The Ras-Erk-ETS-Signaling Pathway Is a Drug Target for Longevity

Abstract

Identifying the molecular mechanisms that underlie aging and their pharmacological manipulation are key aims for improving lifelong human health. Here, we identify a critical role for Ras-Erk-ETS signaling in aging in Drosophila. We show that inhibition of Ras is sufficient for lifespan extension downstream of reduced insulin/IGF-1 (IIS) signaling. Moreover, direct reduction of Ras or Erk activity leads to increased lifespan. We identify the E-twenty six (ETS) transcriptional repressor, Anterior open (Aop), as central to lifespan extension caused by reduced IIS or Ras attenuation. Importantly, we demonstrate that adult-onset administration of the drug trametinib, a highly specific inhibitor of Ras-Erk-ETS signaling, can extend lifespan. This discovery of the Ras-Erk-ETS pathway as a pharmacological target for animal aging, together with the high degree of evolutionary conservation of the pathway, suggests that inhibition of Ras-Erk-ETS signaling may provide an effective target for anti-aging interventions in mammals.

Graphical abstract

Figure 1

Ras Inhibition Functions Downstream of IIS to Extend Lifespan (A) Chico protein sequence with amino acid substitutions used to generate the Chico-Grb2/Drk- and Chico-PI3K-binding site mutants. (B) BiFC in S2 cells co-expressing the indicated Chico constructs with the Drosophila Drk protein. Proportion of YFP-positive cells ± SE; n numbers are indicated above each bar; ^∗∗p < 0.005 Chi-square test to no insulin control. Scale bar, 10 μm. (C) Egg-to-adult development time of the indicated genotypes. See also Figure S1. (D) Fresh weight of adult females of the indicated genotypes. Mean body mass (n = 10 for each genotype) ± SEM, Anova, p < 0.0001, ^∗∗p < 0.05 t test (compared to WT). (E) Survival of wild-type and chico/+ heterozygous females carrying the indicated chico genomic rescue constructs. chico/+ flies were long-lived compared to WT (p = 0.0006), which was rescued by the chico rescue construct (p = 0.58). Both the Grb2/Drk mutant and the PI3K mutant failed to rescue the longevity of chico/+ flies (compared to WT construct, p = 8.36 × 10⁻¹⁰ and p = 1.32 × 10⁻⁹, respectively). See Table S1A. (F) Expression of constitutively active Ras blocks the beneficial effects of chico mutation on survival. daGS/chico > ras^[CA] flies show increased lifespan compared to daGS > ras^[CA] in the absence of RU486 (p = 3.07 × 10⁻¹⁸), but not in the presence of RU486 (p = 0.18). See Table S1B.

Figure 2

Direct Inhibition of Ras-Erk Signaling Extends Lifespan (A) Ubiquitous adult-onset expression of ras^[DN] increases lifespan (p = 1.26 × 10⁻⁷). See Table S2A. (B) Ubiquitous adult onset of ras85D knockdown by RNAi increases lifespan (p = 2 × 10⁻³). See Table S2B. (C) Ubiquitous adult-onset knockdown of rl expression by RNAi increases lifespan (p = 1.46 × 10⁻⁷). See Table S2C. (D) Adult gut/fat body expression of ras^[DN] extends lifespan (p = 3 × 10⁻³). See Table S2D. (E) Adult gut/fat body knockdown of ras85D expression by RNAi extends lifespan (p = 4 × 10⁻⁴). See Table S2E. (F) Adult gut/fat body knockdown of rl expression by RNAi extends lifespan (p = 10⁻⁵). See Table S2F.

Figure 3

Aop Functions Downstream of Ras in the Adult Fly to Control Lifespan (A) Nuclear localization of Aop protein (red) increases in intensity in dissected adult abdominal fat bodies of S₁106 > ras^[DN] females fed RU486. (n = 5, p = 5 × 10⁻³, t test). Nuclei are stained with DAPI (white). Scale bar, 25 μm. Intensity quantifications are shown as box plots. (B) Box plots of log-transformed levels of lacosta (lcs) and CG1678 mRNAs relative to actin in females of the indicated genotypes. S₁106 > aop^[ACT]: mixed effects linear model (MELM) (n = 3–4): significant effect of RU486 (p < 10⁻⁴), transcript (p = 2 × 10⁻³) and their interaction (p = 4 × 10⁻³), with significant differences between – and + RU486 for each transcript (p < 0.05, post hoc t test); S₁106>pnt^[P1]: significant effect of RU486 only (p = 0.02, n = 4, MELM); S₁106>ras^[DN] significant effect of RU486 only (p = 0.02, n = 4, MELM). (C) S₁106 > ras^[DN] show increased lifespan in the presence of RU486 (p = 2 × 10⁻⁶), but S₁106 > ras^[DN]aop^[RNAi] females do not (p = 0.95). See Table S3A. (D) S₁106 > aop^[ACT] females show increased lifespan in the presence of RU486 (p = 2 × 10⁻⁵). S₁106 > aop^[ACT]ras^[DN] females also show increased lifespan in the presence of RU486 (p = 4x10⁻⁶) but are no longer lived than S₁106 > aop^[ACT]. See Table S3B.

