The HIF-1 Hypoxia-Inducible Factor Modulates Lifespan in C. elegans

1During development, homeostasis, and disease, cells integrate diverse environmental inputs and implement the appropriate changes in gene expression to survive stresses or execute developmental programs. Of the many environmental challenges that animals encounter, oxygen deprivation (hypoxia) is of particular concern. Oxygen is the final electron acceptor in cellular respiration, and is, therefore, necessary for most metazoan life. Successful adaptation to hypoxia involves changes in genetic programs that modulate metabolism, cell death, growth, and cellular differentiation. Many of these processes also have roles in cellular aging. This suggests that the networks that govern oxygen homeostasis and aging may be intertwined, but the molecules and mechanisms that underpin these networks are not fully understood.

2In animals as diverse as humans and C. elegans, the hypoxia-inducible factor (HIF) heterodimeric transcription factors are responsible for most of the hypoxia-induced changes in gene expression,. HIFs have been studied intensively by the biomedical research community, because HIFs have central roles in cancer and cardiovascular disease. Small molecules that inhibit HIF may be effective chemotherapeutic agents. Conversely, for patients in which injury, anemia, or ischemia has decreased oxygen supply to tissues, treatments that increase HIF levels may improve cell survival and stimulate angiogenesis.

3HIF regulatory networks are acutely sensitive to oxygen. The HIF transcription complex is heterodimeric, and both subunits belong to the family of DNA-binding transcription factors that contain basic-helix-loop-helix and PAS domains. HIFα stability is regulated by oxygen-dependent degradation. When oxygen levels are sufficiently high, HIFα is hydroxylated by members of the EGL-9/PHD family of Fe(II)- and 2-oxoglutarate-dependent dioxygenases. This modification allows VHL, a ubiquitin E3 ligase, to bind HIFα and target it for proteasomal degradation. In hypoxic conditions, HIFα is stable, and the HIF complex activates a battery of genes that enable adaptation to low oxygen conditions. HIF stability and activity are influenced by reactive oxygen species, citric acid cycle intermediates, Fe(II) availability, and intracellular signaling events,. HIFs are, therefore, well-positioned to integrate information about oxygen availability and cellular redox and to effect changes that influence stress resistance.

4Here, we test the hypothesis that HIF modulates post-mitotic aging. Previous studies in human cells had demonstrated that HIF promoted telomerase expression and that this contributed to increases in the replicative lifespan of primary human lung fibroblasts in hypoxic conditions, but the roles of HIF in post-mitotic aging were not understood. The somatic cells of adult C. elegans are post-mitotic, and the rate of aging can be modulated by genes that control insulin-like receptor signaling, mitochondrial function, protein folding, or responses to reactive oxygen species and other stresses,. The C. elegans genome encodes a single HIF alpha subunit, hif-1, and mutants lacking a functional hif-1 gene are viable in normal culture conditions,. In this study, we report that over-expression of HIF-1 protein extends the lifespan of C. elegans in a dose-dependent manner. Interestingly, we find that loss-of-function mutations in hif-1 gene can also increase C. elegans longevity and stress tolerance. The HIF-1 over-expression and hif-1 loss-of-function longevity phenotypes are not equivalent, as they differ in their dependence on daf-16/FOXO and skn-1/NRF2. These data provide insights to the intersection between oxygen homeostasis and aging, and they suggest that post-mitotic aging may be influenced by treatments or mutations that alter HIF-1 signaling.

5A small group of transcriptional regulators lying in the core of the stress response network, including DAF-16, SKN-1, SIR-2.1, PHA-4 and HSF-1, had been shown to extend lifespan when activated or over-expressed in transgenic C. elegans,,,,,,. To determine whether HIF-1 had similar effects, we generated a series of transgenic animals with extra copies of hif-1 integrated into the genome. The hif-1 minigene is a fusion of genomic and cDNA sequences, and it includes a c-myc tag (Fig. 1A). We confirmed that the epitope-tagged HIF-1 protein, like endogenous HIF-1, was over-expressed in animals that lacked vhl-1 or egl-9 (Fig. 1B). Further, the transgene was able to restore expression of a HIF-1-dependent reporter, Pnhr-57:GFP, to a hif-1 deletion mutant (Fig. 1C). In two independent lines (iaIs27 and iaIs28), the hif-1 integrated transgenes conferred moderate lifespan extension (22 or 23 days, p<0.0001, Tables 1 and S1).

