consistently involved in aging of even very distantly related taxa from yeast to mammals [28]. In addition, there are accumulating data that this is even more the case the closer related the taxa are [29, 30]. On this basis we hypothesize that there is a considerable overlap between the genes/pathways that are involved in aging among analyzed branches.

To evaluate these hypotheses, we performed RNA-seq and subsequently compared gene expression in liver from old vs. young males of both long-lived naked mole-rats (>21 vs. 2–4 years) as well as short-lived rats (24 vs. 6 months; S17–S19 Tables). Indeed, the union set of PSGs across all examined branches showed during aging preference for down-regulation in naked mole-rat and for up-regulation in rats in respect to all regulated genes (p = 0.0029, Lancaster procedure [31]). Moreover, 68 PSGs were both down-regulated in the long-lived naked mole-rat and up-regulated in the short-lived rat during aging (Fig 2), resulting in a highly significant preference for quadrant I (down in naked mole-rat, up in rat; p = 0.0017, one-sided fisher test, quadrant I against the sum of II, III, IV). These results indicate that identified PSGs are associated with expression changes during aging of long- and short-lived rodents consistent with the antagonistic pleiotropy theory of aging.

thumbnail
Download:
Fig 2. Changes of PSG expression in liver during aging of long-lived naked mole-rat and short-lived rat.

The roman numbers describe the quadrant, the colored numbers below that show the number of PSGs in the respective quadrant and the black numbers at the bottom give the total regulated genes in the quadrant. The red marked quadrant (I) represents PSGs that were down regulated in naked mole-rat and up regulated in rat. This was tested against the sum of three blue marked quadrants (II, III, IV) with Fisher’s exact test (one-sided). The resulting p-value is shown in quadrant I. The total number of PSGs shown in this plot (185) is lower than the unique number of all PSGs (250) due to insufficient expression of genes (DEseq2) in at least one of the species.

https://doi.org/10.1371/journal.pgen.1007272.g002

Positively selected genes are enriched for functional terms related to aging

To test whether the PSGs are enriched for genes known to be aging-related, we performed gene ontology (GO) term enrichment analysis. Among 16,639 naked mole-rat genes that were analyzed in total with regard to their expression, 2,220 genes were down-regulated and 3,212 genes up-regulated during naked mole-rat aging. Regarding all genes, there was a significant perturbation towards down-regulation during aging in 159 GO terms while 8 terms were perturbed towards up-regulation (FDR<0.1, GAGE; S20 Table). The enriched 159 terms were summarized into 7 categories (REVIGO, S21/S22 Tables). Among the three by far largest categories based on the number of covered genes are “cellular respiration” (GO:0045333, rank one), “translation” (GO:0006412, three) previously linked to aging (see below). While the other process “SRP-co-translational protein targeting to the membrane” (GO:0006614, two) may be related to aging as it is functionally connected to “translation”.

With respect to possible pleiotropic effects, translation and cellular respiration are also key components of the growth program. To evaluate the PSGs in respect to these categories, we built the union of genes for each REVIGO category and tested for overrepresentation of PSGs. Regarding all PSGs, there was a significant overlap with “cellular respiration” (p = 2.3*10−4, one-sided fisher test). Regarding only the 68 PSGs that were down-regulated in naked mole-rat and up-regulated during rat aging (quadrant I, Fig 2), “cellular respiration” showed a significant (p = 2.4*10−7) and “translation” a borderline significant enrichment (p = 0.10; S23 Table). This again suggests that PSGs are linked to aging relevant processes in an antagonistic pleiotropic way. The result is also consistent with the hyperfunction theory of aging that suggests that antagonistic pleiotropy works via a mechanism of “perverted” growth. According to this theory the growth program that is beneficial during youth is not entirely stopped after finishing development and causes damage from that point on. The theory further claims that the master regulator mTOR (mechanistic target of rapamycin) governs this growth program [32, 33].

Regulation of mTOR- and downstream processes show signs of positive selection leading to enhanced longevity

The kinase mTOR operates as a central regulator of cell metabolism, growth, inflammation and proliferation. It was identified as a key regulator of aging and aging-related diseases in yeast, nematodes, fruit flies, and mice [34, 35]. On branch 2 (naked mole-rat), we found RHEB (Ras homolog enriched in brain) coding for the direct, positive regulator of mTOR and on branch 9 (Ansell’s mole-rat) its paralog RHEBL1 to be positively selected, a situation consistent with the concepts of parallel evolution as well as of subfunctionalization of genes after duplication. mTOR can be activated by RHEB either on the surface of the peroxisome [36]–in response to reactive oxygen species (ROS)–or on the surface the lysosome [37]–in response to amino acids. We found peroxisomal genes (GO:0005777) enriched for PSGs both in the union across all examined branches (FDR = 0.026) as well as on branch 1 (LCA of the African mole-rats, FDR = 0.002). With regard to the lysosome, the major lysosomal membrane component LAMP2 (lysosomal associated membrane protein 2), was identified as PSG on branch 11 (blind mole rat) and has a tendency towards positive selection on branch 2 (naked mole-rat, FDR = 0.11). LAMP2 is estimated to contribute together with its paralog LAMP1 about 50% of all lysosomal membrane proteins.

Despite that mTOR is partially regulated at the lysosome, it is also a key regulator of autophagy [38]. Autophagy is a cellular protective cleaning mechanism, required for organelle homeostasis, especially mitochondria. While enhanced autophagy was shown to be associated with lifespan extension in worms, flies and mice, inhibition of autophagy, conversely, leads to premature aging in mice [39]. LAMP2 (see above) acts also as a receptor for chaperone-mediated autophagy. It is required for degradation of individual proteins through direct import into the lysosomal lumen [40, 41]. Aging-dependent decrease of LAMP2 expression was observed in mouse liver. Reinstatement of juvenile LAMP2 levels in aged mice significantly reduces aging-dependent decline of cell function and restores the degree of cell damage to that found in young mice [42].

Besides the lysosome, another cellular protein quality control and degradation system is the proteasome. While impaired proteasome function and subsequent accumulation of misfolded proteins were tightly correlated with aging and aging-related neurodegenerative disorders like Parkinson’s and Alzheimer’s disease, long-lived humans have sustained proteasome activity [43–45]. Two proteasome subunit genes, PSMG1 (proteasome assembly chaperone 1) and PSMB4 (proteasome subunit beta 4), were identified as PSGs on branch 11 (blind mole rat). PSMB4 has been classified as a driver for several types of tumors [46], is down-regulated during naked mole-rat aging (FDR = 0.088; DESeq2) and is a known interaction partner of PRP19 (pre-mRNA-processing factor 19 or senescence evasion factor) that is essential for cell survival and DNA repair [47].

