Mutations That Increase the Life Span of C. elegans Inhibit Tumor Growth

See allHide authors and affiliations

Science  18 Aug 2006:
Vol. 313, Issue 5789, pp. 971-975
DOI: 10.1126/science.1121908


Mutations in gld-1 cause lethal germline tumors in the nematode Caenorhabditis elegans. We find that a wide variety of mutations that extend C. elegans' life span confer resistance to these tumors. The long life spans of daf-2/insulin-receptor mutants were not shortened at all by gld-1 mutations; we attribute this finding to decreased cell division and increased DAF-16/p53–dependent apoptosis within the tumors. Mutations that increase life span by restricting food intake or inhibiting respiration did not affect apoptosis but reduced tumor cell division. Unexpectedly, none of these longevity mutations affected mitosis in normal germlines; this finding suggests that cellular changes that lead to longevity preferentially antagonize tumor cell growth.

In nature, there is a strong correlation between physiological aging and tumor susceptibility. Mice, which have short (2-year) mean life spans, frequently acquire tumors after ∼1 year, whereas dogs do so after ∼10 years and humans only after many decades. Understanding how youthful animals resist tumors may provide new insights into tumor biology, particularly if genes that regulate aging influence tumor susceptibility. Many apoptotic, signaling, and other genes that affect tumors in mammals have orthologs in C. elegans. Moreover, mutations affecting insulin/IGF-1 (insulin-like growth factor–1) signaling, mitochondrial activity, and food intake each extend life span in both worms and mammals (14). Thus, C. elegans may be a valuable organism in which to investigate links between aging and tumors.

In C. elegans gld-1 tumor-suppressor mutants, germ cells in the early stages of oogenesis reenter the mitotic cell cycle and overproliferate (5). The cells eventually break out of the gonad and fill the body, killing the animal early in life. Like transformed vertebrate cells, these cells proliferate in a growth factor–independent manner (5), and they do not undergo programmed cell death (6). As they do not exhibit an angiogenesis-inducing or clear metastatic phenotype, these cells most likely approximate vertebrate cells in the early stages of tumorigenesis. In this study, we asked whether the lethal effects of this overproliferation could be delayed by longevity mutations that inhibit insulin/IGF-1 signaling [the receptor mutations daf-2(e1370) and daf-2(mu150)] (7, 8), a mutation that extends life span via caloric restriction [the feeding mutant eat-2(ad1116)] (9), and mitochondrial mutations that disrupt respiratory chain components [isp-1(qm150)] (10) or ubiquinone production [clk-1(qm30)] (11). These mutations affect distinct pathways (3, 4, 9), although they may converge on common downstream processes. In principle, longevity mutations might not affect the tumor, in which case they would not extend the short gld-1 mutant life span. Alternatively, as certain p53 mutations that inhibit tumors also accelerate aging and shorten life span (12), mutations that lengthen wild-type life span might promote tumor growth and accelerate death. In contrast, all of the longevity mutations tested extended the life spans of gld-1 mutants (Fig. 1). In fact, the long life spans of the two daf-2 mutants were not shortened at all by gld-1 mutations. Thus, we observed a strong correlation between life-span extension and tumor resistance.

Fig. 1.

Mutations that increase longevity delay the death caused by germline-tumor mutations. For each chart, experimental and control animals were grown in parallel. In this and other figures, gld-1(-) refers to the null allele gld-1(q485), and all RNAi-treated animals are labeled accordingly. (A and B) daf-2(e1370) (A) and daf-2(mu150) mutations (B) suppressed the lethality caused by gld-1(-). Similar results were obtained with gld-1 RNAi (table S5). In control experiments, a nontumorous gld-1 mutant (op236) (18) behaved like the wild type (table S5). (C to E) isp-1(qm150) (C), clk-1(qm30) (D), and eat-2(ad1116) mutations (E) delayed the death caused by gld-1 RNAi, although the animals did not live as long as isp-1, clk-1, or eat-2 mutants exposed to control RNAi. The magnitude of the life-span extension caused by these mutations was roughly the same in gld-1(-) and gld-1(+) backgrounds (table S5). (F) daf-2(RNAi) delayed the death caused by another germline tumor mutant, glp-1(ar202). All experiments were repeated at least twice with similar results (table S5). We considered the possibility that germline tumors shorten life span by down-regulating the life span–extending protein DAF-16, because removing the germ cells in the wild type extends life span in a daf-16–dependent manner. However, changes in DAF-16 activity cannot be responsible, because gld-1 RNAi further shortens the life spans of daf-16(null) mutants (table S5). As they near death, the animals become rigid and so packed with germ cells that none of their organs are visible by Nomarski microscopy, a finding consistent with tumors`killing the animal.