6To further test the hypothesis that HIF-1 over-expression increased lifespan in C. elegans, we constructed a transgene in which HIF-1 proline 621 was converted to glycine. In wild-type animals, HIF-1 proline 621 is hydroxylated by EGL-9, and this covalent modification had been shown to mediate binding of HIF-1 to VHL-1. Mutations in vhl-1 or egl-9 can also be used to inhibit HIF-1 degradation, but these mutations have other defects that can complicate interpretation of the phenotypes. The vhl-1(ok161) deletion causes some changes in extracellular matrices that are not suppressed by a mutation in hif-1, and animals carrying strong loss-of-function mutations in egl-9 have an array of morphological and behavioral defects,,,. To understand the effects of HIF-1 over-expression, we generated transgenic animals carrying the hif-1 (P621G) expression construct and analyzed 3 independently isolated lines (iaIs32, iaIs33, and iaIs34). As shown in Fig. 1D, the HIF-1(P621G) gain-of-function mutation released the protein from degradation via the egl-9/vhl-1 pathway. In these strains, the HIF-1 (P621G) protein was expressed at levels 3–14 times greater than HIF-1 expressed from transgenes without the stabilizing mutation (Fig. 2A). The hif-1 (P621G) lines lived 20%–34% longer than wild-type N2 worms (Fig. 2B, Table 1). Moreover, the mean adult lifespans of the five hif-1 transgenic strains were proportionately correlated with HIF-1 expression levels (Fig. 2C). These data established that HIF-1 over-expression caused dose-dependant lifespan extension.

7Mutations in the daf-2 insulin-like receptor gene increase nuclear localization of the DAF-16 FOXO transcription factor and dramatically increase longevity and resistance to heat or oxidative stress. Conversely, daf-16 loss-of-function mutants age quickly,,,,,. We considered the possibility that loss-of-function mutations in daf-16 might be epistatic to HIF-1 over-expression in longevity assays. As shown in Fig. 3A and Table 2, hif-1(P621G) transgenes increased the mean lifespan of daf-16 (mu86) mutants by up to 33%. This was very similar to the effect that HIF-1(P621G) had on wild-type animals (Table 1). These data show that in animals over-expressing HIF-1, DAF-16 and HIF-1 act in parallel to influence post-mitotic aging.

8SKN-1 is a transcription factor that mediates the phase II detoxification response during oxidative stress in C. elegans, and expression of constitutively active SKN-1 can extend lifespan via genetic pathways that are independent of DAF-16. We tested the hypothesis that a loss-of-function mutation in skn-1 could abolish the prolongevity effects of HIF-1 over-expression. As shown in Fig. 3B and Table 2, over-expression of HIF-1 through hif-1(P621G) transgenes extended the lifespan of skn-1(zu67) mutants by up to 25%.

9We then asked whether the longevity effect of increased HIF-1 protein was independent of the DAF-2 insulin-like signaling pathway by crossing the hif-1(P621G) transgene to a long-lived daf-2 loss-of-function mutant. The hif-1 (P621G) transgene did not extend the lifespan of daf-2(e1370) worms (Fig. 3C, Table 2). Taken together, the genetic data indicate that increased dosage of HIF-1 promote longevity via a pathway that acts in parallel to DAF-16 and SKN-1, but may be downstream of the DAF-2 insulin-like receptor.