Among mTOR-regulated processes that are relevant for both growth and aging are translation and cellular respiration [34]. Consistent with the observed antagonistic expression patterns of PSGs in the long-lived naked mole-rat and short-lived rat (see above), lower expression of genes related to these processes as well as pharmacological inhibition of the respective gene products was shown to be associated with longer lifespan in worms [48, 49], killifishes [50] and mice [51, 52].

Finally, mTOR is thought to play a critical role in regulating inflammatory and immune responses [53]. We found inflammatory response (GO:0006954; FDR = 0.027, Fisher’s exact test) and defense response (GO:0006952, FDR = 0.004) to be enriched for PSGs on branch 11 (chinchilla). Aging is tightly associated to the delicate balance between pro-inflammatory responses to resist potentially fatal infections and the inexorable damages that are accumulated by this [28, 54]. Chronic inflammation is described as a major risk factor for aging and aging-related diseases such as atherosclerosis, diabetes, Alzheimer's disease, sarcopenia and cancer [55].

IGF1 (insulin-like growth factor 1), a central regulator of the insulin/IGF1 pathway, was identified as PSG on branch 1 (LCA of African mole-rats). The insulin/IGF1 pathway is another aging relevant signaling pathway that links nutrient sensing to various anabolic and catabolic aspects of the metabolism [56]. Similar as for the mTOR-pathway, reduction of the insulin/IGF1 signaling increases longevity in worms, flies and mice and both pathways are linked by mutual feedback loops [34]. Furthermore, IGF1 is down-regulated during naked mole-rat aging (FDR = 1.6*10−5).

Positive selection leading to enhanced longevity affects regulation of oxidative stress

With cellular respiration and peroxisome activity the main cellular sources of ROS were found to be affected by positive selection (see above). In small concentrations ROS can serve as signaling molecules, e.g. in regulation of mTOR [36] or apoptosis [57]. In higher doses, however, they can cause negative oxidative stress [58], i.e. damages to DNA, proteins and other cellular components [59]. Oxidative stress is thought to play a major role in the pathogenesis of neurodegenerative diseases [60] and even the determination of lifespan in general (“oxidative stress theory of aging”) [61]. On branch 3 (LCA of all African mole-rats except naked mole-rat), we found an enrichment of oxidoreductase activity (GO: GO:0016491; FDR = 0.056) and positive selection of TXN (thioredoxin), coding for an oxidoreductase enzyme that acts as an antioxidant extending lifespan in fly [62] and potentially also in mice [63, 64]. SOD2 (superoxide dismutase 2) and CCS (copper chaperone for superoxide dismutase) are PSGs on branch 10 (chinchilla) and branch 2 (naked mole-rat), respectively. Both genes are involved in ROS defense and affect aging/lifespan in several species [65, 66]. This is interesting because in recent years, it has been repeatedly questioned that the oxidative stress theory of aging has much relevance for bathyergid rodents, given that several studies failed to find improved antioxidant capacities and/or less accumulation of oxidative damage in naked mole-rats compared to the much shorter-lived mice [67–69]. On the other hand, significantly higher levels of oxidative damage on proteins and lipids in non-reproductive as compared to reproductive females of the Damaraland mole-rat were found [70]. Since non-reproductive individuals live shorter (and hence age faster) than their reproductive counterparts in Fukomys sp. [71–73], these results are consistent with the oxidative stress theory of aging. The diverse signs of positive selection on branch 2 (naked mole-rat), 3 (LCA of all African mole-rats except naked mole-rat) and 7 (LCA of Ansell’s mole-rat and giant mole-rat) may suggest that the impact of oxidative stress on aging differs between naked mole-rat and other African mole-rats.

ROS production and ROS-induced damage to biomolecules are intertwined with the formation of advanced glycation end-products (AGEs). AGEs are stable bonds between carbohydrates and proteins/lipids which are formed in a non-enzymatic fashion. AGEs activate membrane-bound or soluble AGER (AGE specific receptor) and AGEs/AGER have been linked to several aging-related diseases including Alzheimer’s disease and diabetes [74]. Interestingly, AGER was found to be a PSG on branch 10 (chinchilla). The role of AGEs/AGER in aging is complex and ambivalent [75]. AGER is up-regulated in liver during naked mole-rat aging (FDR = 0.035). Similarly, in skin AGE levels rise with chronological age in Ansell’s mole-rat, but surprisingly are higher in the skin of slow aging breeders than of faster aging non-breeders [76]

Positive selection leading to enhanced longevity affects metal ion homeostasis and transport

The majority of ROS is generated under participation of redox-active metals [77]. Therefore, disruptions of the metal ion homeostasis are thought to contribute to formation of free radicals. Furthermore, imbalances of metal ions, especially elevated iron levels, were associated with the pathogenesis of aging-related neurodegenerative diseases [78, 79]. Metal ion transport genes were found to be enriched for PSGs on branch 8 (giant mole-rat, GO:0030001, FDR = 0.025) and TF (transferrin) was identified as PSG on branch 4 (LCA of Cape, Cape dune, giant, Ansell’s mole-rat and common mole-rats). TF is an iron-binding protein responsible for transport of iron in the bloodstream and therefore essential for iron homeostasis [80]. Neurons regulate iron intake via the TF receptor and dysregulation of this tightly controlled process in the brain was shown to be highly associated with Parkinson’s and Alzheimer’s disease [81].

Protein homology modeling suggests functional consequences of amino acid changes under positive selection

Our positive selection analysis provides not only candidate genes but also candidate amino acids for functional follow-up studies. Protein homology modeling may reveal insights into the potential structural impact of a predicted positively selected amino acid change. As a proof of concept, we performed homology modeling for the site of highest probability of selection in TF–Ser383Lys. Serum TFs form a bilobal structure, and each lobe contains two dissimilar domains with a single iron-binding site. Inspecting the structure of the Ansell’s mole-rat TF modeled on the rabbit protein (PDP ID: 1JNF; [82]) as template, we realized that Lys 383 is located at the interface between the two lobes (Fig 3A). In the rabbit TF two juxtapositioned Asn residues at position 383 and 312 might form an H-bond and this constellation could stabilize the inter-lobe interactions (Fig 3B). In contrast, the juxtaposition of the positively charged side chains of Lys383 and a conserved Arg312 in the Ansell’s mole-rat structural model would be expected to weaken the lobe-lobe interaction due to electrostatic repulsion. The functional consequences for TF implied by this modeling require experimental investigations.

thumbnail
Download:
Fig 3. Homology models of Ansell’s mole-rat transferrin (TF).