How do these longevity mutations affect the tumor? We found that none of the mutations restored oocyte production, but they all reduced germ cell number (Fig. 2). Thus, these mutations were likely to inhibit germ cell proliferation and/or increase germ cell death.

Fig. 2.

Mutations that increase longevity reduce germ cell number in gld-1(-) mutants. Adult animals were stained with the DNA-intercalating dye DAPI (4′,6′-diamidino-2-phenylindole). Left panels, midpoints of the gonad arms; right panels, heads. In all panels, anterior is to the left; d, day of adulthood. (A) Wild type. (B) gld-1(q485) mutants lack oocytes and have many undifferentiated germ cells in their gonads (left), which later break out of the gonad and fill the head and body (right). (C) gld-1(q485); daf-2(e1370) and (D) gld-1(q485); daf-2(mu150) double mutants lack oocytes; however, they have far fewer undifferentiated germ cells (left) and maintain the integrity of their gonads (right), even into old age [(C), lower panel]. The number of germ cells in gld-1(op236), a nontumorous mutant of gld-1, is unaffected by daf-2 RNAi (table S6). (E to G) isp-1(qm150) (E), clk-1(qm30) (F), and eat-2(ad1116) mutants (G) also have fewer undifferentiated germ cells in the presence of gld-1 RNAi (left), and the spread of germ cells into the body is delayed (right). Scale bars, 20 μm. For germ cell estimates and statistical analysis, see table S6. We considered the possibility that daf-2 mutations increased the (low) frequency of transdifferentiation of gld-1(-) germ cells into somatic cells. We found, using Nomarski optics, DAPI staining, and neuronal green fluorescent protein markers, that this was not the case (fig. S5).

In the wild type, many germ cells undergo apoptosis as they enter oogenesis, but in gld-1 mutants, this apoptosis is blocked (6). Using the dye SYTO12 to label apoptotic cells (6), we found that daf-2 mutations stimulated apoptosis in the gld-1 germlines (Fig. 3, A and B). daf-2 mutations also stimulated apoptosis in wild-type germ cells undergoing oogenesis (Fig. 3D). Therefore, daf-2 mutations appear to trigger a general increase in germline cell death.

Fig. 3.

p53- and DAF-16–dependent germline apoptosis is triggered by insulin/IGF-1 but not by mitochondrial or feeding mutations. Germ cell corpses were identified by SYTO12 labeling of day 2 (48-hour) adults. Each bar represents the mean (±SE) of at least three combined experiments (for details, see table S1); *P < 0.0001. (A) daf-2(e1370) mutations partially restored germ cell death in gld-1(q485) mutants. This was blocked by a ced-3(-) mutation (table S1). (B) Representative SYTO12-labeled gld-1(q485) mutant (left) and gld-1(q485); daf-2(e1370) double mutant (right). Arrows, stained germ cell corpses; scale bar, 20 μm. (C) Life spans of gld-1(RNAi); daf-2(e1370) animals were reduced ∼20% by either ced-3(n1286) (left), cep-1(gk138) (right), or ced-4(n1162) (table S2) mutations. The number of germ cells in a gld-1(-); daf-2(-) animal was increased by either a ced-3(-) or cep-1(-) mutation (table S6). Dashed line, mean life span of animals treated with gld-1 RNAi. (D to F) Left panels, gld-1(RNAi) animals; right panels, gld-1(+) animals; *P < 0.0001. (D) daf-2 mutations increased germ cell apoptosis in wild-type as well as gld-1(-) animals; this increase was daf-16 dependent. Residual germ cell apoptosis in gld-1(RNAi) animals was likely due to incomplete RNAi knockdown. The requirement of daf-16 for this apoptosis is consistent with life-span data (fig. S6). (E) The daf-2–dependent increase in germ cell apoptosis was reversed by cep-1(gk138). (F) isp-1(qm150), clk-1(qm30), and eat-2(ad1116) mutations did not affect germ cell apoptosis in gld-1(-) or wild-type animals. (G) daf-16 was necessary for DNA damage–induced apoptosis (left). Germ cell corpses were scored 18 hours after irradiation. The checkpoint genes hus-1, mrt-1, and clk-2 were required for daf-2 RNAi to increase apoptosis in the wild type (right). For details and statistical data, see table S1.

