The science of aging and life extension: studies in C. elegans and human cells

(Updated, August, 2012)

Part 1:  C. elegans: mechanisms of life extension

Our laboratory discovered that the C. elegans insulin/IGF-1 receptor DAF-2, and the FOXO-family transcription factor DAF-16, regulate the rate of aging of C. elegans.  These findings showed that the aging process is subject to endocrine and transcriptional regulation. Our work has now led to the discovery that mammalian aging is also regulated hormonally by insulin and IGF-1 endocrine systems, and has catalyzed a fundamental shift in the way scientists view the aging process, from one that is inevitable and intractable to one that is plastic and subject to regulation. These findings extend to humans, too: In the last few years, DNA variants in this pathway have been associated with exceptional longevity in human populations around the world. The long-lived animal mutants are resistant to many age-related diseases, raising the possibility that multiple diseases can be countered at once by targeting aging itself.

An evolutionarily conserved regulatory system for aging

When we started our studies in the early 1990s, aging was generally thought to be a random and haphazard process.  I wanted to investigate aging because I thought that aging, like so many other processes in biology (including pattern formation, which we were studying), would be subject to evolutionarily conserved mechanisms of regulation. A mutant living ~50% longer than normal, age-1(hx546), had been identified by Klass and Johnson, but at the time this mutant was suspected to live long because of a reproductive trade-off.  We looked for long-lived mutants, and discovered that mutations in daf-2 double the lifespan of the animal (independently of reproduction) (Kenyon, et al., 1993, A C. elegans mutant that lives twice as long as wild type.  Nature 366: 461). These animals are magical: they are like 90-year olds who look and act 45.  daf-2 was cloned in the Ruvkun lab, which had been studying a developmental process involving this gene, and shown to encode an insulin/IGF-1-like receptor. Together these findings indicated that the aging process is subject to active, endocrine regulation. We made another very important discovery in 1993: in order for daf-2 animals to live so long, they require the activity of a second gene, daf-16 (Kenyon et al., 1993, ibid.), which we subsequently showed encodes a FOXO-family transcription factor (Lin, et al., 1997, Science 278: 1319; this was also shown independently by the Ruvkun lab, 1977].  This finding really cemented the concept that aging is subject to regulation, since new patterns of gene expression are required to extend lifespan.  DAF-16 has proven to be a key node in a regulatory network that affects aging. Its function is required for a wide variety of conditions to extend lifespan, including overexpression of the heat-shock transcription factor (below), AMP kinase, and the timing microRNA lin-4. 

This longevity pathway is conserved.  Other showed that inhibiting insulin/IGF-1 signaling can increase lifespan in flies and mice. In fact, the lifespans of different strains of mice are inversely correlated with their IGF-1 levels (Yuan et al, PNAS, 2012).  Likewise, small dogs live longer than large dogs, and a single IGF-1 allele is a major determinant of body size in dogs (Ostrander lab, Science, 2007). FOXO(DAF-16) proteins even regulate lifespan in yeast.  The most exciting new findings, though, are from studies on humans.  For example, functionally significant IGF-1 (coding-sequence) mutations are overrepresented among centenarians in an Ashkenazi-Jewish population (Suh et al, PNAS, 2008).  Likewise, analyzing a European longevity GWAS study at the level of pathways highlighted insulin/IGF-1 signaling (Deelen et al., 2011; Age).  FOXO3A (DAF-16) DNA variants have been linked to exceptional longevity in at least eight human populations around the world (see references in Kenyon, 2010, Nature 464: 504), though how these DNA variants affect gene function is not yet known.