(A) Overview of the modeled Ansell’s mole-rat TF structure (green) superimposed onto the rabbit TF template structure (1JNF, grey). The position of the Asn383Lys site discussed in the text at the boxed center of the lobe interface numbered and indicated in cyan. Brown spheres: Fe3+ coordinated in the template structure (PDB ID: 1JNF; the radius of the metal ion is enlarged for better visibility). (B) Detail of the TF lobe interface. Shown is a magnification of the boxed region in (A). Coloring and numbering as in (A), side chain nitrogen atoms (blue), oxygen atoms (red). Potential hydrogen bond in 1JNF (light blue) as discussed in the text. Numbering (black) according to positions in the rabbit TF structure (1JNF).

https://doi.org/10.1371/journal.pgen.1007272.g003

Conclusions

We provided a systematic scan for PSGs on evolutionary branches of the African mole-rat family and other rodents leading to enhanced longevity that examine d considerably more extant and ancestral branches and provides a high resolution of positive selection than previous studies.

Analyzing the gene expression of PSGs, we found a highly significant pattern of down-regulation in the long-lived NMR and up-regulation in the short-lived rat, fitting the antagonistic pleiotropy theory of aging [83]. The PSGs and enriched functional terms cover many of the processes that are regulated by the mTOR pathway, e.g. translation, autophagy and cellular respiration. Furthermore, with RHEB and RHEB1L, direct regulators of mTOR [84], and with IGF1, a central component of insulin-signaling, were found to be under positive selection. In addition, we linked positive selection with inflammation and the antioxidant defense, processes known to be involved in regulation of lifespan.

By providing statistical evidences that the set of identified PSGs represents a meaningful resource for aging research, our results may encourage experimental follow-up studies–regardless whether the gene is currently linked to aging or determination of lifespan or not–since all sequences and alignments including the identified positively selected sites are accessible via supplement data.

Methods

Ethics statement

For tissue collection, rats were euthanized with CO2, whereas mole-rats were either anaesthetized by 3% isoflurane inhalation (naked mole-rats) or 6 mg/kg ketamine combined with 2.5 mg/kg xylazine (giant mole-rats; see Garcia et al. 2015, doi: 10.1111/vaa.12180) and then euthanized by surgical decapitation. Animal housing and tissue collection was compliant with national and state legislation (Berlin: breeding allowance #ZH 156; animal experimentation approval G 0221/12, Landesamt für Gesundheit und Soziales, Berlin; Essen: breeding allowances 32-2-1180-71/328 and 32-2-11-80-71/345, ethics/animal experimentation approval 84–02.04.2013/A164, Landesamt für Natur-, Umwelt- und Verbraucherschutz Nordrhein-Westfalen).

CDS data

We examined nine African mole-rat species covering all six genera. Additionally, our analysis comprises eight outgroup species, including the long-lived blind mole rat and the chinchilla. mRNA sequences of seven distantly related outgroup species were obtained from RefSeq along with their CDS annotation (S1 Table). For the naked mole-rat we used a recently published de novo transcriptome assembly [85]. RNA-seq data for six mole-rat species was obtained from GenBank Sequence Read Archive, study SRP061925 [17]. The reads were assembled and annotated using FRAMA as described in [85].

For Ansell’s and giant mole-rat, purification of RNA from 13 and 17 tissues, respectively, was done using Qiagen RNeasy Mini Kit following the manufacturer's description. Novel RNA-seq was performed for both species as described in S2 Table. De novo transcriptome assemblies of the generated data were performed using FRAMA with human as reference species [85]. In cases in which multiple isoforms per gene were annotated based on the reference, all of them were used in subsequent analyses. The completeness of the assemblies, estimated using BUSCO [86], was 54–100% (S1 Table).

For the silvery mole-rat and the greater cane rat genome sequencing was performed to complement the transcriptome data. DNA was isolated from liver tissue of two female silvery mole-rats and a male greater cane rat using DNeasy Blood & Tissue (Qiagen). DNA was then converted to Illumina libraries and sequencing was done as given in S2 Table. Sequence reads were cleaned by removal of adaptors. Then, the reads were trimmed using the CLC quality trim program (parameters -l 0.5, -f 33 and -b 0.1), i.e. the largest region of each read was identified that that has at max. 10% of its bases with quality scores ≤20 and this region was used for further analysis only if it contained at least 50% of the total bases of the respective read (otherwise the read was discarded). Additionally, duplicons were discarded. De novo genome sequence assembly was performed using CLC assembler (Qiagen) with default settings. The CDS annotation was done using AUGUSTUS [87] with Ansell’s mole-rat CDSs as hint. The estimated completeness for these assemblies using BUSCO was 39% and 13% for silvery mole-rat and greater cane rat, respectively (S1 Table). For the greater cane rat, an additional, reference based coding sequence assembly was created by mapping greater cane rat reads against giant mole-rat transcripts using CLC mapper (-a local -l 0.4 -s 0.9) and introducing changes to the giant mole-rat transcripts using CLC variation finder (-z unknown -c 2 –I). The estimated completeness of this assembly was 91% (S1 Table).

All animals were housed and euthanized compliant with national and state regulations. Read data was deposited as ENA (European Nucleotide Archive) study PRJEB20584.

Identification of positively selected genes

To scan on a genome-wide scale for genes under positive selection, we fed the CDSs of the described species set along with the branches we wanted to examine (Fig 1) into the PosiGene pipeline [88]. Giant mole-rat was used as PosiGene’s anchor species. Orthology was determined by PosiGene via best bidirectional BLAST searches [89, 90] against the orthologs groups defined in the HomoloGene [91] database (PosiGene parameter -hs for HomoloGene species and -nhs for species not included in the database). Regarding the silvery mole-rat, for which we had both a genome and a transcriptome assembly, we used generally the transcriptome assembly, except for those ortholog groups in which the silvery mole-rat ortholog could not be assigned via transcriptomic but via genomic data. This was accomplished by calling the three PosiGene modules separately, feeding both assemblies independently in the first module (ortholog assignment) and deleting all genome-based silvery mole-rat sequences in those ortholog groups that contained transcriptome-based silvery mole-rat CDSs before calling the second module. We considered all genes with Benjamini-Hochberg corrected p-values (FDR) < 0.1 as candidate genes. To exclude that these candidates were products of computational convergence problems that are known to potentially cause false positives in such analyses [92], we performed for each candidate two control runs with PosiGene. Only those candidates that were approved in both control runs were considered as PSGs and used in subsequent analyses (across all branches 8 candidates were removed, S5–S15 Tables).