10Studies in mammalian cell lines have shown that the insulin receptor pathway can influence translation of HIF-1α, and recent studies in C. elegans have suggested that DAF-16 might regulate HIF-1 expression in some developmental or environmental contexts,. To directly test the hypothesis that DAF-2 regulated total HIF-1 protein levels, we compared the expression of epitope-tagged HIF-1 in wild-type animals and in daf-2 (e1370) mutants. We also assayed the expression of two HIF-1 targets: Pnhr-57:GFP and K10H10.2. Prior studies had shown that K10H10.2 mRNA levels were induced by hypoxia in wild-type animals in a hif-1-dependent manner and that mutations that caused over-expression of HIF-1 protein resulted in over-expression of K10H10.2 mRNA,. As shown in Fig. 4A, the daf-2 (e1370) mutation did not have a significant effect on the expression levels of HIF-1 or HIF-1(P621G) proteins (Fig. 4A) or on the expression of two HIF-1 targets that we assayed (Figs. 4B, 4C).

11Prior studies did not clearly predict the effects a hif-1 loss-of-function mutation might have on longevity. HIF-1 had been shown to promote the expression of genes that inhibit formation of stress resistant dauer larvae. In accordance with this, hif-1 (ia04) mutants arrest larval development as partial dauers when cultured at 27°C. Since mutations that confer stress resistance and dauer formation are often correlated with longer lifespan, it was reasonable to expect that hif-1 mutants might age slowly. However, loss-of-function mutations in the daf-16 or skn-1 prolongevity genes are associated with premature aging,,, and our data in Fig. 2 had established that over-expression of HIF-1 extended lifespan. To address this question directly, we assayed the lifespans of wild-type animals and hif-1 (ia04) mutants. Under normal culture conditions (20°C), hif-1 (ia04) mutants lived approximately 20% longer than wild-type animals (24 days, compared to 20 days for wild-type N2) (Table 1). To further validate this finding, we isolated an additional loss-of-function allele in hif-1 [hif-1 (ia07), described in Figs. 5A and 5B]. Sequence analysis predicts that hif-1 (ia07) should encode a truncated protein lacking a transcriptional activation domain. We also characterized hif-1 (ok2564), a deletion allele isolated by the C. elegans Genome Consortium (Figs. 5A). Phenotypic analyses confirmed that both hif-1 (ia07) and hif-1 (ok2564) decreased expression of HIF-1 targets (Fig. 5C). Each of the hif-1 loss-of-function alleles increased the mean adult lifespan by at least 10% (Fig. 5D, Table 1). hif-1 RNAi also extended the lifespan of wild-type animals (Table S2).

12To understand whether the longevity phenotypes conferred by HIF-1 over-expression and hif-1 loss-of-function mutations had similar genetic underpinnings, we asked whether the hif-1(ia04) mutation could extend the lifespan of daf-16- or skn-1-deficient mutants. As illustrated in Fig. 5E and 5F, daf-16 (mu86) animals and daf-16 (mu86) hif-1 (ia04) double mutants had similarly short lifespans (∼15 days), and there was not a significant difference between daf-2(e1370) and daf-2(e1370) hif-1(ia04) animals. We also determined that the hif-1 (ia04) deletion mutation did not extend the lifespan of skn-1(zu67) animals (Fig. 5G, Table 2). These data suggest that hif-1 loss-of-function mutations extend life span via a mechanism that requires daf-16 and skn-1 function. Thus, the longevity phenotypes of HIF-1 over-expressing animals and hif-1 (ia04) animals have differing requirements for daf-16 and skn-1 function.

13Thermotolerance and stress resistance have been shown to be correlated with C. elegans longevity,. To determine whether changes in hif-1 dosage that extend lifespan also conferred resistance to oxidative stress, we assessed relative resistance to t-butyl-peroxide. We assayed two loss-of-function alleles of hif-1 and two lines that expressed the stabilized HIF-1 (P621G). All of these strains exhibited significantly higher rates of survival in the oxidizing agent compared to wild-type (Fig. 6A). We also tested the ability of these strains to withstand heat stress. Prior studies had determined that both hif-1 (ia04) mutants and egl-9-deficient mutants were more resistant to heat stress than were wild-type animals. We extended these studies and determined that, similar to the findings in peroxide treatment, loss-of-function mutations in hif-1 or over-expression of HIF-1 increased the ability of the animals to survive at 35°C (Fig. 6B).