How might daf-2 mutations affect apoptosis? Timed RNA interference (RNAi) experiments showed that daf-2 acts during adulthood to influence cell death (fig. S1). We found that the transcription factor DAF-16/FOXO, which is required for the longevity of daf-2 mutants (8), was required for daf-2 mutations to increase germline apoptosis in both the wild type and gld-1 mutants (Fig. 3D). The C. elegans p53 gene cep-1 (13) was required as well (Fig. 3E). p53/cep-1 does not appear to be a DAF-16 transcriptional target (14, 15). Together, these findings suggest that daf-2 inhibition during adulthood increases the expression of DAF-16 target genes whose products function, directly or indirectly, along with p53/CEP-1 to stimulate germline apoptosis.

In C. elegans, p53 stimulates germline apoptosis in response to genotoxic stress but plays only a minor role in germline apoptosis under normal conditions (13). Thus, we hypothesized that daf-2 mutations might affect apoptosis by shifting cells into a physiological state resembling that induced by genotoxic stress. This idea is attractive because inhibiting daf-2 activity is known to activate multiple stress resistance processes (1, 3). Consistent with this, we found that DAF-16 was required for the increase in germline apoptosis caused by gamma irradiation in wild-type animals (Fig. 3G). In addition, the checkpoint genes hus-1, mrt-2, and clk-2, which act upstream of p53 in the DNA damage–induced apoptotic pathway (16), were required for daf-2 mutations to increase apoptosis in otherwise wild-type animals (Fig. 3G).

The cell death that occurs in wild-type and gld-1; daf-2(-) double-mutant germ cells is restricted to cells entering oogenesis. For this reason, we also examined a germline-tumor mutant in which cells do not progress to meiosis. The glp-1(ar202gf) mutation renders the GLP-1/Notch receptor, which maintains the germ cells in the stem-cell state, constitutively active (17). We found that daf-2 mutations were unable to stimulate apoptosis in this germline tumor (table S1). Thus, daf-2 mutations may trigger cell death specifically in cell populations that are already “primed” for apoptosis. Consistent with this interpretation, p53 levels are known to be elevated in gld-1 mutants (18).

The life spans of gld-1(RNAi); daf-2(-) animals were shortened when cell death was prevented with either the ced-3(n1286) caspase mutation (19) or the cep-1(gk138) p53 mutation (20). [Cell death mutations do not reduce wild-type or gld-1(+); daf-2(-) life spans (7) (table S2).] Nonetheless, daf-2 mutations still more than doubled the life spans of both of these apoptosis-defective tumor strains (Fig. 3C). This result suggested that daf-2 mutations might influence cell proliferation as well as apoptosis.

We assayed cell proliferation using an M phase–specific (phospho–histone H3) antibody (21) and found that daf-2 mutations decreased the number of actively dividing germ cells in gld-1 mutants (Fig. 4, A and B). We observed a similar decrease when apoptosis was blocked by a ced-3 mutation (table S3), indicating that the decrease in labeling was not due to increased apoptosis. daf-2 mutations also inhibited cell division in the glp1/Notch germline tumors; consistent with this finding, these animals had a smaller overall germ cell number and increased life spans (Fig. 1F and fig. S2).

Fig. 4.

Longevity mutations reduce the number of M-phase germ cells in gld-1(-) but not wild-type animals. Dividing germ cells were scored by antibody labeling to phospho–histone H3 at 36 hours of adulthood. Each bar represents the mean (±SE) of at least three combined experiments; *P < 0.0001. (A) daf-2(mu150) and daf-2(e1370) mutations reduced labeled cells in gld-1(-) mutants by about 50%. Similar results were obtained with gld-1(RNAi) (table S3). (B) Representative gonads of wild-type (top), gld-1(q485) (middle), and gld-1(q485); daf-2(e1370) (bottom) animals labeled with antibody to phospho–histone H3. Scale bar, 20 μm. (C) The reduction of mitotic germ cells by daf-2(e1370), but not daf-2(mu150), mutations was incompletely reversed by daf-16 RNAi. Similar results were obtained in gld-1(RNAi) daf-16(mu86); daf-2(e1370) animals (table S3). In contrast, a cep-1(-) mutation had no effect on the number of mitotic cells (table S3). In a wild-type background, daf-2 mutations had no effect on germ cell divisions (right). (D) clk-1(qm30), isp-1(qm150), and eat-2(ad1116) mutations reduced the number of M-phase cells in gld-1(RNAi) animals by about 35%, 50%, and 30%, respectively, but had no effect in a wild-type background. The scale in (D) differs from that in (A) and (C) because gld-1 RNAi has a less severe effect on proliferation than a gld-1(q485) mutation. For details and statistical data, see table S3.

daf-16 was required for daf-2(mu150) mutations to inhibit gld-1 germline mitosis. However, a small portion of the effect caused by the stronger allele, daf-2(e1370), was daf-16 independent (Fig. 4C). Likewise, daf-2(e1370) mutations slightly extended the life spans of gld-1(RNAi) daf-16(null) mutants (∼20%, P < 0.0001) (fig. S3).