Figure 4

The Ras-Regulated Transcription Factor, Aop, Is Required Downstream of chico to Increase Survival (A) chico extends lifespan in daGS > aop^[RNAi] females in the absence of RU486 (p = 1.2 × 10⁻⁵), but not in the presence of RU486 (p = 0.95). See Table S4A. (B) chico extends lifespan in daGS > pnt^[P1] females in the absence of RU486 (p = 2.34 × 10⁻⁹), but not in the presence of RU486 (p = 0.64). See Table S4B. (C) The effects of chico/₊ mutation and S₁106 > aop^[ACT] are not additive for lifespan (S₁106 > aop^[ACT] versus S₁106/chico > aop^[ACT] in the RU486− condition, p = 8.84 × 10⁻⁷; S₁106 > aop^[ACT] RU486− versus RU486+ conditions, p = 3.20 × 10⁻¹³; S₁106/chico > aop^[ACT] RU486− versus RU486+, p = 0.29). See Table S4C. (D) dfoxo is required for chico/+ lifespan extension. chico/+ females show increased lifespan (p = 3.71 × 10⁻⁶), but not in the absence of dfoxo (p = 0.63). See Table S4D.

Figure 5

Pharmacological Inhibition of Ras-Erk Signaling Using Trametinib Extends Lifespan (A) Drosophila S2 cells treated with trametinib (Tram) at the indicated concentrations show inhibition of insulin-stimulated (Ins) Erk phosphorylation. (B) S2 cells treated with a high concentration of trametinib (10 μM) show no inhibition of insulin-dependent phosphorylation of Akt or S6K. (C) Effects of trametinib dose on female egg laying. Means ± SEM, ^∗∗p < 0.05 t test (compared to 0 μM condition). (D) Trametinib extends lifespan in wDah females.(1.56 μM, p = 2.65 × 10⁻⁴; 15.6 μM, p = 1.92 × 10⁻¹⁰). See Table S5A. (E) Later-life (day 30) post-reproductive (see also Figure S5) exposure to 15.6 μM trametinib extends lifespan (p = 5.02 × 10⁻⁵). See Table S5B. (F) Trametinib increases survival of daGS > ras^[CA] (p = 1.24 × 10⁻³⁹), but not daGS > pnt^[P1] flies (p = 0.59) in the presence of RU486. See Table S5C. (G) Age-related changes in gut ISC proliferation in animals exposed to trametinib for 15 or 65 days. Mean number of PH3+ cells per gut ± SEM. Number of guts analyzed are indicated above each bar. GLM with Poisson distribution and overdispersion parameter: significant effect of age (p < 0.001), but not trametinib concentration (p = 0.84) or their interaction (p = 0.79). (H) Age-related changes in intestinal integrity in animals exposed to trametinib for 15 or 65 days. Proportion of smurfs present in the population ± SE. Total numbers of flies examined for each condition are indicated above each bar. GLM with binomial distribution and overdispersion parameter: significant effect of age (p < 0.001), but not trametinib concentration (p = 0.84) or their interaction (p = 0.79). (I) No significant differences in survival of GS5961 > aop^[ACT] females in the presence or absence of RU486 (p = 0.45). See Table S5D.

Figure 6

Model of Aop-Foxo Function Downstream of IIS We propose that, downstream of the insulin receptor substrate, Chico, signaling via the IIS pathway bifurcates into two branches: Ras-Erk and PI3K-Akt. At the transcriptional level, these two branches subsequently re-join, acting on the Aop and Foxo TFs in a non-additive manner. The two TFs then co-operatively regulate the expression of a subset of target genes required for lifespan extension.

