Anticonvulsant Medications Extend Worm Life-Span

See allHide authors and affiliations

Science  14 Jan 2005:
Vol. 307, Issue 5707, pp. 258-262
DOI: 10.1126/science.1105299


Genetic studies have elucidated mechanisms that regulate aging, but there has been little progress in identifying drugs that delay aging. Here, we report that ethosuximide, trimethadione, and 3,3-diethyl-2-pyrrolidinone increase mean and maximum life-span of Caenorhabditis elegans and delay age-related declines of physiological processes, indicating that these compounds retard the aging process. These compounds, two of which are approved for human use, are anticonvulsants that modulate neural activity. These compounds also regulated neuromuscular activity in nematodes. These findings suggest that the life-span–extending activity of these compounds is related to the anticonvulsant activity and implicate neural activity in the regulation of aging.

Aging is characterized by widespread degenerative changes. Although treatments for aging would be desirable, the development of such treatments is challenging. Approaches based on rational design require information about the aging process, but little information is currently available. Approaches based on random screens of potential treatments require relevant and feasible assays of aging, but the time and effort necessary to measure aging are substantial obstacles.

To address these challenges, we exploited the C. elegans model system. These animals age rapidly, and many processes are conserved between nematodes and vertebrates, including aspects of the aging process (1). To identify compounds that delay aging, we assayed 19 drugs from a variety of functional or structural classes that have known effects on human physiology (2). We reasoned that such compounds might have an undiscovered effect on aging. For each drug, hermaphrodites were cultured with three different concentrations from before fertilization until death, and the adult life-span [fourth larval (L4) stage to death] of about 50 animals was measured. To focus on aging, we excluded dead worms that displayed internally hatched progeny, an extruded gonad, or desiccation due to crawling off the agar.

Ethosuximide had the greatest effect on adult life-span, extending mean adult life-span from 16.7 to 19.6 days (17% increase) (Fig. 1B and Table 1). A dose-response analysis revealed that worms cultured with external concentrations of 2 and 4 mg/ml ethosuximide displayed the largest extensions of mean life-span; lower concentrations caused smaller extensions, whereas higher concentrations caused toxicity and reduced life-span (3). This effect was temperature sensitive; ethosuximide extended mean life-span by 35% at 15°C, 17% at 20°C, and insignificantly at 25°C (3).

Fig. 1.

Anticonvulsants extend adult worm life-span. (A) Compounds. (B to M) Hermaphrodite survival of [(B) to (D)] wild type (WT), (E) eat-2(ad465), (F) daf-16(m26) (G) daf-16(mu86), (H) osm-3(p802), (I) tax-4(p678), (J) unc-31(e928), (K) unc-64(e246), (L) aex-3(ad418), and (M) daf-2(e1370). Worms were exposed to ethosuximde (+ETH), DEABL (+DEABL), trimethadione from fertilization until L4 (+TRI 4/0), trimethadione from L4 until death (+TRI 0/4), or trimethadione from fertilization until death (+TRI 4/4 and +TRI). Dosages are shown in Table 1.

Table 1.

Mean and maximum life-spans. Most strains were fed live E. coli OP50 and cultured at 20°C. Exceptions were wild-type strain N2 cultured at 15°C (WT, 15°C), N2 fed live B. subtilis (WT, B. subtilis), and N2 fed UV-killed OP50 (WT, UV/E. coli). External drug concentrations are shown in milligrams per milliliter for ethosuximide (ETH), trimethadione (TRI), and succinimide (SUC). (4/0) and (0/4) indicate culture with drug from fertilization to L4 and L4 to death, respectively. Genotypes with no drug treatment are compared with line 1, and differences were not analyzed for statistical significance. Otherwise, comparisons are to the same genotype with no drug treatment. For these comparisons in the columns showing life-spans, numbers with no asterisks are not significant (P > 0.05); *, P < 0.05; **, P < 0.005; ***, P < 0.0001. Maximum adult life-span is the mean life-span of the 10% of the population that had the longest life-spans. N, number of hermaphrodites analyzed, with number of independent experiments in parentheses. N.D., not determined.