The above described procedure was performed separately for the branches 2–11 and for branch 1 with different greater cane rat assemblies. While for the branches 2–11 the de novo greater cane rat assembly was used, for branch 1 the reference based greater cane rat assembly was used. The reason for this separation is the lack of completeness of the initial de novo greater cane rat assembly (S1 Table). The impact of this for branches 2–11 can assumed to be minimal as the greater cane rat is only one of nine potential outgroups. For branch 1 (LCA of African mole-rats), however, the greater cane rat is the sister taxon whose presence is required in an alignment to ensure that detected signs of positive selection do not predate branch 1 [18]. Therefore, to avoid a considerably lower sensitivity with respect to branch 1 than for other branches, the second, more complete, reference based greater cane rat assembly was used (S1 Table).

An overview about the number of genes and sequences tested for positive selection in the different branches is shown in S4 Table.

Differentially expressed genes during naked mole-rat and rat aging

The young and old rats (strain Wistar) had an age of 6 (n = 4) and 24 (n = 5) months, respectively. The young naked mole-rats had an age of 3.42±0.58 years (average±sd, n = 6). The old naked mole-rats were at least 21 years old (recorded lifetime in captivity, n = 3). All examined animals were males. All animals were housed and euthanized compliant with national and state regulations. For both species, purification of RNA from liver samples was done using Qiagen RNeasy Mini Kit following the manufacturer's description. In short, we performed RNA-seq using Illumina HiSeq 2500 with 50 nt single read technology and a sequencing depth of at least 20 mio reads/sample (S17 Table). For naked mole-rat, the read mapping was performed with STAR [93] (—outFilterMismatchNoverLmax 0.06—outFilterMatchNminOverLread 0.9—outFilterMultimapNmax 1) against the public genome (Bioproject: PRJNA72441) that we had annotated before by aligning the above-mentioned naked mole-rat transcriptome reference using BLAT [94] and SPLIGN [95]. Rat reads were aligned against the mentioned RefSeq reference using bwa aln [96] (-n 2 -o 0 -e 0 -O 1000 -E 1000). Read data and counts were deposited as GEO (Gene Expression Omnibus) series GSE98746. Differentially expressed genes (FDR≤0.1, S18 and S19 Tables) and fold-changes were determined with DESeq2 [97]. GAGE [98] was used to determine enriched gene ontologies based on fold-changes (S20 Table). Gene ontologies with FDR≤0.1 were summarized using REVIGO (allowed similarity = 0.5) [99]. Four of the six largest summarized categories of the resulting treemap (S21/S22 Tables) were further analyzed due their aging relevance (representative terms given): “translation” (GO:0006412), “cellular respiration” (GO:0045333), “response to oxidative stress” (GO:0006979) and “iron ion homeostasis” (GO:0055072). For each of these categories the union of genes across gene ontology terms was built. These unions were tested for significant overlaps with (i) the union of PSGs across branches and (ii) the union of PSGs across branches that were down-regulated during aging in naked mole-rat and up-regulated in rat (Fisher’s exact test). Functional annotation of the PSGs in respect to the four categories is given in S23 Table).

Gene ontologies

We determined enrichments for GO categories with Fisher’s exact test based on the R package GOstats (S24 Table). The resulting p-values were corrected using the Benjamini-Hochberg method [100]. We used throughout the manuscript 0.1 as significance threshold.

Homology modeling of protein structure

Models were built in SWISS-MODEL (http://swissmodel.expasy.org;) [101, 102]. No further optimization was applied to the resulting models. Superimposition of the model and template structures and rendering was carried out using CHIMERA [103].

Supporting information

S1 Fig. Taxon coverage.

https://doi.org/10.1371/journal.pgen.1007272.s001

(PDF)

S2 Fig. Lengths of examined sequences.

https://doi.org/10.1371/journal.pgen.1007272.s002

(PDF)

S3 Fig. GC−content of examined sequence.

https://doi.org/10.1371/journal.pgen.1007272.s003

(PDF)

S4 Fig. Standard deviation of GC−content within alignments.

https://doi.org/10.1371/journal.pgen.1007272.s004

(PDF)

S1 Table. Data sources for assemblies and sequence statistics.

https://doi.org/10.1371/journal.pgen.1007272.s005

(XLSX)

S2 Table. Samples that were sequenced to create genome/transcriptome assemblies.

https://doi.org/10.1371/journal.pgen.1007272.s006

(XLSX)

S3 Table. PSGs on multiple branches.

https://doi.org/10.1371/journal.pgen.1007272.s007

(XLSX)

S4 Table. Overview of positively selected genes (FDR<0.1) on examined branches.

https://doi.org/10.1371/journal.pgen.1007272.s008

(XLSX)

S5 Table. Results on branch 1.

https://doi.org/10.1371/journal.pgen.1007272.s009

(XLSX)

S6 Table. Results on branch 2.

https://doi.org/10.1371/journal.pgen.1007272.s010

(XLSX)

S7 Table. Results on branch 3.

https://doi.org/10.1371/journal.pgen.1007272.s011

(XLSX)

S8 Table. Results on branch 4.

https://doi.org/10.1371/journal.pgen.1007272.s012

(XLSX)

S9 Table. Results on branch 5.

https://doi.org/10.1371/journal.pgen.1007272.s013

(XLSX)

S10 Table. Results on branch 6.

https://doi.org/10.1371/journal.pgen.1007272.s014

(XLSX)

S11 Table. Results on branch 7.

https://doi.org/10.1371/journal.pgen.1007272.s015

(XLSX)

S12 Table. Results on branch 8.

https://doi.org/10.1371/journal.pgen.1007272.s016

(XLSX)

S13 Table. Results on branch 9.

https://doi.org/10.1371/journal.pgen.1007272.s017

(XLSX)

S14 Table. Results on branch 10.

https://doi.org/10.1371/journal.pgen.1007272.s018

(XLSX)

S15 Table. Results on branch 11.

https://doi.org/10.1371/journal.pgen.1007272.s019

(XLSX)

S16 Table. Overlaps between this and previous studies.

https://doi.org/10.1371/journal.pgen.1007272.s020

(XLSX)

S17 Table. Samples that were RNA-sequenced to examine gene regulation during aging.

https://doi.org/10.1371/journal.pgen.1007272.s021

(XLSX)

S18 Table. DESeq2 result for gene expression comparison of young (Ø 3.42 years) vs. old (>21 years) naked mole-rats.

https://doi.org/10.1371/journal.pgen.1007272.s022

(XLSX)