Figure S1

Expression Levels and Developmental Phenotypes of chico Genomic Rescue Constructs, Related to Figure 1 (A) Log-transformed levels of chico mRNA relative to actin in females of the indicated genotypes. chico expression was significantly reduced in chico mutants (p < 0.05, t test compared to WT controls), but not in the presence of any of the genomic rescue constructs. Means ± SEM. (B–D) The PI3K-binding site, but not the Grb2/Drk-binding site, of Chico is essential for female sterility (B) and increased metabolic stores (C and D). Means ± SEM, ^∗∗p < 0.05, t test (compared to WT controls).

Figure S2

Effects of Yeast Concentration on the Survival of Flies Expressing Constitutively Active Ras or Constitutively Active Pointed, Related to Figures 1 and 4 (A) Survival of daGS > ras^[CA] flies in the presence of RU486 under different dietary yeast concentrations. Log-rank detected a significant increase in survival between 0.5x and 1.0x yeast concentrations (increase in median lifespan of 13%, p = 6.28x10⁻⁸) and between 1.0x and 2.0x yeast concentrations (increase in median lifespan of 11%, p = 7.41x10⁻⁴). 0.5x yeast: n = 97 deaths/0 censors, median lifespan = 16 days; 1.0x: n = 89/0, median lifespan = 18 days; 2.0x: n = 86/0, median lifespan = 20 days. (B) Survival of daGS > pnt^[P1] flies in the presence of RU486 under different dietary yeast concentrations. Log-rank detected a significant increase in survival between 1.0x and 0.5x yeast concentrations (increase in median lifespan of 40%, p = 4.68x10⁻¹⁴) and between 2.0x and 1.0x yeast concentrations (increase in median lifespan of 150%, p = 1.62x10⁻³⁷). 0.5x yeast: n = 163 deaths/0 censors, median lifespan = 28 days; 1.0x: n = 159/0, median lifespan = 20 days; 2.0x: n = 152/0, median lifespan = 8 days.

Figure S3

Mortality Analysis of Flies Exposed to 15.6 μM Trametinib, Related to Figure 5 (A) Combined survival data from three independent trials of wDah females exposed to 15.6 μM trametinib. Log-rank detected a significant difference in survival between trametinib-treated flies and untreated controls (increase in median lifespan of 9%, p < 0.0001). Controls: n = 338 deaths/8 censors, median lifespan = 75 days; trametinib-treated: n = 325/28, median lifespan = 82 days. (B) Age-specific mortality analysis of the survival data in (A). Mortality (μx) was calculated as: μx = -ln(1-qx), where qx (the probability of dying in a time interval) was averaged per day over a 5-day interval and is shown at the end of the interval. Intervals with zero deaths were removed. Estimates of the parameters of the Gompertz mortality model: λ (baseline mortality) controls = 7.3x10⁻⁶, trametinib-treated = 2.3x10⁻⁶, p = 0.045; γ (change in mortality with age) controls = 0.13, trametinib-treated = 0.13, p = 0.67.

Figure S4

Trametinib Does Not Affect Feeding Behavior, Related to Figure 5 Feeding events as a proportion of total events observed per vial for wDah females on food containing 0 μM, 1.56 μM and 15.6 μM trametinib (15 vials per condition). Data were analyzed using a Generalized Linear Model with binomial distribution and overdispersion parameter. There was no significant effect of trametinib concentration on feeding (p = 0.66).

Figure S5

Egg Laying Over Time in wDah Females, Related to Figure 5 Egg-laying in wDah females peaks at around day 10 post-mating and then declines to plateau by day 28. Means ± SEM.

Figure S6

Trametinib Extends the Lifespan of daGS > ras^[CA] and daGS > pnt^[P1] Flies in the Absence of Transgene Expression, Related to Figure 5 (A and B) In the absence of RU486 and hence transgene expression, trametinib treatment at 15.6 μM increased the survival of daGS > ras^[CA] flies (increase in median lifespan of 9%, p = 2.07x10⁻⁸) and daGS > pnt^[P1] flies (increase in median lifespan of 9%, p = 2.69x10⁻⁷). Untreated daGS > ras^[CA]: n = 90 deaths/6 censors, median lifespan = 82 days; trametinib-treated daGS > ras^[CA]: n = 95/1, median lifespan = 89 days; untreated daGS/ pnt^[P1], n = 141/6, median lifespan = 79 days; trametinib-treated daGS> pnt^[P1], n = 132/19, median lifespan = 86 days.