Unexpectedly, daf-2(-) mutations did not affect the number of M-phase germline cells in a wild-type background (although we may not have detected a subtle change with this antibody) (Fig. 4C). Thus, daf-2(-)'s antiproliferative effect appeared to be specific to tumor cells. This was particularly intriguing for the glp-1/Notch mutant, because its tumors are thought to consist of an over-abundance of normal germline stem cells. Perhaps the insulin/IGF-1 pathway, which controls food usage and growth in many species, becomes limiting for cell proliferation in the face of such a large metabolic load.

In Drosophila, insulin/IGF-1–pathway mutations were recently reported to prevent normal germ cell growth and maturation (22). The discrepancy between this finding and ours may reflect the strength of the mutations examined, because stronger daf-2 alleles in C. elegans inhibit reproduction severely (23). In contrast, the daf-2(mu150) mutants reproduced normally, in terms of both reproductive timing and brood size, and daf-2(e1370) mutants grown at 20°C had only a slight reduction in fecundity (table S4), as reported previously (7, 8).

In humans, PTEN tumor-suppressor mutations stimulate oncogenesis by elevating insulin/IGF-1 signaling (24). This inhibits the DAF-16 ortholog FOXO3a, which in turn prevents apoptosis and promotes cell proliferation (25). Here, we decreased insulin/IGF-1 signaling (and elevated DAF-16 activity) throughout the animal and obtained a strong tumor-suppressive effect. This may have implications for mammalian tumors, including those that do not carry mutations in the insulin/IGF-1 signaling pathway. Mutations that inhibit growth hormone signaling extend life span and increase tumor resistance in rodents (26). Our findings suggest that reduced insulin/IGF-1 signaling, which results from growth hormone deficiency, may be responsible for their cancer resistance, and that this resistance is mediated by FOXO (and possibly p53) activity.

Are these tumor-suppressive mechanisms activated by other longevity pathways? To address this question, we analyzed mutations that cause caloric restriction [eat-2(ad1116)] (9) or impair mitochondrial activity [isp-1(qm150) and clk-1(qm30)] (10, 11). None of these mutations affected germline apoptosis in gld-1 tumor mutants (Fig. 3F and fig. S4). Thus, this effect appears to be specific to the insulin/IGF-1 pathway. However, all of the mutations reduced the number of M-phase germ cells (Fig. 4D). Remarkably, as with daf-2 mutations, none of these mutations reduced the number of M-phase germ cells in gld-1(+) animals (Fig. 4D). Thus, all of the longevity mutations we examined preferentially affected the proliferation of tumor cells. This implies a molecular link between processes that protect or repair cells and processes that disadvantage tumor cell growth.

Caloric restriction inhibits tumorigenesis in mammals, and our findings suggest that inhibiting the mammalian orthologs of isp-1 and clk-1 may do so as well. isp-1 mutations are thought to increase life span by inhibiting respiration (3, 4, 10). Perhaps tumor cells, which have low rates of respiration, are particularly sensitive to further reductions, or perhaps reducing respiration rates in surrounding cells inhibits tumor growth. The mechanism by which clk-1 mutations increase life span is not known. C. elegans clk-1 mutants, which receive essential quinones from the environment (27), do not have reduced metabolic output (28). The longevity of clk-1+/– mice has been associated with increased stress resistance (2). It will be interesting to investigate the effects of clk-1 mutations on mammalian tumors.

Our results show that many C. elegans longevity mutations are tumor protective. The generality of our findings, along with the cancer resistance of long-lived calorically restricted and endocrine-mutant rodents, argues that mutations such as the p53 alleles that inhibit tumors but accelerate aging (12) may be the exception rather than the rule. It may be that during evolution, longevity and delayed tumorigenesis arose together—for example, from endocrine pathway and mitochondrial mutations that influenced both processes.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Tables S1 to S6


References and Notes

View Abstract

Navigate This Article