DAF-2 and DAF-16 function in C. elegans

We showed that C. elegans DAF-2 receptor activity regulates DAF-16/FOXO, at least in part, by controlling its nuclear localization via AKT-dependent phosphorylation (Lin et al., 2001, Nature Genetics 28: 139). We also found that a conserved Cullin/SCF E3-ligase complex also stimulates DAF-16 activity, presumably via regulated proteolysis, though the target of this complex has not been identified (Ghazi et al, 2007, PNAS 104: 5947).

daf-2 mutations influence development, reproduction and stress resistance as well as aging. We showed that both daf-2 and daf-16 act exclusively during adulthood to regulate aging and stress resistance, whereas they act earlier to influence development and reproduction (Dillin, et al., 2002, Science 298: 830). These findings have exciting therapeutic implications, and, they confirmed that the insulin/IGF-1 pathway controls reproduction and aging independently of one another.  More recently, DAF-16/FOXO was shown also to act in the adult to influence aging in flies, and the IGF-1 receptor was shown to act independently of reproduction to influence aging in mice.

How is insulin/IGF-1 signaling is distributed among the different tissues of the animal to affect lifespan? We showed that DAF-2/insulin/IGF-1-receptor.and DAF-16/FOXO act in a cell non-autonomous fashion to control multiple downstream signals that influence lifespan (Apfeld and Kenyon. 1998, Cell 95: 199; Libina et al., 2003, Cell 115: 489).  We found that the intestine, which also serves as the animal’s adipose tissue, plays a key role in lifespan regulation. This is the case for adipose tissue in flies and mice as well.  We find that DAF-16 activity in the intestine/adipose affects other tissues in two ways: first, it feedback-regulates expression of an insulin gene, ins-7, which causes DAF-16 activity to rise in other tissues (Murphy et al., 2007, PNAS, 104: 19046).  Second, it acts through a different, unknown, pathway to keep the other tissues alive even if they are daf-16(-) (Libina et al., ibid; Zhang et al., submitted).  We would like to learn the identity of this other pathway.

How does the DAF-16 transcription factor affect aging? Using microarrays followed by RNAi we showed that DAF-16 controls expression of a wide variety of functionally-significant downstream genes, each of which, on its own, has only a modest effect on lifespan.  These downstream genes encode cell-protective proteins, chaperones, antimicrobial proteins, metabolic proteins and novel proteins (Murphy, C. et al., 2003, Nature 424: 277). Thus the animal contains a regulatory module for aging, in which diverse gene activities can be coordinated by DAF-16 and other upstream regulators to produce large effects on lifespan. Other labs have identified additional genes that act downstream of DAF-16 to influence lifespan.

Heat-shock factor regulates the rate of aging and is part of the insulin/IGF-1 signaling system

We discovered that the heat-shock transcription factor regulates aging.  We wanted to identify mutants that aged too fast, so we taught ourselves what old worms look like when viewed with Nomarskin optis. With this knowledge, we discovered that RNAi of the C. elegans heat-shock transcription factor HSF-1, which regulates the heat-shock response, causes progeria (rapid aging) (Garigan et al., 2002, Genetics 161: 1101).  We then showed that overexpression of HSF-1 extends lifespan (Hsu et al., 2003, Science 300: 1145). HSF-1, like DAF-16, is absolutely required for the longevity of daf-2 mutants.  HSF-1 is only required for the expression of a subset of DAF-16 target genes. These genes may include essential effectors of longevity.  They include the small heat-shock proteins, whose up-regulation in daf-2 mutants requires HSF-1 as well as DAF-16.  In fact, we found that small heat-shock proteins do contribute to the longevity of daf-2 mutants (Murphy et al, ibid; Hsu et al, ibid).

DAF-2 mutants have a more potent ER stress response, which contributes to their longevity

We found that daf-2 mutants are resistant to ER stress (Sivan-Korenblit et al., 2010, PNAS 107: 9730).  This ER stress-resistance is completely dependent on the XBP-1 UPR transcription factor. Paradoxically, XBP-1 RNA levels, and XBP-1 target genes like hsp-4/BIP, were expressed at lower, not higher, levels in daf-2 mutants.  After much thought, we came up with a model; namely, that in daf-2 mutants, XBP-1 collaborates with activated transcription factors such as DAF-16 to express powerful new ER-homeostasis genes. Using microarrays followed by genetics, we identified one gene up-regulated by both XBP-1 and DAF-16 in daf-2 mutants that contributes to the longevity of daf-2 mutants.  We suspect there are more.

