|Home | About | Journals | Submit | Contact Us | Français|
Early life experiences have a major impact on adult phenotypes [1–3]. However, the mechanisms by which animals retain a cellular memory of early experience are not well understood. Here we show that adult wild-type C. elegans that transiently passed through the stress-resistant dauer larval stage exhibit distinct gene expression profiles and life history traits, as compared to adult animals that bypassed this stage. Using chromatin immmunoprecipitation experiments coupled with massively parallel sequencing, we find that genome-wide levels of specific histone tail modifications are markedly altered in post-dauer animals. Mutations in subsets of genes implicated in chromatin remodeling abolish, or alter, the observed changes in gene expression and life history traits in post-dauer animals. Modifications to the epigenome as a consequence of early experience may contribute in part to a memory of early experience, and generate phenotypic variation in an isogenic population.
Early developmental or environmental experience profoundly affects adult behaviors. The effects of modality-specific experience during early ‘critical periods’ for the correct development of defined neural circuits is well-studied (reviewed in ). Early experiences also have more general effects on animal behavior and development. For instance, childhood abuse has been linked to increased rates of mental and behavioral disorders in adult humans [2, 3]. The molecular mechanisms by which early experience results in global, long-lasting changes in the adult are poorly understood.
Early environmental experience plays a major role in the lifecycle of the nematode C. elegans. Environmental cues are assessed prior to the first larval molt to regulate the decision between entry into the stress-resistant dauer developmental stage, or continuation in the reproductive cycle at the second larval molt (Figure 1A) [5, 6]. When conditions improve, dauer larvae exit the dauer stage and resume reproductive growth. However, despite their starkly distinct developmental histories, phenotypic differences between adults that bypassed the dauer stage (henceforth referred to as control animals), and animals that transiently passed through the dauer stage (henceforth referred to as post-dauer animals) (Figure 1A), have previously been uncovered only in mutant backgrounds [7–9], leaving open the question of whether a memory of developmental history also influences phenotypes in wild-type control and post-dauer adults.
To address this issue at the organismal level, we first investigated whether control and post-dauer wild-type adult animals exhibited global differences in gene expression. To ensure that populations were matched for age and growth conditions, we developed a protocol for precise regulation of dauer entry and exit (Figure S1), and compared the expression profiles of control and post-dauer animals that had remained in the dauer stage for 1 day. The expression of 1,181 and 946 genes was significantly up or down-regulated, respectively, in post-dauer animals compared to controls (Tables S1 and S2). Expression changes in post-dauer animals ranged from an 8-fold decrease to a 5-fold increase compared to control expression levels; however, ~60% of these genes exhibited expression changes that were altered by less than 1.5-fold (Figure 1B; Tables S1 and S2). Since the expression profiles of whole adult animals were examined, we expect that these smaller expression changes are biologically relevant, and may result from an overall small change in expression in many tissues, or large changes in a subset of expressing cells. Expression changes predicted from microarray hybridization experiments were further verified via qRT-PCR (Figure S2).
In order to determine whether specific biological processes were preferentially affected upon passage through the dauer stage, we identified the Gene Ontology (GO) terms  associated with each gene exhibiting altered expression in post-dauer animals. Affected genes were associated with multiple cellular and biological pathways; over-represented categories included genes predicted to encode proteins involved in regulation of the cell cycle, transcription, phosphorylation and dephosphorylation, metabolism, reproduction, and G-protein coupled receptor signaling (Table S3). Expression changes of subsets of these genes in different categories are summarized in Table S2.
Since one of the largest groups of genes identified was associated with reproduction, we further compared this data set with genes previously identified as sperm or oocyte-enriched in expression profiling experiments . The expression of 23% of genes identified as sperm-enriched was significantly down-regulated in post-dauer animals, whereas the expression of only 1% was up-regulated (Figure 1B). Notably, the expression of 65 of 88 predicted major sperm cytoskeletal protein (msp) genes implicated in oocyte maturation, sheath cell contraction, ovulation and sperm movement [12, 13] was down-regulated in post-dauer animals (Table S1). In contrast, the expression of 32% of oocyte-enriched genes was up-regulated in post-dauer animals, and none were down-regulated (Figure 1B; Table S1). Passage through the dauer stage has been shown to increase male survival and facultative outcrossing rates in some predominantly self-fertilizing C. elegans isolates . Together, these results suggest that changes in gene expression may contribute to altered life history traits such as reproduction in post-dauer adults.