14This study advances our understanding of the regulatory networks that link oxygen homeostasis and aging. Cells respond to oxygen deprivation by changing their patterns of gene expression to optimize energy production and mitigate cellular damage. Oxygen deprivation can extend the replicative lifespan of cells in culture, and C. elegans cultured in hypoxic conditions live longer. The hypoxia-inducible factors have been termed master regulators of oxygen homeostasis, and in C. elegans, HIF-1 is required for most hypoxia-induced changes in gene expression. HIF-1 targets include genes involved in metabolism, cell signaling, transcription and translation. We show here that over-expression of HIF-1 extends C. elegans lifespan in a dose-dependent manner. We hypothesize that multiple HIF-1 targets act in concert to slow cellular deterioration over time.

15When over-expressed, HIF-1 acts in parallel to the stress responsive transcription factors DAF-16/FOXO and SKN-1/NRF to promote longevity. In wild-type animals, DAF-16 and SKN-1 are inhibited by the sole C. elegans insulin-like growth factor receptor, DAF-2. Analogous to mammalian IGF signaling pathways, DAF-2 regulates phosphoinositide 3-kinase (PI3K) signaling. The effectors of PI3K include kinases (AKT-1, AKT-2, and SGK-1) that phosphorylate DAF-16 and SKN-1 to repress transcription factor nuclear localization,,. As illustrated in Fig. 3, HIF-1 over-expression can extend the lifespan of daf-16 or skn-1 loss-of-function mutants. Microarray studies have suggested that HIF-1 and DAF-16 may co-regulate some genes, but HIF-1 and DAF-16 clearly have independent functions as well,. The relationship between HIF-1 and DAF-2 is less clear. The genetic data leave open the possibility that the prolongevity functions of HIF-1 are regulated by DAF-2. DAF-2 and the PI3K pathway have been shown to have other DAF-16-independent functions, including regulation of SKN-1 nuclear localization. Additionally, mutations in daf-2 were shown to protect C. elegans from the lethal effects of high-temperature anoxia, and to confer resistance to hypoxia-induced defects in neural development. Here, we have demonstrated that mutations in daf-2 did not cause significant changes in whole animal levels of HIF-1 protein or in the expression of two HIF-1 targets that we assayed (Fig. 4), but DAF-2 may have tissue-specific or stage-specific roles in HIF-1 regulation.

16A recent independent study from the Kaeberlein research group provides added support for the conclusion that HIF-1 over-expression extends C. elegans lifespan. Mehta et al reported that vhl-1 or egl-9 RNAi increased C. elegans longevity and decreased the toxic effects of transgenes that expressed polyglutamine repeats or beta amyloid. Further, they showed that daf-16 RNAi did not completely suppress the longevity conferred by a vhl-1 deletion mutation, and they found that vhl-1 RNAi had no effect on the nuclear localization of DAF-16:GFP. This further substantiates models in which HIF-1 over-expression acts in parallel to DAF-16 to modulate aging.

17While moderate increases in HIF-1 protein expression promote longevity, dramatic over-expression of HIF-1 target genes can be detrimental. C. elegans egl-9 mutants express HIF-1 targets at extremely high levels, and this causes a range of morphological and behavioral defects,,. We and others have shown that animals carrying a strong loss-of-function mutation in egl-9 do not live longer than wild-type animals (Table 1 and ). This suggests that animals treated with egl-9 RNAi or animals carrying transgenes expressing stabilized HIF-1 benefit from increased HIF-1 function, but only to a point. Forcing gene expression beyond normal parameters can be deleterious. A similar pattern has been reported for yeast LAG1 and for C. elegans SKN-1. Moderate over-expression of SKN-1 extends life span, but over-expression from high copy transgenes is toxic.