Regulation of lifespan by sensory perception

We discovered that sensory perception regulates lifespan in C. elegans (Apfeld and Kenyon, 1999, Nature 402: 804). The effect is largely DAF-16 dependent. Two labs have now showed that sensory perception also regulates lifespan in flies and, interestingly, insulin levels rise to a greater extent in people when they smell the food they are eating.  When we zeroed in on specific neurons using a laser (Alcedo and Kenyon, 2004, Neuron 41: 45), we found baroque complexity: for example, loss of a gustatory neuron called ASI extends lifespan, and this lifespan extension requires another neuron, ASJ.  Interestingly, ASJ is not required for loss of olfactory neurons to extend lifespan. How does this circuitry function at the molecular level? We are investigating this fascinating question now using the new techniques that have been developed for perturbing and visualizing neuronal activity.  

Regulation of lifespan by the reproductive system

We discovered that depleting the gonad of germ cells (specifically germline stem cells) extends C. elegans’ lifespan (Hsin and Kenyon, 1999, Nature 399: 362; Arantes-Oliveira et al., 2001, Science 295: 502). A similar situation was recently shown to exist in flies, which is interesting because flies and worms are pretty different anatomically.  In mammals, too, reproductive signaling can extend lifespan (Mason et al., J. Gerontol, 2009; Aging Cell, 2011).

The lifespan extension of germline-less C. elegans requires DAF-16/FOXO.  We found that removing the germ cells stimulates DAF-16 nuclear localization in the intestine/adipose tissue, specifically in the adult (Lin et al., 2001, ibid).  This localization is significant, because expressing DAF-16 only in the intestine is sufficient to account for the entire lifespan extension produced by germline removal (Libina et al, ibid).  DAF-16 nuclear localization completely requires KRI-1, a conserved ankyrin-containing protein (Berman and Kenyon, 2006, Cell, 124: 1055).  In addition, DAF-16’s function (but not its localization) requires a putative transcription-elongation factor called TCER-1 (Ghazi and Kenyon, 2009, PLoS Genetics, Sep;5(9):e1000639). There are many interesting differences between DAF-16’s role in the germline pathway and its role in the DAF-2/insulin/IGF-1 pathway. For example, the timing and tissue specificity of DAF-16 localization are different; and TCER-1 and KRI-1 are not required for daf-2 mutations to extend lifespan.

A particularly interesting feature of the germline longevity pathway is that (again, unlike the daf-2 pathway) it requires the nuclear hormone receptor DAF-12 (Hsin and Kenyon, ibid).  DAF-12 is the worm ortholog of the human vitamin D and LXR receptors.  The Antebi and Manglesdorf labs identified ligands for DAF-12, specific bile acids called dafachronic acid.  We were able to show that dafachronic acid and DAF-12 are the means by which the somatic gonad extends lifespan when the germ cells are gone (Yamawaki, et al., 2010, PLoS Biology, 10.1371/journal.pbio.1000468). DAF-16 can still enter the nucleus when the germline is removed in DAF-12 mutants (at least partially), so there are other signals from the germline to the intestine as well.  We are trying to identify these factors, and (along with other labs) dissect this new signaling pathway in more detail.

Amazing lifespan extensions

By perturbing both insulin/IGF-1 and reproductive signaling in the same animal, we were able to extend the mean lifespan of C. elegans six fold without apparently decreasing its health or activity until it is near death (Arantes-Oliveira et al., 2003, Science 302: 611).  More recently, the Shmookler-Reis lab showed that inactivating the age-1 PI3-kinase, which acts downstream of daf-2, can increase lifespan by 10-fold! Learning how these extremely long lifespans are achieved will be very valuable. Perhaps related mechanisms are used to achieve the remarkable longevity of naked mole rats and bats, which live so much longer than mice.