To determine when altered gene expression profiles arise in post-dauer adults, we examined expression levels of the set of 2,127 genes described above in dauer animals that were in the dauer stage for 1 day (Figure 1C). Based on expression in control, dauer and post-dauer animals, these genes were further clustered into 4 groups (Figure 1D; Table S1). Genes in group 1 (164 genes) exhibited similar expression levels in post-dauer adults and dauer animals, indicating that the expression levels of these genes was altered during the dauer stage and stably maintained thereafter (Figure 1C,D). The expression levels of genes in group 2 (143 genes) were similar in dauer animals and control adults, but altered (either up- or down-regulated) in post-dauer animals (Figure 1D), indicating that altered expression of this gene set in post-dauer adults was likely to occur following exit from the dauer stage. Group 3 (400 genes) included genes whose expression levels were either up- or down-regulated in both dauer and post-dauer animals, compared to controls (Figure 1D). Expression of members of this group was altered in the dauer stage, and was further up- or down-regulated in post-dauer adults. The largest group (Group 4; 1,420 genes) included genes whose expression levels were distinct and uncorrelated in control, dauer and post-dauer adults (Figure 1D). These results indicate that the altered gene expression observed in post-dauer animals arises from multiple regulatory mechanisms acting both during the dauer stage, as well as upon subsequent resumption of reproductive growth. Moreover, since genes associated with specific GO functional terms were overrepresented in different groups (Table S4), genes with related functions may be regulated via similar mechanisms. Previous work identified genes whose expression is altered over a 12 hour time course during the transition from the dauer to the post-dauer stage . However, there was little overlap of genes identified in this work with either the dauer or the post-dauer gene sets described here (Figure S3), likely due to differences in experimental methods (; see Supplemental Experimental Procedures).
The expression of chromatin-associated genes such as histones was previously shown to be altered in the dauer stage , and we identified genes implicated in nucleosome assembly and chromatin remodeling in the gene set whose expression is altered in post-dauer animals (Tables S1 and S2). We investigated whether the observed developmental experience-dependent changes in gene expression profiles were associated with global alterations in the epigenome. We performed ChIP-Seq (chromatin immunoprecipitation followed by sequencing) [17, 18] analyses to generate genome-wide maps of histone modifications using DNA isolated from populations of control and post-dauer animals grown under conditions similar to those used for transcriptional profiling (Figure S1). ChIP was performed using antibodies directed against histone H3 to assess nucleosome density, as well as against histone modifications associated with euchromatin [pan-acetylation of histone H4 (H4ac) and trimethylation of histone H3 at lysine 4 (H3K4me3)], and heterochromatin (trimethylation of histone H3 at lysine 9 (H3K9me3) and lysine 27 (H3K27me3)].
We first correlated histone modifications in gene regulatory and coding sequences with overall expression levels in control and post-dauer animals. Since few gene transcriptional start sites are known in C. elegans due to trans-splicing , we instead averaged the number of ChIP-Seq reads for each histone modification in the regions 2 kb up- and downstream of the translation initiation sites (TISs), and compared the histone modification profiles with gene expression levels. Nucleosome occupancy and overall histone H3 levels were similar in control and post-dauer animals regardless of gene expression levels (Figure 2A; Figure S4), suggesting that histone modifications, rather than nucleosome content, may be correlated with gene activity under the examined conditions.
