To seek out potential drivers of age-related changes in gene expression, Budovskaya et al. 3 carried out microarray studies in C. elegans at 4, 7, 10, and 14 days of adulthood, finding 1,254 genes that change expression during aging. (By and large, gene expression decreases in older animals - a general pattern similar to that seen in human subjects 7.) Microarray studies of aging nematodes have been carried out previously (for reviews, see 89), with results largely consistent with the current study. However, instead of simply categorizing the genes identified, Budovskaya et al. then looked for common patterns of transcriptional control. As in other work (for example, in C. elegans 10 and yeast 11), computational techniques were brought to bear on the identification of DNA sequence motifs enriched in the regions upstream of genes suspected to be co-regulated. Using the CompareProspector tool, which minimizes false positives by looking for motifs with evolutionary conservation 12, the investigators found conserved GATA motifs upstream of approximately half of the genes that change expression with age.

The authors also found that GATA motifs were enriched in genes identified by previous aging microarray studies 13 and in genes identified by microarray analyses of mutations in the insulin/IGF-I pathway. Somewhat surprisingly, the overall trend that emerged from this analysis is that long-lived insulin/IGF-I pathway mutants (which have a salubrious increase in DAF-16 activity) have gene-expression patterns similar to those of old animals, whereas gene-expression patterns in daf-16 mutants tend towards those in young animals. We hope that this, perhaps counterintuitive, result, which has now been replicated by two aging time-courses 13, will provoke future studies.

Budovskaya et al. 3 then selected 10 of the 14 known C. elegans GATA transcription factors to examine for potential longevity phenotypes. RNAi knockdowns of these yielded no longevity phenotype; however, three factors - ELT-3, EGR-1 and EGL-27 - were shown to suppress the phenotype of a daf-2 mutant, indicating that, although their loss does not impair lifespan in lab conditions, they are required for the lifespan-prolonging effects of decreased insulin/IGF-I signaling. Furthermore, knockdown of ELT-3 also suppresses the longevity of the feeding-deficient eat-2 mutant, suggesting a second longevity pathway that requires ELT-3. (The dietary-restriction-induced longevity of eat-2 does not require active DAF-16 and is additive with daf-2 mutations, indicating that these pathways are independent 14.)

The authors then examined several of the 602 age-regulated, conserved-GATA-bearing genes they had identified and confirmed that the majority of those examined were downstream targets of elt-3; furthermore, all genes so examined were regulated via the GATA sites in their promoter regions. Together, these data indicate that elt-3 is indeed a regulator of several of age-related genes; further studies to establish whether elt-3 plays a role in the regulation of the remaining age-modulated genes may prove enlightening.

Although elt-3 was not among the age-regulated genes identified by Budovsakaya et al. 3 in their microarray experiment, they found that an elt-3::GFP reporter shows an age-related decline in expression, indicating that perhaps the decline in elt-3 expression is responsible for driving some portion of the age-related decline in overall gene expression - overexpression of elt-3 may thus provide several interesting phenotypes. As a next step, the authors sought to identify the upstream regulator of elt-3 responsible for its age-related decline. RNAi experiments showed that age-1 (and by implication, the insulin/IGF-I pathway) exerts a negative control over the expression of elt-3; this regulation was independent of age, however. One theory of aging holds that senescent decline is driven by accumulation of damage from a lifetime of endogenous and environmental stressors (see, for example 15161718); perhaps, therefore, stressors such as heat shock, oxidation, DNA damage or bacterial infection might downregulate elt-3? As it happens, the answer is no: Budovskaya et al. 3 did not find evidence for this hypothesis.

Another theory of aging notes that as organisms age toward a post-reproductive phase, the force of natural selection weakens. Thus, alleles that have beneficial early-life but detrimental late-life phenotypes can be evolutionarily advantageous, and genes that are silenced after a certain developmental stage may become derepressed later in life 19. During development, elt-3, which itself has GATA motifs in its promoter region, is regulated by several other GATA factors: elt-1, a positive regulator of elt-3 5; and elt-5 and elt-6, which are negative regulators 4. As such, antagonistic pleiotropy theories of aging suggest that this transcriptional circuit may become dysregulated in aging animals in a way that is immune to the effects of natural selection. Indeed, Budovskaya et al. 3 found tantalizing evidence for age-related dysregulation of the elt circuit. For one thing, the expression of elt-5 and elt-6, which are downregulated at the beginning of adulthood, drifts upward with time. Moreover, RNAi against these factors in adult C. elegans prevents much of the age-related decline in elt-3 expression. Another piece of evidence is the fact that such RNAi substantially prolongs lifespan compared with wild-type animals, and does so in an elt-3-dependent manner. As expected by antagonistic pleiotropy theories of aging, RNAi against elt-5 and elt-6, while lifespan-extending in older animals, is detrimental in younger ones.

What, then, drives senescence in elderly animals? Do the parts simply wear out, or is gene dysregulation to blame? Many genes in well-known longevity pathways such as insulin/IGF-I signaling, mitochondrial oxygen transport, or the response to dietary restriction seem to be involved in modulating physiological stress responses 220. Indeed, the ability to mobilize an effective heat-shock response is a potent marker for the eventual longevity of an individual C. elegans 21. These and other observations 15 augur well for the damage-accumulation view of aging, in which stress-response and damage repair are key to lifespan extension. On the other hand, the lack of a well-defined role for these pathways in normal aging, despite intensive investigation, is most curious. The identification by Budovskaya et al. of age-related transcriptional (dys)regulation, in a protein required for various lifespan-prolonging interventions, certainly suggests that antagonistic pleiotropy may play a significant part in aging.

Several observations in this work, however, suggest that these two views on aging need not be mutually exclusive. First, the authors note that the long lifespan of daf-2 mutants (insulin/IGF-I signaling deficient) and eat-2 mutants (feeding deficient) both require expression of elt-3. These pathways are generally thought to be genetically independent 14, so any common thread must be fairly far downstream, perhaps in the stress-response pathways. This suggests that future experiments to identify more closely the positions of elt-3, elt-5, and elt-6 in these genetic pathways will be most informative. One perspective on the insulin/IGF-I pathway is that it serves to repress the stress response in times of plenty 20; thus, repression of elt-3 by insulin/IGF-I signaling is consistent with a role for that gene in stress responses. The similarity of gene-expression patterns in old age (after accumulation of many cellular stressors) and those in insulin/IGF-I signaling deficient mutants (with derepressed stress responses) may also be telling. (A comparison of stress-regulated and aging-related genes would be illuminating on this point.) Finally, Budovskaya et al. 3 observe that elt-3-null C. elegans are more sensitive to heat shock and oxidative stress than are wild-type animals, whereas elt-5 RNAi (which increases the level of elt-3) causes a slight stress resistance. If elt-3 were involved in regulating stress responses, its downregulation later in life (through dysregulation of elt-5 and elt-6 levels) might lead to impairment of damage repair and decreased longevity. Although theoretical arguments have suggested that antagonistic pleiotropy is most likely to appear in regulators of stress-response and damage-repair pathways 22, this work may be the first demonstration of that principle.