Regulation of lifespan by mitochondrial respiration

We discovered that mildly inhibiting mitochondrial respiration or ATP synthesis increases lifespan (Dillin et al., 2002, Science 298: 2398; similar findings were made independently by the Hekimi and Ruvkun labs).  Interestingly, we found that respiration must be inhibited during development for lifespan to be extended, suggesting that a switch is thrown during development that influences later life.  We found that the hypoxia-inducible transcription factor HIF-1 is required for this lifespan extension (Lee, S-J., et al., 2010, Current Biology 20: 2131). HIF-1, in turn, is activated by ROS, whose levels rise when respiration is inhibited. Other labs have implicated additional factors in this life extension pathway. It is a very interesting one, because inhibiting respiration can extend lifespan in yeast, worms, flies and mice.  Large mammals tend to have lower metabolic rates and they also tend to have longer lifespans (though exceptions exist).  Thus this longevity pathway could potentially have played a role in the evolution of longevity.

One interesting finding to come from this study is that low levels of paraquat extend lifespan, presumably by inducing a cell-protective response (Lee et al, ibid; the same finding was made independently by others as well).  Perhaps ROS is required for other conditions to extend life; we are investigating this idea now.

Lifespan extension by inhibition of translation

From an RNAi screen, we found that reducing the levels of ribosomal proteins extends lifespan.  So does inhibition of ribosomal-protein S6 kinase or translation-initiation factors (Hansen et al., 2007, Aging Cell, 6: 95).  These perturbations, as well as inhibition of the nutrient sensor TOR, which was known to increase lifespan, all increase thermal-stress resistance.  Thus inhibiting translation may extend lifespan by shifting to a physiological state that favors maintenance and repair. We went on to show that TOR inhibition, as well as caloric restriction (which inhibits TOR), triggers autophagy, and that autophagy is also required for life extension (Hansen et al., 2008, PLoS Genetics 4:e24) Others independently discovered that inhibiting ribosomal proteins in yeast, translation-initiation factors in worms, and S6K in flies extends lifespan.

 Mutations that increase lifespan slow tumor growth

Many people think that mutations that delay aging would accelerate cancer growth. However, we found that long-lived insulin/IGF-1, caloric restriction and mitochondrial mutations are all resistant to germline tumors in a C. elegans tumor model, due to increased DAF-16 and p53-dependent apoptosis, and/or decreased mitosis  (Pinkston et al., 2006, Science, 313: 971). Interestingly, none of these longevity mutations affected germline mitosis in normal animals, suggesting that the cellular changes that lead to longevity specifically antagonize excessive cell growth.  LIkewise, long-lived mouse insulin/IGF-1-pathway mutants are also cancer resistant.  Thus there need not be a tradeoff between longevity and cancer.  In fact, longer-lived species remain cancer-free longer than shorter-lived species. Our findings suggest a molecular mechanism for the co-evolution of longevity and tumor resistance; namely by changes in the activities of genes like the ones we analyzed. We went on to identify ~30 genes that act downstream of DAF-16 as tumor suppressors or tumor stimulators in our system.  25% of these are othologous to known cancer genes in human (Pinkston and Kenyon, 2007, Nature Genetics 39: 1403). As one expects only 2% by random chance, perhaps these studies in worms have identified some new human tumor genes. 

Regulation of lifespan by thermosensory neurons

Many ectotherms (“cold blooded animals”), including C. elegans, have shorter life spans at high temperature than at low temperature. High temperature is generally thought to increase the “rate of living” simply by increasing chemical reaction rates. However, many processes that seem to happen passively turn out to be regulated, so we took a closer look. We found that thermosensory neurons play a regulatory role. Inhibiting their function causes animals to have even shorter life spans at warm temperature. Thermosensory neurons affect lifespan via daf-12-dependent steroid-signaling. The thermosensory system may allow C. elegans to reduce the effect that warm temperature would otherwise have on aging, something that warm-blooded animals do by controlling temperature itself.