S19 Table. DESeq2 result for gene expression comparison of young (6 months) vs. old (24 months) rats.

https://doi.org/10.1371/journal.pgen.1007272.s023

(XLSX)

S20 Table. GAGE gene ontology enrichment for expression changes during naked mole-rat aging (Ø 3.42 vs > 21 years, FDR<0.1).

https://doi.org/10.1371/journal.pgen.1007272.s024

(XLSX)

S21 Table. REVIGO treemap result of GAGE enrichment for differential expression during naked mole-rat aging.

https://doi.org/10.1371/journal.pgen.1007272.s025

(XLSX)

S22 Table. REVIGO representative categories (representative term given) of GAGE enrichment for differential expression during naked mole-rat aging.

https://doi.org/10.1371/journal.pgen.1007272.s026

(XLSX)

S23 Table. PSGs in aging relevant summarized REVIGO categories and quadrant 1 (up-regulated in rat and down-regulated in naked mole-rat).

https://doi.org/10.1371/journal.pgen.1007272.s027

(XLSX)

S24 Table. Gene ontologies enriched for PSGs on examined branches based on GOStats and Fisher's exact test (FDR<0.1).

https://doi.org/10.1371/journal.pgen.1007272.s028

(XLSX)

Acknowledgments

We thank Ivonne Görlich, Christiane Vole and Yoshiyuki Henning for excellent assistance, Debra Weih for proofreading the manuscript and Christoph Kaether for helpful discussions.