18HIF-1 over-expression and hif-1 loss-of-function mutations exert their prolongevity effects through different pathways. We demonstrated that three independently isolated hif-1 mutations or depletion of hif-1 by RNAi extended C. elegans lifespan. The hif-1 (ia07) allele is predicted to encode a truncated HIF-1 protein that lacks a transcriptional activation domain, and the hif-1(ia07) phenotype is slightly more severe than the phenotypes of the hif-1 (ia04) and hif-1 (ok2564) deletion alleles (Fig. 5A). In agreement with our findings, the Kapahi group determined that hif-1 RNAi increased C. elegans lifespan, and they demonstrated that this effect was abrogated by dietary restriction. These data show that the hif-1 loss-of-function phenotype can be strongly influenced by nutrients.

19There are also some intriguing differences between our findings and those from the other two groups that have recently examined the roles of HIF-1 in C. elegans aging. While we report and Chen et al have reported that hif-1-deficient animals live longer when food is abundant, Mehta et al did not detect differences between the lifespans of hif-1 (ia04) and wild-type animals in their experiments. This appears to reflect differences in assay conditions. Small differences in the bacterial food source, the temperature, or other conditions may alter the longevity phenotypes of hif-1-deficient animals. Chen et al conducted their experiments at 25°C, and they assayed the effects of varying amounts of live bacterial food on plates containing FUdR and antibiotics. Mehta et al performed their experiments at 20°C and fed the animals UV-treated bacteria on plates containing FUdR and antibiotics. All of the longevity assays that we report here were conducted in nutrient rich conditions in the absence of DNA synthesis inhibitors. Although our findings regarding the hif-1 loss-of-function phenotype largely agree with those of Chen et al, there are open questions about the role of daf-16 in hif-1-deficient animals. Chen et al found that hif-1 RNAi could extend the mean lifespan of daf-16 (mgDf47) animals by 18%. In our experiments, double mutants carrying loss-of-function mutations in both hif-1 and daf-16 had similar lifespans to daf-16 (mu86) single mutants (Fig. 3).

20Why might apparently small differences in assay conditions influence the phenotypes of hif-1-deficient animals? The answer to this question may require a fuller understanding of the roles of HIF-1 in heat acclimation and hormesis. Prior studies demonstrated that wild-type C. elegans that have been incubated at 25°C were more thermotolerant than animals that had been incubated at 20°C. This process of heat acclimation required hif-1 function. This suggests that while loss-of-function mutations in hif-1 or HIF-1 over-expression can increase thermal resistance (Fig. 6 and ), hif-1-deficient animals are less able to adapt to or benefit from changes in the environment. Future studies will investigate the ways in which culture conditions and environmental stresses can influence HIF-1 function and hif-1-deficient phenotypes and may discover additional roles for HIF-1 in aging, stress, and hormesis.

21To our knowledge, this is the first case in a multicellular organism in which wild-type animals (grown in nutrient-rich conditions) have been shown to live shorter than either loss-of-function mutants or animals over-expressing the protein. However, in yeast, both loss-of-function mutations in the LAG1 longevity assurance gene and moderate over-expression of LAG1 can extend replicative lifespan. Other studies in mammalian cells have shown that similar phenotypes can be caused by loss-of-function or gain-of-function mutations that force a signaling pathway outside of its normal range. For example, in models for TNF-induced apoptosis, kinase-dead and constitutively active MK2 mutants have been shown to have similar effects on apoptosis. When considering the complex roles of HIF-1 in aging, it is important to recognize that HIF-1 over-expression and hif-1 loss-of-function phenotypes are qualitatively different and have distinct genetic requirements.

22In the hundreds of millions of years since nematodes and mammals diverged from a common ancestor, some of the key molecules and feedback loops that modulated HIF signaling have been conserved,. The EGL-9/VHL-1 pathway enables rapid up-regulation of HIF in hypoxic conditions, but the system is tightly controlled. In humans and in C. elegans, activation of HIF results in increased expression of the HIF prolyl hydroxylases (PHD/EGL-9),. This establishes negative feedback loops to attenuate HIF expression. While either hif-1 depletion or HIF-1 over-expression increase longevity in optimal lab culture conditions, the conservation of HIF regulatory networks suggests that there is a strong selective pressure for keeping HIF signaling within a normal range, where HIF can effectively respond to subtle changes in the environment.