Genotype Drug Mean life-span ± SD (days) % Change in mean life-span Maximum life-span ± SD (days) % Change in maximum life-span N
WT None 16.7 ± 3.7 23.3 ± 1.7 976(19)
ETH(2) 18.9 ± 6.0*** +13 28.9 ± 1.7*** +24 479(10)
ETH(4) 19.6 ± 5.3*** +17 28.5 ± 1.8*** +22 458(10)
TRI(4) 24.6 ± 8.4*** +47 36.5 ± 2.2*** +57 482(9)
TRI (0/4) 20.7 ± 6.7*** +24 30.1 ± 2.3*** +29 124(2)
TRI (4/0) 15.5 ± 3.4 -7 21.0 ± 1.0 -10 110(2)
DEABL(2) 21.8 ± 7.6*** +31 34.7 ± 1.6*** +49 92(2)
SUC(2)View inline 16.0 ± 4.0 -4 23.5 ± 1.9 +1 94(2)
WT, 15°C None 23.6 ± 5.3 +41 32.9 ± 2.7 +41 116(2)
ETH(4) 31.9 ± 8.2*** +35 45.2 ± 2.4*** +37 93(2)
WT, B. subtilis None 19.5 ± 5.4 +17 28.8 ± 1.5 +24 108(2)
TRI(4) 27.1 ± 6.6*** +39 38.2 ± 1.8*** +33 115(2)
WT, UV/E. coli None 20.0 ± 6.3 +20 30.4 ± 2.3 +30 51(2)
ETH(4) 22.8 ± 4.8* +14 29.8 ± 1.4 -2 45(2)
TRI(4) 28.1 ± 6.0*** +41 37.3 ± 0.7** +23 49(2)
daf-16 (m26) None 14.4 ± 3.3 -14 20.0 ± 1.0 -14 123(3)
ETH(2) 16.7 ± 3.0*** +16 21.3 ± 0.8** +7 119(3)
TRI(2) 16.7 ± 3.2*** +16 22.0 ± 0.0 (N.D.) +10 58(1)
daf-16 (mu86) None 14.4 ± 3.2 -14 19.6 ± 0.9 -16 113(2)
ETH (0.5) 16.0 ± 3.5** +11 21.1 ± 0.5** +8 105(2)
TRI(4) 17.4 ± 4.5** +21 24.2 ± 1.2** +23 53(1)
daf-2 (e1370) None 34.6 ± 10.9 +107 52.6 ± 5.0 +126 142(3)
ETH(4) 39.3 ± 11.5** +14 56.8 ± 2.9* +8 118(3)
TRI(4) 37.6 ± 8.5* +9 50.0 ± 3.5 -5 50(1)
unc-31 (e928) None 22.8 ± 8.4 +37 36.2 ± 3.3 +55 56(3)
ETH(2) 27.4 ± 11.1** +20 44.4 ± 2.4*** +23 95(3)
TRI(4) 28.3 ± 5.4*** +24 35.8 ± 2.4 -1 55(1)
unc-64 (e246) None 22.6 ± 10.7 +35 42.8 ± 3.3 +84 227(4)
ETH(2) 24.4 ± 10.7* +8 43.7 ± 2.6* +2 160(3)
TRI(4) 28.1 ± 10.2*** +24 43.3 ± 2.9 +1 245(2)
aex-3 (ad418) None 19.3 ± 5.3 +16 29.1 ± 2.5 +25 209(4)
ETH(4) 21.8 ± 6.8*** +13 34.4 ± 2.1*** +18 236(4)
TRI(4) 26.0 ± 6.0*** +35 34.3 ± 1.0*** +18 113(2)
tax-4 (p678) None 20.1 ± 6.6 +20 31.4 ± 2.4 +35 144(3)
TRI(4) 27.4 ± 4.9*** +36 35.2 ± 1.2* +12 52(1)
osm-3 (p802) None 20.2 ± 6.6 +21 32.1 ± 4.1 +38 161(4)
ETH(2) 22.1 ± 7.1 +9 34.0 ± 3.6 +6 131(3)
TRI(4) 23.6 ± 5.4** +17 32.7 ± 3.0 +2 30(1)
eat-2 (ad465) None 20.1 ± 6.3 +20 31.4 ± 3.1 +35 192(4)
TRI(4) 28.6 ± 10.0*** +42 47.2 ± 7.3** +50 68(1)
  • View inline Concentrations of 0.5, 5, or 10 mg/ml succinimide also did not significantly increase mean life-span.

  • Ethosuximide is a small heterocyclic ring compound that prevents absence seizures in humans and has been a preferred drug for treating this disorder since its introduction in the 1950s (4, 5) (Fig. 1A). An important question is whether the anticonvulsant activity in humans and the life-span extension activity in worms have a similar mechanism. If this is the case, then other drugs with similar structures and anticonvulsant activity might also affect life-span. Trimethadione and 3,3-diethyl-2-pyrrolidinone (DEABL) have anticonvulsant activity and structures similar to that of ethosuximide (4, 6) (Fig. 1A). Trimethadione is approved for human use and the treatment of absence seizures. DEABL is not used to treat humans. Both compounds caused significant extensions of mean and maximum life-span (Fig. 1, C and D, and Table 1). Trimethadione caused the largest extension of mean (47%) and maximum (57%) life-span of the three compounds. Succinimide, similar in structure but lacking in anticonvulsant activity in vertebrates, did not extend life-span (Fig. 1A and Table 1). These findings suggest that ethosuximide, trimethadione, and DEABL may extend life-span by a similar mechanism that may be related to the mechanism of anticonvulsant activity.