References

  1. 1. Austad SN. Comparative biology of aging. J Gerontol A Biol Sci Med Sci. 2009;64(2):199–201. pmid:19223603; PubMed Central PMCID: PMCPMC2655036.
  2. 2. Austad SN. Diverse aging rates in metazoans: targets for functional genomics. Mech Ageing Dev. 2005;126(1):43–9. pmid:15610761.
  3. 3. de Magalhaes JP, Costa J, Church GM. An analysis of the relationship between metabolism, developmental schedules, and longevity using phylogenetic independent contrasts. J Gerontol A Biol Sci Med Sci. 2007;62(2):149–60. pmid:17339640; PubMed Central PMCID: PMCPMC2288695.
  4. 4. Tacutu R, Craig T, Budovsky A, Wuttke D, Lehmann G, Taranukha D, et al. Human Ageing Genomic Resources: integrated databases and tools for the biology and genetics of ageing. Nucleic Acids Res. 2013;41(Database issue):D1027–33. pmid:23193293.
  5. 5. Fushan AA, Turanov AA, Lee SG, Kim EB, Lobanov AV, Yim SH, et al. Gene expression defines natural changes in mammalian lifespan. Aging Cell. 2015;14(3):352–65. pmid:25677554; PubMed Central PMCID: PMCPMC4406664.
  6. 6. Van Daele PA, Verheyen E, Brunain M, Adriaens D. Cytochrome b sequence analysis reveals differential molecular evolution in African mole-rats of the chromosomally hyperdiverse genus Fukomys (Bathyergidae, Rodentia) from the Zambezian region. Mol Phylogenet Evol. 2007;45(1):142–57. pmid:17576078.
  7. 7. Gorbunova V, Seluanov A, Zhang Z, Gladyshev VN, Vijg J. Comparative genetics of longevity and cancer: insights from long-lived rodents. Nat Rev Genet. 2014;15(8):531–40. pmid:24981598; PubMed Central PMCID: PMCPMC4353926.
  8. 8. Beltran-Sanchez H, Finch C. Age is just a number. Elife. 2018;7. pmid:29364114; PubMed Central PMCID: PMCPMC5783609.
  9. 9. Buffenstein R. Negligible senescence in the longest living rodent, the naked mole-rat: insights from a successfully aging species. J Comp Physiol B. 2008;178(4):439–45. pmid:18180931.
  10. 10. Delaney MA, Ward JM, Walsh TF, Chinnadurai SK, Kerns K, Kinsel MJ, et al. Initial Case Reports of Cancer in Naked Mole-rats (Heterocephalus glaber). Vet Pathol. 2016;53(3):691–6. pmid:26846576.
  11. 11. Taylor KR, Milone NA, Rodriguez CE. Four Cases of Spontaneous Neoplasia in the Naked Mole-Rat (Heterocephalus glaber), A Putative Cancer-Resistant Species. J Gerontol A Biol Sci Med Sci. 2017;72(1):38–43. pmid:27129918.
  12. 12. Kosiol C, Vinar T, da Fonseca RR, Hubisz MJ, Bustamante CD, Nielsen R, et al. Patterns of positive selection in six Mammalian genomes. PLoS Genet. 2008;4(8):e1000144. pmid:18670650.
  13. 13. Roux J, Privman E, Moretti S, Daub JT, Robinson-Rechavi M, Keller L. Patterns of positive selection in seven ant genomes. Mol Biol Evol. 2014;31(7):1661–85. pmid:24782441.
  14. 14. Sahm A, Bens M, Platzer M, Cellerino A. Parallel evolution of genes controlling mitonuclear balance in short-lived annual fishes. Aging Cell. 2017. pmid:28295945.
  15. 15. Kim EB, Fang X, Fushan AA, Huang Z, Lobanov AV, Han L, et al. Genome sequencing reveals insights into physiology and longevity of the naked mole rat. Nature. 2011;479(7372):223–7. pmid:21993625.
  16. 16. Fang X, Nevo E, Han L, Levanon EY, Zhao J, Avivi A, et al. Genome-wide adaptive complexes to underground stresses in blind mole rats Spalax. Nat Commun. 2014;5:3966. pmid:24892994.
  17. 17. Davies KT, Bennett NC, Tsagkogeorga G, Rossiter SJ, Faulkes CG. Family Wide Molecular Adaptations to Underground Life in African Mole-Rats Revealed by Phylogenomic Analysis. Mol Biol Evol. 2015;32(12):3089–107. pmid:26318402.
  18. 18. Sahm A, Platzer M, Cellerino A. Outgroups and Positive Selection: The Nothobranchius furzeri Case. Trends Genet. 2016;32(9):523–5. pmid:27423541.
  19. 19. Fang X, Seim I, Huang Z, Gerashchenko MV, Xiong Z, Turanov AA, et al. Adaptations to a subterranean environment and longevity revealed by the analysis of mole rat genomes. Cell Rep. 2014;8(5):1354–64. pmid:25176646.
  20. 20. Hughes KA, Reynolds RM. Evolutionary and mechanistic theories of aging. Annu Rev Entomol. 2005;50:421–45. pmid:15355246.
  21. 21. Bartke A. Healthy aging: is smaller better?—a mini-review. Gerontology. 2012;58(4):337–43. pmid:22261798; PubMed Central PMCID: PMCPMC3893695.
  22. 22. Gout JF, Kahn D, Duret L, Paramecium Post-Genomics C. The relationship among gene expression, the evolution of gene dosage, and the rate of protein evolution. PLoS Genet. 2010;6(5):e1000944. pmid:20485561; PubMed Central PMCID: PMCPMC2869310.
  23. 23. Zhang J, Nielsen R, Yang Z. Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level. Mol Biol Evol. 2005;22(12):2472–9. pmid:16107592.
  24. 24. Sahm A, Bens M, Platzer M, Cellerino A. Parallel evolution of genes controlling mitonuclear balance in short-lived annual fishes. Aging Cell. 2017;16(3):488–96. pmid:28295945; PubMed Central PMCID: PMCPMC5418189.
  25. 25. Valenzano DR, Benayoun BA, Singh PP, Zhang E, Etter PD, Hu CK, et al. The African Turquoise Killifish Genome Provides Insights into Evolution and Genetic Architecture of Lifespan. Cell. 2015;163(6):1539–54. pmid:26638078; PubMed Central PMCID: PMCPMC4684691.
  26. 26. Durlinger AL, Visser JA, Themmen AP. Regulation of ovarian function: the role of anti-Mullerian hormone. Reproduction. 2002;124(5):601–9. pmid:12416998.
  27. 27. Semeiks J, Grishin NV. A method to find longevity-selected positions in the mammalian proteome. PLoS One. 2012;7(6):e38595. pmid:22701678; PubMed Central PMCID: PMCPMC3372525.
  28. 28. Pitt JN, Kaeberlein M. Why is aging conserved and what can we do about it? PLoS Biol. 2015;13(4):e1002131. pmid:25923592; PubMed Central PMCID: PMCPMC4414409.
  29. 29. Mercken EM, Capri M, Carboneau BA, Conte M, Heidler J, Santoro A, et al. Conserved and species-specific molecular denominators in mammalian skeletal muscle aging. NPJ Aging Mech Dis. 2017;3:8. pmid:28649426; PubMed Central PMCID: PMCPMC5460213.
  30. 30. Kim SK. Common aging pathways in worms, flies, mice and humans. J Exp Biol. 2007;210(Pt 9):1607–12. pmid:17449826.
  31. 31. Dai H, Leeder JS, Cui Y. A modified generalized Fisher method for combining probabilities from dependent tests. Front Genet. 2014;5:32. pmid:24600471; PubMed Central PMCID: PMCPMC3929847.
  32. 32. Blagosklonny MV. Aging: ROS or TOR. Cell Cycle. 2008;7(21):3344–54. pmid:18971624.
  33. 33. Blagosklonny MV. Answering the ultimate question "what is the proximal cause of aging?". Aging (Albany NY). 2012;4(12):861–77. pmid:23425777; PubMed Central PMCID: PMCPMC3615154.
  34. 34. Johnson SC, Rabinovitch PS, Kaeberlein M. mTOR is a key modulator of ageing and age-related disease. Nature. 2013;493(7432):338–45. pmid:23325216; PubMed Central PMCID: PMCPMC3687363.