Sweet but deadly

Because glucose triggers insulin release in mammals, and just out of curiosity, we gave our worms low levels of glucose (2%) and measured their lifespans.  Glucose shortened the lifespan of wild-type animals, but not animals lacking daf-16/FOXO. This finding suggested that glucose shortens lifespan by up-regulating insulin/IGF-1 signaling.  (The day we learned this, in the spring of 2002, I went on a low-glycemic-index diet and have been on one ever since.)  Our findings suggest that glucose shortens lifespan via a pathway that involves not only DAF-16 but also insulins, HSF-1 and an aquaporin glycerol channel (Lee et al., 2009, Cell Metabolism 10: 379-391).

Studies related to neurodegenerative disease

Many neurodegenerative diseases are age-related, and we study these in worm models.  For example, we found that small heat-shock proteins delay polyQ (Huntingtin-like) protein aggregation in C. elegans (Hsu et al., 2003, ibid).  More recently, we found that loss of the progranulin gene, whose insufficiency potentiates human neurodegenerative disease, accelerates the rate at which apoptotic cells are engulfed in C. elegans.  Likewise, macrophages from progranulin-mutant mice engulf apoptotic cells faster than do wild-type macrophages (Kao et al., 2011, PNAS 108: 4441). Perhaps accelerated cell engulfment contributes to neurodegenerative disease in humans; for example, by clearing damaged cells that might otherwise recover. 

Endogenous protein aggregation

Human neurodegenerative disease proteins like Alzheimer’s Ab and Huntingtin become aggregation-prone with age.  Our finding that small heat-shock proteins counteract polyQ aggregation made us wonder whether normal worm proteins might also aggregate with age. Using mass spectrometry, we found that about 450 endogenous proteins reproducibly become insoluble with age (David, D. et al., 2010, PLoS Biology 8(8): e1000450; the Lithgow lab made a similar observation; 2011, Aging Cell). In the animal, the six proteins we examined all formed FRAP-insoluble aggregates.  In what may be a vicious cycle, proteins that maintain proteostasis themselves aggregate, including proteosomal and ribosomal proteins.  Different proteins aggregate in different parts of the cell, including the nucleolus, the centrosome-associated aggresome, and the general cytoplasm.  Amazingly about half of the minor components of human Alzheimer’s plaques are present in this set of C. elegans insoluble proteins.  So we may have discovered something of general significance. In daf-2 mutants, we see aggregates but they are FRAP-soluble! Why is this? Might the answer be part of the reason daf-2 mutants live so long?

Neurons and Aging

In 2002, the Driscoll lab reported that neurons in C. elegans do not age. We are fascinated by this finding, and would like to find out why this is. When we began to look more closely, however, we found that neurons do exhibit an age-dependent change: they begin to elaborate new branches! (Tank et al., J. Neuroscience. 31: 9279; others made the same discovery independently.)  This is a rare example of growth in old age; cancer is another. We have learned that Jun kinase signaling and also insulin/IGF-1 signaling both influence the timing of neuronal branching.  Both pathways act cell-autonomously in neurons. This means that it is possible for the overall animal to age more quickly while at the same time neuronal branching is slowed (or vice versa).  We are continuing to study the mechanism of this surprising phenomenon.

More worm projects

In addition to these ongoing studies, we (like others) are starting to learn more about the biology of aging itself, because this information could target specific biological processes in humans.  We are also asking whether aspects of aging can be reversed. In addition, we have developed some powerful new bioinformatic ways to analyze genomic data, and we are applying these methods to the biology of aging; for example, asking whether disparate aging pathways converge on common downstream processes.

Part II: Trying to extend human lifespan and healthspan