    For the treatment of seizures, the therapeutic range of ethosuximide in humans is 40 to 100 μg/ml (5). Worms cultured with an external concentration of 2 mg/ml ethosuximide had an internal concentration (±SD) of 30.5 ± 22.2 μg/ml. This value is near the therapeutic range, suggesting that the anti-convulsants may have similar targets in worms and humans.

    To determine the developmental stage at which the drugs function to extend life-span, trimethadione was administered from fertilization until the L4 stage or from the L4 stage until death. Exposure to trimethadione only during embryonic and larval development had no effect on life-span. In contrast, exposure to trimethadione only during adulthood caused a significant extension of mean life-span (24%) (Fig. 1D and Table 1).

    To determine whether these drugs delay age-related declines of physiological processes, we analyzed self-fertile reproduction, body movement, and pharyngeal pumping. The declines of pharyngeal pumping and body movement are positively correlated with each other and with life-span (7). The decline of self-fertile reproduction is not correlated with life-span, suggesting that this age-related change is regulated independently (7). Treatments with ethosuximide and/or trimethadione significantly extended the span of time that animals displayed fast body movement, fast pharyngeal pumping, and any pharyngeal pumping (Fig. 2, B to D and F). Neither compound significantly extended the span of time that animals displayed self-fertile reproduction (Fig. 2, A and F). These measurements can be used to define stages of aging (7). Both compounds extended Stage II, the postreproductive period characterized by vigorous activity (Fig. 2E). Trimethadione also extended Stage IV, the terminal phase characterized by minimal activity. These findings indicate that ethosuximide and trimethadione delay the aging process.

    Fig. 2.

    Anticonvulsants delay age-related declines of physiological processes. Wild-type hermaphrodites were cultured with no drug (WT), 2 mg/ml ethosuximide (+ETH), or 4 mg/ml trimethadione (+TRI). We measured the time from L4 to the cessation of self-fertile progeny production (A), to the cessation of fast body movement (B), to the cessation of fast pharyngeal pumping (≥25 contractions per 10 s) (C), and to the cessation of all pharyngeal pumping (≥1 contraction per 10 s) (D). (E) Stages I to IV end at the mean self-fertile reproductive span, the mean fast body movement span, the mean pharyngeal pumping span, and the mean life-span, respectively. (F) Mean values in days ± SD for data in (A) to (D). Stars indicate P values compared with no drug (Table 1).

    Several genetic and environmental manipulations can extend C. elegans life-span. To investigate the relationships between the anti-convulsants and these regulators of aging, we examined the effect of combining two treatments. Worms cultured on nonpathogenic Bacillus subtilis or ultraviolet (UV)–irradiated E. coli display an extended life-span (8, 9). Trimethadione extended the life-span of worms cultured on B. subtilis and UV-irradiated E. coli (Table 1), indicating that the primary mechanism of the anticonvulsant life-span extension is not a reduction of bacterial pathogenicity.

    Nutrient limitation extends life-span and can be caused by a mutation of the eat-2 gene that is important for pharyngeal pumping (1, 10, 11). Trimethadione significantly extended the life-span of eat-2 mutants (42%) (Fig. 1E and Table 1), indicating that the primary mechanism of life-span extension is not nutrient limitation. Furthermore, wild-type animals treated with ethosuximide or trimethadione were not nutrient limited, because they displayed normal pharyngeal pumping, food ingestion, and body morphology (they did not appear thin or starved), and they produced an approximately normal number of progeny (3).

    An insulin-like signaling pathway regulates C. elegans life-span. This pathway requires the function of sensory neurons that may mediate the release of an insulin-like ligand, the daf-2 insulin-like growth factor (IGF) receptor gene, and a signal transduction cascade that regulates the daf-16 forkhead transcription factor gene. Loss-of-function daf-16 mutations reduce life-span and suppress the life-span extensions caused by mutations in upstream signaling pathway genes such as daf-2 (12). Treatment with ethosuximide or trimethadione significantly extended the life-span of two loss-of-function mutants, daf-16(m26) (16%) and daf-16(mu86) (11 to 21%) (Fig. 1, F and G, and Table 1), although the percentage change caused by trimethadione was less than that in wild-type animals (47%). These results indicate that part of the anticonvulsant action is independent of daf-16. Part of the anti-convulsant action may require daf-16. However, the reduced effect of trimethadione is consistent with other possibilities, such as deleterious consequences of combining a mutation and a drug that both cause pleiotropic effects (13).