  35. 35. Kenyon CJ. The genetics of ageing. Nature. 2010;464(7288):504–12. pmid:20336132.
  36. 36. Zhang J, Kim J, Alexander A, Cai S, Tripathi DN, Dere R, et al. A tuberous sclerosis complex signalling node at the peroxisome regulates mTORC1 and autophagy in response to ROS. Nat Cell Biol. 2013;15(10):1186–96. pmid:23955302; PubMed Central PMCID: PMCPMC3789865.
  37. 37. Betz C, Hall MN. Where is mTOR and what is it doing there? J Cell Biol. 2013;203(4):563–74. pmid:24385483; PubMed Central PMCID: PMCPMC3840941.
  38. 38. Jung CH, Ro SH, Cao J, Otto NM, Kim DH. mTOR regulation of autophagy. FEBS Lett. 2010;584(7):1287–95. pmid:20083114; PubMed Central PMCID: PMCPMC2846630.
  39. 39. Rubinsztein DC, Marino G, Kroemer G. Autophagy and aging. Cell. 2011;146(5):682–95. pmid:21884931.
  40. 40. Bandyopadhyay U, Kaushik S, Varticovski L, Cuervo AM. The chaperone-mediated autophagy receptor organizes in dynamic protein complexes at the lysosomal membrane. Mol Cell Biol. 2008;28(18):5747–63. pmid:18644871; PubMed Central PMCID: PMCPMC2546938.
  41. 41. Cuervo AM, Dice JF. A receptor for the selective uptake and degradation of proteins by lysosomes. Science. 1996;273(5274):501–3. pmid:8662539.
  42. 42. Zhang C, Cuervo AM. Restoration of chaperone-mediated autophagy in aging liver improves cellular maintenance and hepatic function. Nat Med. 2008;14(9):959–65. pmid:18690243; PubMed Central PMCID: PMCPMC2722716.
  43. 43. Kevei E, Hoppe T. Ubiquitin sets the timer: impacts on aging and longevity. Nat Struct Mol Biol. 2014;21(4):290–2. pmid:24699075.
  44. 44. Saez I, Vilchez D. The Mechanistic Links Between Proteasome Activity, Aging and Age-related Diseases. Curr Genomics. 2014;15(1):38–51. pmid:24653662; PubMed Central PMCID: PMCPMC3958958.
  45. 45. Chondrogianni N, Petropoulos I, Franceschi C, Friguet B, Gonos ES. Fibroblast cultures from healthy centenarians have an active proteasome. Exp Gerontol. 2000;35(6–7):721–8. pmid:11053662.
  46. 46. Lee GY, Haverty PM, Li L, Kljavin NM, Bourgon R, Lee J, et al. Comparative oncogenomics identifies PSMB4 and SHMT2 as potential cancer driver genes. Cancer Res. 2014;74(11):3114–26. pmid:24755469.
  47. 47. Beck BD, Park SJ, Lee YJ, Roman Y, Hromas RA, Lee SH. Human Pso4 is a metnase (SETMAR)-binding partner that regulates metnase function in DNA repair. J Biol Chem. 2008;283(14):9023–30. pmid:18263876; PubMed Central PMCID: PMCPMC2431028.
  48. 48. Hansen M, Taubert S, Crawford D, Libina N, Lee SJ, Kenyon C. Lifespan extension by conditions that inhibit translation in Caenorhabditis elegans. Aging Cell. 2007;6(1):95–110. pmid:17266679.
  49. 49. Houtkooper RH, Mouchiroud L, Ryu D, Moullan N, Katsyuba E, Knott G, et al. Mitonuclear protein imbalance as a conserved longevity mechanism. Nature. 2013;497(7450):451–7. pmid:23698443; PubMed Central PMCID: PMCPMC3663447.
  50. 50. Baumgart M, Priebe S, Groth M, Hartmann N, Menzel U, Pandolfini L, et al. Longitudinal RNA-Seq Analysis of Vertebrate Aging Identifies Mitochondrial Complex I as a Small-Molecule-Sensitive Modifier of Lifespan. Cell Syst. 2016;2(2):122–32. pmid:27135165.
  51. 51. Miwa S, Jow H, Baty K, Johnson A, Czapiewski R, Saretzki G, et al. Low abundance of the matrix arm of complex I in mitochondria predicts longevity in mice. Nat Commun. 2014;5:3837. pmid:24815183; PubMed Central PMCID: PMCPMC4024759.
  52. 52. Hofmann JW, Zhao X, De Cecco M, Peterson AL, Pagliaroli L, Manivannan J, et al. Reduced expression of MYC increases longevity and enhances healthspan. Cell. 2015;160(3):477–88. pmid:25619689; PubMed Central PMCID: PMCPMC4624921.
  53. 53. Powell JD, Pollizzi KN, Heikamp EB, Horton MR. Regulation of immune responses by mTOR. Annu Rev Immunol. 2012;30:39–68. pmid:22136167; PubMed Central PMCID: PMCPMC3616892.
  54. 54. Licastro F, Candore G, Lio D, Porcellini E, Colonna-Romano G, Franceschi C, et al. Innate immunity and inflammation in ageing: a key for understanding age-related diseases. Immun Ageing. 2005;2:8. pmid:15904534; PubMed Central PMCID: PMCPMC1166571.
  55. 55. Chung HY, Cesari M, Anton S, Marzetti E, Giovannini S, Seo AY, et al. Molecular inflammation: underpinnings of aging and age-related diseases. Ageing Res Rev. 2009;8(1):18–30. pmid:18692159; PubMed Central PMCID: PMCPMC3782993.
  56. 56. Lopez-Otin C, Blasco MA, Partridge L, Serrano M, Kroemer G. The hallmarks of aging. Cell. 2013;153(6):1194–217. pmid:23746838; PubMed Central PMCID: PMCPMC3836174.
  57. 57. Hekimi S, Wang Y, Noe A. Mitochondrial ROS and the Effectors of the Intrinsic Apoptotic Pathway in Aging Cells: The Discerning Killers! Front Genet. 2016;7:161. pmid:27683586; PubMed Central PMCID: PMCPMC5021979.
  58. 58. Moldovan L, Moldovan NI. Oxygen free radicals and redox biology of organelles. Histochem Cell Biol. 2004;122(4):395–412. pmid:15452718.
  59. 59. Balaban RS, Nemoto S, Finkel T. Mitochondria, oxidants, and aging. Cell. 2005;120(4):483–95. pmid:15734681.
  60. 60. Kim GH, Kim JE, Rhie SJ, Yoon S. The Role of Oxidative Stress in Neurodegenerative Diseases. Exp Neurobiol. 2015;24(4):325–40. pmid:26713080; PubMed Central PMCID: PMCPMC4688332.
  61. 61. Barja G. The mitochondrial free radical theory of aging. Prog Mol Biol Transl Sci. 2014;127:1–27. pmid:25149212.
  62. 62. Umeda-Kameyama Y, Tsuda M, Ohkura C, Matsuo T, Namba Y, Ohuchi Y, et al. Thioredoxin suppresses Parkin-associated endothelin receptor-like receptor-induced neurotoxicity and extends longevity in Drosophila. J Biol Chem. 2007;282(15):11180–7. pmid:17301052.
  63. 63. Mitsui A, Hamuro J, Nakamura H, Kondo N, Hirabayashi Y, Ishizaki-Koizumi S, et al. Overexpression of human thioredoxin in transgenic mice controls oxidative stress and life span. Antioxid Redox Signal. 2002;4(4):693–6. pmid:12230882.
  64. 64. Perez VI, Cortez LA, Lew CM, Rodriguez M, Webb CR, Van Remmen H, et al. Thioredoxin 1 overexpression extends mainly the earlier part of life span in mice. J Gerontol A Biol Sci Med Sci. 2011;66(12):1286–99. pmid:21873593; PubMed Central PMCID: PMCPMC3210956.
  65. 65. Flynn JM, Melov S. SOD2 in mitochondrial dysfunction and neurodegeneration. Free Radic Biol Med. 2013;62:4–12. pmid:23727323; PubMed Central PMCID: PMCPMC3811078.
  66. 66. Son M, Fu Q, Puttaparthi K, Matthews CM, Elliott JL. Redox susceptibility of SOD1 mutants is associated with the differential response to CCS over-expression in vivo. Neurobiol Dis. 2009;34(1):155–62. pmid:19320055; PubMed Central PMCID: PMCPMC2835407.
  67. 67. Andziak B, Buffenstein R. Disparate patterns of age-related changes in lipid peroxidation in long-lived naked mole-rats and shorter-lived mice. Aging Cell. 2006;5(6):525–32. pmid:17129214.