Unexpectedly, we found that H3K4me3 and H4ac modifications were decreased genome-wide in post-dauer animals, despite similar overall gene expression levels (Figure 2B–C, Figure S5A, B; Table S5). This decrease was primarily observed for highly expressed genes and not genes expressed at lower levels, perhaps due to a floor effect in detection sensitivity. H3K27me3 and H3K9me3 levels were similar in both control and post-dauer animals (Figure 2D, E; Table S5). H3K4me3 and H4ac modifications were positively correlated with gene expression levels across gene regulatory and coding sequences in both control and post-dauer animals (Figure 2B,C, Figure S5A, B; Table S5) . In contrast, H3K27me3 and H3K9me3 modifications were not strongly correlated with gene expression levels, (Figure 2D, E, Figure S5A, B; Table S5), as has previously been observed in human CD4+ T cells . However, H3K9me3 levels were previously shown to be enriched across genes expressed at low levels in C. elegans . Differences in the developmental stages of animals, or the methodologies used in the two studies may account in part for these different observations.
To further examine the observed changes in chromatin state in post-dauer animals, we selected individual genes from the highest expressed category, and quantified levels of H3K4me3 and H4ac modifications across their regulatory and coding sequences. Average levels of both modifications were decreased across both upstream and downstream sequences of examined genes (Figure 3A), although their overall expression levels were unaltered. To verify the chromatin modification level changes, we performed chromatin immunoprecipitation from a biologically independent population of control and post-dauer adult animals, and quantified histone modification levels across the regulatory sequences by quantitative PCR. We again observed decreases in both H3K4m3 and H4ac levels in the sequences examined (Figure 3B). Taken together, these results imply that the genome-wide chromatin modification profiles in post-dauer animals are markedly distinct from those in control animals.
Since the expression of only ~10% of predicted genes was altered in post-dauer animals (Figure 1B), any chromatin changes specific to this subset may be masked in the genome-wide histone modification maps. We, therefore, next examined the chromatin states associated specifically with this gene set in control and post-dauer animals. The 2,127 genes identified via transcriptional profiling were first separated into two groups based on whether expression was up- or down-regulated in post-dauer animals, and further binned based on expression levels. We found that the overall chromatin state changes were similar to those observed in the genome-wide experiments, such that global levels of both H3K4me3 and H4ac modifications were decreased in the highest expressed gene category in post-dauer animals (Figure 2F; Figure S5C–F; Table S5). This decrease was primarily observed for the up- but not down-regulated genes, presumably due to overall lower levels of these modifications in the down-regulated gene set (Figure 2F,G). Levels of both modifications were positively correlated with gene activity in both the up- and down-regulated gene subsets (Figure S5C–F; Table S5). No correlation was observed between the fold-change in expression between control and post-dauer animals and chromatin modifications (Figure S6), although this analysis does not preclude correlations at the level of individual genes. These results indicate that the gene subset whose expression is altered in post-dauer animals is subject to overall developmental history-dependent chromatin modification changes similar to those observed across the genome. Changes in the expression of individual genes may, therefore, arise from additional genetic or epigenetic mechanisms acting at the local level.
To begin to identify mechanisms that may play a role in the regulation of gene expression and reproduction in post-dauer animals, we next investigated whether mutations in chromatin remodeling genes affect post-dauer expression changes. We surveyed post-dauer gene expression changes of the highly down-regulated major sperm protein gene msp-64, and the highly up-regulated choline/carnitine O-acyltransferase gene W03F9.4, in mutants defective in different chromatin remodeling pathways (Figure 4). The selected mutants were viable, and did not exhibit defects in entry into, or exit from, the dauer stage. Mutations in many of the selected chromatin remodeling genes affected expression levels in control adult animals (Figure S7); we, therefore, determined whether the ratio of post-dauer to control expression was significantly different between wild-type and mutant animals. We found that while the expression changes of msp-64 and W03F9.4 in a subset of mutant post-dauer animals were similar to those in wild-type post-dauer animals, mutations in additional chromatin remodeling genes either decreased, abolished, or reversed the expected levels of up- or down-regulation in post-dauer animals (Figure 4A,B). Each mutation had different effects on msp-64 and W03F9.4 expression. Similarly, the post-dauer expression changes of three additional genes identified by transcriptional profiling were differentially affected in animals mutant for the zfp-1 chromatin-associated zinc finger protein gene [21, 22], although the post-dauer expression of the act-2 actin gene remained unaltered (Figure 4C). These observations imply that remodeling of chromatin architecture plays a causal role in the establishment or maintenance of the post-dauer expression changes, and that different mechanisms may affect the expression of different gene sets.