23Small molecule inhibitors of the HIF prolyl hydroxylases are being developed as treatments for anemia or other pathologies that may be ameliorated by increased HIF expression. There is also great interest in pharmaceuticals that can inhibit HIF and may be effective in combinatorial treatments for certain cancers. We anticipate that drugs that target HIF may have widespread, dose-dependent impacts on stress resistance and on somatic aging.

24The mutant alleles were as follows: LGI: daf-16(mu86)lf; LGII: daf-2(e1370) reduction-of-function; LGV: hif-1(ia04)lf, hif-1(ia07)lf, hif-1(ok2564)lf, egl-9(sa307)lf; LGX: vhl-1(ok161)lf.

25The pFOX1 and pFOX4 constructs encode wild-type HIF-1 and HIF-1(P621G), respectively. pFOX1 includes 5.2 kb genomic sequence 5′ to the hif-1 coding region. The first exon and first intron of hif-1 are fused to a cDNA fragment including exon 2 to exon 9 of hif-1 (for the predominant mRNA isoform: hif-1a). Five copies of c-myc epitope (5×myc) (from clone CD3-128, Arabidopsis Biology Resources Center) were inserted C-terminal to hif-1 coding sequences, and this was followed by a stop codon and 400 bp of genomic sequence just 3′ to the hif-1 coding region. To create pFOX4, the P621G point mutation was introduced to the pFOX1 sequence. Microparticle bombardment was used to create C. elegans carrying integrated copies of each construct. The hif-1 constructs were introduced to hif-1 (ia04), unc-119(ed3) animals, and the unc-119 rescuing plasmid pPD#MM016b was the co-transformation marker. Animals carrying integrated copies of the transgenes were isolated and backcrossed at least four times prior to further characterization.

26Longevity assays were conducted as described previously unless noted otherwise. Prior to life span assays, all strains were maintained at 20°C for at least two generations with abundant food. The OP50 bacterial food was grown in L-broth to a 600 nm optical density of approximately 0.4, and 250 microliters of live bacterial culture was spotted onto each 60 mm NGM plate. The plates did not contain FUdR or antibiotics. One to three days later, young adult hermaphrodites were allowed to lay eggs overnight on the bacterial food. The L4 larvae grown from these eggs were transferred to fresh plates. Each plate included 20–25 animals, and at least 50 animals were used in each independent trial. Longevity assays for daf-2-deficient strains were performed at 25°C. All other lifespan assays were conducted at 20°C. The worms were transferred to new plates every two days for the first two weeks and every six days thereafter. The data from individual experiments are listed in Table S1. Any experiments in which the assay conditions were modified are explicitly noted in the text and in the legends for Tables S2 and S3. Viability was scored every 2 days. Worms that crawled off the plates, burst at the vulva or died because progeny hatched in utero were excluded from final statistical analyses. Life span statistical analyses were carried out using JMP software (version 7.0) to determine the means and percentiles. p values were calculated using the log-rank (Mantel-Cox) method.

27In oxidative stress assays, young adult worms (one day after L4 molt) were transferred to NGM plates containing 7.2 mM t-butyl-peroxide (Sigma) and OP50 food. Animals were incubated at 20°C and were scored for survival at the time points shown. For thermal stress assays, young adult worms were incubated at 35°C on NGM plates seeded with OP50 and were scored for survival. At least 50 worms were tested in each trial and at least three independent trials were performed for each stress assay.

28For each lane of an immunoblot, ten (for Pnhr57:GFP) to 100 (for the epitope-tagged HIF-1) L4 or young adult worms were assayed. Mouse anti-AHA-1 antibody (cell culture supernatant) was produced in ISU Hybridoma Facility and used at 1∶100. Mouse anti-myc ascites (clone 9E10) from the Developmental Studies Hybridoma Bank was used at 1∶1000. Mouse anti-GFP antibody (from Roche) was used at 1∶1000.

29Quantitative RT-PCR was performed as described previously. inf-1 is not regulated by hypoxia and was used as an input control. The standard curve method was used to analyze the expression levels. Student t-tests were used to determine the p value for protein expression and quantitative RT-PCR data.