  68. 68. Andziak B, O'Connor TP, Buffenstein R. Antioxidants do not explain the disparate longevity between mice and the longest-living rodent, the naked mole-rat. Mech Ageing Dev. 2005;126(11):1206–12. pmid:16087218.
  69. 69. Andziak B, O'Connor TP, Qi W, DeWaal EM, Pierce A, Chaudhuri AR, et al. High oxidative damage levels in the longest-living rodent, the naked mole-rat. Aging Cell. 2006;5(6):463–71. pmid:17054663.
  70. 70. Schmidt CM, Blount JD, Bennett NC. Reproduction is associated with a tissue-dependent reduction of oxidative stress in eusocial female Damaraland mole-rats (Fukomys damarensis). PLoS One. 2014;9(7):e103286. pmid:25068591; PubMed Central PMCID: PMCPMC4113376.
  71. 71. Dammann P, Burda H. Sexual activity and reproduction delay ageing in a mammal. Curr Biol. 2006;16(4):R117–8. pmid:16488857.
  72. 72. Dammann P, Sumbera R, Massmann C, Scherag A, Burda H. Extended longevity of reproductives appears to be common in Fukomys mole-rats (Rodentia, Bathyergidae). PLoS One. 2011;6(4):e18757. pmid:21533255; PubMed Central PMCID: PMCPMC3076438.
  73. 73. Schmidt CM, Bennett NC, Jarvis JU. The long-lived queen: reproduction and longevity in female eusocial Damaraland mole-rats (Fukomys damarensis). African Zoology 2013;48:193–6.
  74. 74. Vistoli G, De Maddis D, Cipak A, Zarkovic N, Carini M, Aldini G. Advanced glycoxidation and lipoxidation end products (AGEs and ALEs): an overview of their mechanisms of formation. Free Radic Res. 2013;47 Suppl 1:3–27. pmid:23767955.
  75. 75. Simm A, Muller B, Nass N, Hofmann B, Bushnaq H, Silber RE, et al. Protein glycation—Between tissue aging and protection. Exp Gerontol. 2015;68:71–5. pmid:25536383.
  76. 76. Dammann P, Sell DR, Begall S, Strauch C, Monnier VM. Advanced glycation end-products as markers of aging and longevity in the long-lived Ansell's mole-rat (Fukomys anselli). J Gerontol A Biol Sci Med Sci. 2012;67(6):573–83. pmid:22156473; PubMed Central PMCID: PMCPMC3348492.
  77. 77. Chen X, Guo C, Kong J. Oxidative stress in neurodegenerative diseases. Neural Regen Res. 2012;7(5):376–85. pmid:25774178; PubMed Central PMCID: PMCPMC4350122.
  78. 78. Finefrock AE, Bush AI, Doraiswamy PM. Current status of metals as therapeutic targets in Alzheimer's disease. J Am Geriatr Soc. 2003;51(8):1143–8. pmid:12890080.
  79. 79. Bush AI. The metallobiology of Alzheimer's disease. Trends Neurosci. 2003;26(4):207–14. pmid:12689772.
  80. 80. Macedo MF, de Sousa M. Transferrin and the transferrin receptor: of magic bullets and other concerns. Inflamm Allergy Drug Targets. 2008;7(1):41–52. pmid:18473900.
  81. 81. Hare D, Ayton S, Bush A, Lei P. A delicate balance: Iron metabolism and diseases of the brain. Front Aging Neurosci. 2013;5:34. pmid:23874300; PubMed Central PMCID: PMCPMC3715022.
  82. 82. Hall DR, Hadden JM, Leonard GA, Bailey S, Neu M, Winn M, et al. The crystal and molecular structures of diferric porcine and rabbit serum transferrins at resolutions of 2.15 and 2.60 A, respectively. Acta Crystallogr D Biol Crystallogr. 2002;58(Pt 1):70–80. pmid:11752780.
  83. 83. Medawar PB. An unsolved problem of biology. (Printed lecture: University College London). 1952.
    • 84. Groenewoud MJ, Zwartkruis FJ. Rheb and Rags come together at the lysosome to activate mTORC1. Biochem Soc Trans. 2013;41(4):951–5. pmid:23863162.
    • 85. Bens M, Sahm A, Groth M, Jahn N, Morhart M, Holtze S, et al. FRAMA: from RNA-seq data to annotated mRNA assemblies. BMC Genomics. 2016;17:54. pmid:26763976; PubMed Central PMCID: PMCPMC4712544.
    • 86. Simao FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics. 2015;31(19):3210–2. pmid:26059717.
    • 87. Stanke M, Schoffmann O, Morgenstern B, Waack S. Gene prediction in eukaryotes with a generalized hidden Markov model that uses hints from external sources. BMC Bioinformatics. 2006;7:62. pmid:16469098; PubMed Central PMCID: PMCPMC1409804.
    • 88. Sahm A, Bens M, Platzer M, Szafranski K. PosiGene: automated and easy-to-use pipeline for genome-wide detection of positively selected genes. Nucleic Acids Res. 2017. pmid:28334822.
    • 89. Overbeek R, Fonstein M, D'Souza M, Pusch GD, Maltsev N. The use of gene clusters to infer functional coupling. Proc Natl Acad Sci U S A. 1999;96(6):2896–901. pmid:10077608.
    • 90. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, et al. BLAST+: architecture and applications. BMC Bioinformatics. 2009;10:421. PubMed PMID: 20003500. pmid:20003500
    • 91. Geer LY, Marchler-Bauer A, Geer RC, Han L, He J, He S, et al. The NCBI BioSystems database. Nucleic Acids Res. 2010;38(Database issue):D492–6. pmid:19854944.
    • 92. Yang Z, dos Reis M. Statistical properties of the branch-site test of positive selection. Mol Biol Evol. 2011;28(3):1217–28. pmid:21087944.
    • 93. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. pmid:23104886; PubMed Central PMCID: PMCPMC3530905.
    • 94. Kent WJ. BLAT—the BLAST-like alignment tool. Genome Res. 2002;12(4):656–64. Article published online before March 2002. pmid:11932250; PubMed Central PMCID: PMCPMC187518.
    • 95. Kapustin Y, Souvorov A, Tatusova T, Lipman D. Splign: algorithms for computing spliced alignments with identification of paralogs. Biol Direct. 2008;3:20. pmid:18495041; PubMed Central PMCID: PMCPMC2440734.
    • 96. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60. pmid:19451168; PubMed Central PMCID: PMCPMC2705234.
    • 97. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. pmid:25516281; PubMed Central PMCID: PMCPMC4302049.
    • 98. Luo W, Friedman MS, Shedden K, Hankenson KD, Woolf PJ. GAGE: generally applicable gene set enrichment for pathway analysis. BMC Bioinformatics. 2009;10:161. pmid:19473525; PubMed Central PMCID: PMCPMC2696452.
    • 99. Supek F, Bosnjak M, Skunca N, Smuc T. REVIGO summarizes and visualizes long lists of gene ontology terms. PLoS One. 2011;6(7):e21800. pmid:21789182; PubMed Central PMCID: PMCPMC3138752.
    • 100. Benjamini Y, Y. H. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289–300.
    • 101. Arnold K, Bordoli L, Kopp J, Schwede T. The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics. 2006;22(2):195–201. pmid:16301204.
    • 102. Biasini M, Bienert S, Waterhouse A, Arnold K, Studer G, Schmidt T, et al. SWISS-MODEL: modelling protein tertiary and quaternary structure using evolutionary information. Nucleic Acids Res. 2014;42(Web Server issue):W252–8. pmid:24782522; PubMed Central PMCID: PMCPMC4086089.
    • 103. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, et al. UCSF Chimera—a visualization system for exploratory research and analysis. J Comput Chem. 2004;25(13):1605–12. pmid:15264254.