We next determined whether the altered gene expression and chromatin modification profiles were associated with altered phenotypes in post-dauer adult animals. Since the expression of a number of genes implicated in the regulation of C. elegans adult lifespan and reproduction was affected in post-dauer animals (Table S1–3), we quantified the adult lifespan and brood sizes of post-dauer and control animals. We found that the mean adult lifespan of post-dauer animals was significantly extended when compared to values in control animals (Figure 5A). Additionally, post-dauer animals produced more progeny than controls (Figure 5B). Discrepancies between these and previous observations [23, 24] are likely due to comparison of non age-matched populations of control and post-dauer animals, or induction of dauer entry through the use of non-physiological methods in previous reports [24, 25]. Similar to their effects on gene expression changes in post-dauer animals, mutations in a subset of chromatin remodeling genes abolished the brood size differences between control and post-dauer animals (Figure 5B). The embryonic lethality and adult survival rates were similar between the control and post-dauer populations in all genetic backgrounds (Figure S8). These results indicate that the developmental experience of spending one day in the dauer stage is sufficient to significantly alter physiology and life history traits in post-dauer animals, and further support the notion that changes in chromatin state may be causal to these phenotypic differences.
Brood sizes in the self-fertilizing C. elegans hermaphrodite are limited by the number of sperm produced; animals lay unfertilized oocytes once sperm are depleted . To determine whether increased spermatogenesis may account for the increased brood size in post-dauer animals, we quantified the timecourse of progeny production by control and post-dauer animals. Although the rate of egg-laying was unaltered between the two populations (Figure S9A), post-dauer animals produced more progeny on later days, suggesting that total sperm number may be increased in post-dauer animals (Figure S9B). Moreover, sperm from control males was similarly effective in competing with endogenous sperm for the production of progeny in both control and post-dauer animals (Figure S9C). These results indicate that under the examined conditions, the observed downregulation of sperm-enriched genes in post-dauer animals (Figure 1B) may be compensated for by other mechanisms, resulting in a larger brood size.
Our results indicate that C. elegans retains a cellular memory of its developmental history that is reflected in altered life history traits, gene expression and global chromatin state in post-dauer animals. These experience-dependent changes in chromatin modifications may serve as a global signature of the specific developmental experience of transient passage through the dauer stage. The chromatin state of dauer animals has also been proposed to be distinct , and may contribute to the observed changes in post-dauer animals. At a subset of loci, additional chromatin modifications, or other transcriptional mechanisms may act at the local level to establish or maintain expression changes in post-dauer animals. The complexity of these regulatory mechanisms is reflected in the different temporal patterns of gene expression changes in post-dauer animals, as well as the differential effects of chromatin remodeling genes on the expression changes of individual genes. We suggest that at other loci, these global modifications poise genes for further regulation by additional mechanisms upon subsequent exposure to specific external or internal cues. This is analogous to key regulatory genes in ES cells being associated with ‘bivalent’ marks of H3K4 and H3K27 methylation, allowing either gene activation or silencing upon lineage-specific differentiation [18, 27].
Alteration in gene expression, and consequent phenotypic differences among animals exposed to different environmental or development experiences creates phenotypic variation in a genetically identical population, and may provide critical evolutionary advantages. In rodents and humans, early experience shapes adult behaviors via alterations in DNA methylation state and histone modification profiles at specific loci [28–30]. Similarly, early mechanosensory stimulation has been shown to modulate adult behavior and gene expression in C. elegans . The establishment of C. elegans as a model system in which to explore the roles of early experience on adult phenotypes now allows for studies of the underlying genetic and epigenetic mechanisms at high resolution in an experimentally tractable organism.
We are grateful to A. Grishok, B. Bernstein, P. Kolasinska-Zwierz, and S. Young for advice and technical assistance, and the Caenorhabditis Genetics Center for providing strains. We thank A. Grishok, M. Marr, O. Hobert, C. Bargmann and J. Haber for comments on the manuscript, and the Sengupta lab for discussion. This work was supported by the NIH (RO1 GM56223 – P.S.; F32 GM083593 – S.E.H.), and an NSF IGERT training grant (DGE-0549390 – M.B.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.