    Life-span extension is caused by loss-of-function mutations of genes important for the function of sensory neurons (osm-3 and tax-4), for neurotransmission (unc-31, unc-64, and aex-3), and for transmission of the insulin-like signal (daf-2) (12, 14, 15) (Table 1). Ethosuximide and/or trimethadione significantly increased the life-span of osm-3, tax-4, unc-31, unc-64, aex-3, and daf-2 loss-of-function mutants from 8 to 36% (Fig. 1, H to M, and Table 1). These results indicate that part of the anticonvulsant action may be different than the action of these mutations. The effects of ethosuximide and/or trimethadione were only partially additive with several mutations, notably daf-2, unc-64, and osm-3. Thus, part of the activity of the anticonvulsants may be similar to the effects of these mutations, several of which affect neural function. However, an absence of full additivity is also consistent with other possibilities (13).

    Anticonvulsants affect the neural activity of vertebrates. To determine whether these drugs have a similar activity in nematodes, we analyzed neuromuscular behaviors. C. elegans egg laying is mediated by HSN neurons that innervate the vulval muscles (16, 17). Wild-type hermaphrodites lay eggs that have matured to about the 30-cell stage of development. Trimethadione and ethosuximide caused wild-type hermaphrodites to lay eggs at much earlier stages of development, often the 1- to 7-cell stage (Fig. 3C). The control drug, succinimide, did not stimulate egg laying (Fig. 3C). A delay in egg laying can result in an egg-laying defective (Egl) phenotype characterized by progeny that hatch internally. Approximately 8.9% of wild-type hermaphrodites displayed an Egl phenotype during their lifetime; ethosuximide and trimethadione reduced this to 2.9 and 1.2%, respectively (Fig. 3A). To investigate whether the anticonvulsants act presynaptically on the HSN neurons or postsynaptically on the vulval muscles, we analyzed an egl-1 mutant that lacks HSNs as a result of a developmental abnormality (17). Ethosuximide did not cause egl-1 mutants to lay eggs at earlier stages of development (Fig. 3D), indicating that the vulval muscles are not sufficient and the HSN neurons are necessary for the anticonvulsant to stimulate egg laying. This result is consistent with the model that the drug acts presynaptically.

    Fig. 3.

    Anticonvulsants stimulate neuromuscular activity. (A) The percent of dead hermaphrodites that displayed internally hatched progeny (Egl) with no drug (WT), 2 mg/ml ethosuximide (+ETH), or 4 mg/ml trimethadione (+TRI) (n > 150). (B) Motility of wild-type young adult hermaphrodites with no drug (n = 13), 2 mg/ml ethosuximide (n = 8), 4 mg/ml trimethadione (n = 14). Stars indicate P values compared to no drug (Table 1). (C and D) The developmental stage of embryos at the time of egg laying. (C) Wild-type young adult hermaphrodites treated with no drug (n = 107), 2 mg/ml ethosuximide (n = 92), 4 mg/ml trimethadione (n = 119), or 2 mg/ml succinimide (+SUC) (n = 44). (D) egl-1(n487) young adult hermaphrodites treated with no drug (n = 37) or 2 mg/ml ethosuximide (n = 41). (E) A time course of paralysis induced by aldicarb in wild-type young adult hermaphrodites treated with no drug (n = 99) or 4 mg/ml trimethadione (n = 99).

    Treatment with ethosuximide or trimethadione caused wild-type hermaphrodites to display hyperactive motility, indicating that these drugs stimulate neuromuscular activity (Fig. 3B). To analyze this phenotype, we examined sensitivity to the acetylcholinesterase inhibitor aldicarb. Aldicarb causes paralysis of body movement resulting from the accumulation of acetylcholine at the neuromuscular junction (18). Mutations that reduce synaptic transmission cause resistance to aldicarb (18). In contrast, mutations that stimulate synaptic transmission cause hypersensitivity to aldicarb-mediated paralysis (19). Trimethadione treatment of wild-type animals caused hypersensitivity to aldicarb-mediated paralysis (Fig. 3E). The control drug, succinimide, did not cause hyperactive motility or aldicarb hypersensitivity (3). These results indicate the anticonvulsants stimulate synaptic transmission in the neuromuscular system that controls body movement.

    Ethosuximide and trimethadione effectively treat absence seizures in humans by regulating neural activity. A likely target of ethosuximide is T-type calcium channels, although it is possible that these compounds act on multiple targets (2022). These anti-convulsants also affected neural activity in nematodes, and the anticonvulsant and the life-span extension effects of the compounds may act through similar mechanisms. The findings presented here are consistent with the model that the effect on neural activity causes the life-span extension, although they do not exclude the possibility that the drugs affect neural activity and aging by different mechanisms. Furthermore, the interactions with the insulin-signaling mutants suggest the intriguing possibility that neural activity regulates aging by both daf-16–dependent and daf-16–independent mechanisms.

    Supporting Online Material

    Materials and Methods

    Fig. S1


    References and Notes

    Stay Connected to Science

    Navigate This Article