Report

RNA Pol II Accumulates at Promoters of Growth Genes During Developmental Arrest

See allHide authors and affiliations

Science  03 Apr 2009:
Vol. 324, Issue 5923, pp. 92-94
DOI: 10.1126/science.1169628

Abstract

When Caenorhabditis elegans larvae hatch from the egg case in the absence of food, their development is arrested (L1 arrest), and they show increased stress resistance until food becomes available. To study nutritional control of larval development, we analyzed growth and gene expression profiles during L1 arrest and recovery. Larvae that were fed responded relatively slowly to starvation compared with the rapid response of arrested larvae to feeding. Chromatin immunoprecipitation of RNA polymerase II (Pol II) followed by deep sequencing showed that during L1 arrest, Pol II continued transcribing starvation-response genes, but the enzyme accumulated on the promoters of growth and development genes. In response to feeding, promoter accumulation decreased, and elongation and messenger RNA levels increased. Therefore, accumulation of Pol II at promoters anticipates nutritionally controlled gene expression during C. elegans development.

Animals cope with fluctuating nutrient availability by altering resource allocation between growth and survival. For the free-living nematode Caenorhabditis elegans, life in the wild is characterized by feast or famine, so that organismal fitness presumably depends on developmental responses to nutrient availability. C. elegans L1 arrest is an acute response to starvation, in which development is suspended and environmental stress resistance is increased without morphological modification (1). L1 arrest offers an opportunity to identify regulatory mechanisms mediating nutritional control of gene expression and, therefore, resource allocation, in a developmental system.

We measured growth and gene expression in C. elegans larvae as they hatched from mature eggs and were cultured in the presence or absence of E. coli as food (Fig. 1A) (2). We also switched conditions (fed or starved) 12 hours after hatching to further dissociate the effects of development and nutrition. Using a flow cytometer (COPAS) to measure growth (2), we found that larvae hatched in the presence of food increased in mass within an hour (Fig. 1B). Arrested L1s also responded rapidly to feeding, increasing in mass within 2 hours and achieving the normal L1 growth rate within 3 hours.

Fig. 1.

(A) A schematic of the experimental design for growth and expression analysis. Time points sampled are indicated below, where 0 hours is precisely mid-hatch (fig. S1). Crossed arrows indicate when starved cultures were fed and when fed cultures were starved. (B) Optical extinction (arbitrary units) was measured as a proxy for mass for synchronous populations of about 5000 individuals at each time point, and the median is reported.

In three independent experiments, we collected samples for gene expression analysis using a high-density oligonucleotide array (table S1). During L1 arrest, multiple energy homeostasis signaling pathways and two known regulators of metabolic gene expression (fig. S3) (3, 4) were up-regulated (fig. S2). Numerous metabolic genes were modulated as expected (fig. S4) (5), and reporter gene analysis further validated our results (figs. S5 and S6, table S2).

Using a two-factor analysis of variance (ANOVA) to isolate the effects of nutrition and time, we found that expression of more genes was affected by nutrient availability than by development (fig. S7). At the smallest P value (10–16), 20.6% (2386 genes) of the transcriptome was affected by time and 26.9% (3117 genes) by nutrition, which demonstrated the profound influence of energy homeostasis on gene regulation.

We used principal components analysis (PCA) to visualize the dynamics of the transcriptome. This PCA projects multidimensional gene expression data into the two orthogonal factors that capture the most variance; transcriptome-wide dynamics of gene expression can be visualized by the relative positions of data points in the resulting “phase plane.” The clustering of starvation data points obtained after >6 hours indicates that the gene expression response to starvation was largely established within 6 hours of hatching in the absence of food (Fig. 2A), a conclusion also supported by pairwise t tests (fig. S8). In contrast, gene expression continued to change in the fed, developing larvae as indicated by the lack of clustering among data points for fed larvae. Notably, arrested L1s responded to feeding more rapidly than fed L1s responded to starvation. The dashed lines in Fig. 2A plot the transcriptome-wide effects of switching conditions. Feeding arrested L1s caused the transcriptome to change so much in 3 hours that it switched sides of the PCA graph, becoming more similar to that of fed than starved L1s (Fig. 2A). In contrast, starving fed L1s had less effect on the transcriptome; it remained more similar to fed larvae than starved. Moreover, pairwise t tests showed that feeding starved L1s for 3 hours caused 381 genes to change (P ≤ 10–3) so that only 47 differed between them and L1s hatched in the presence of food 3 hours earlier, but starving fed L1s for 3 hours caused only 56 genes to change, leaving 544 that differed between them and L1s hatched in the absence of food 3 hours earlier (fig. S9). We infer that arrested L1s are primed for rapid response to food.

Fig. 2.

(A) Principal components analysis reveals the dynamics of gene expression in response to starvation and feeding. Note the large and rapid response of arrested L1s to food during recovery (dashed green line). (B) The 3117 genes most significantly affected by nutrition (fig. S7) were Z transformed and hierarchically clustered by pairwise correlation. Yellow indicates relatively high levels of normalized expression; blue indicates low levels.

We used cluster analysis to visualize the diversity and relative proportion of nutritionally regulated gene expression patterns. The genes divided cleanly into two clusters that were up-regulated during either L1 arrest or larval development (Fig. 2B). The relatively complex expression dynamics of larval development compared with those of L1 arrest are evident in the clustered expression patterns. Cluster analysis was also consistent with the observed asymmetric response to switching conditions. The expression profile of larvae fed for 12 hours and then starved for 3 hours resembled larvae fed for 15 hours rather than those starved for 3 hours. Conversely, the expression profile of larvae starved for 12 hours and then fed for 3 hours did not resemble larvae starved for 15 hours but rather larvae fed for 3 hours.

Expression analysis indicated that the response to starvation is complete within about 6 hours and that the transcriptome is in steady state thereafter (Fig. 2), possibly as a result of transcriptional quiescence during arrest. We performed chromatin immunoprecipitation of RNA Pol II followed by deep sequencing (ChIP-Seq) (6, 7) to map the genomic pattern of Pol II binding. Contrary to our hypothesis, Pol II was enriched on thousands of genes during L1 arrest (Fig. 3A). Consistent with this, destabilized reporter genes for 16 transcription factors up-regulated during L1 arrest were transcribed for at least the first few days of L1 arrest, whereas signal from transient heat-shock induction of the reporter was undetectable after 12 hours (fig. S6). ChIP-Seq showed that the genome-wide pattern of transcription did not change between 6 and 12 hours of L1 arrest (Fig. 3A), consistent with the transcriptome's maintenance of the steady state.

Fig. 3.

(A) Fold enrichment of Pol II relative to input (preimmunoprecipitation) over each gene in the genome is plotted for 6 and 12 hours of L1 arrest. (B) The amount of elongating Pol II per gene changes dramatically within 1 hour of feeding after 12 hours of L1 arrest. The plot is the same as in (A), except genes that increase or decrease in Pol II fold enrichment by at least two units are colored green and red, respectively.

The amount of elongating Pol II per gene changed rapidly in response to feeding (Fig. 3B). After just 1 hour of recovery from L1 arrest, Pol II enrichment over the coding region increased on 387 genes and decreased on 183 genes. Genes with increased elongation included numerous constituents of the ribosome and translational regulators (table S3). Genes with decreased elongation in response to feeding were enriched for effectors of aging and long-chain fatty acid metabolism (table S3), which indicated that they overlap with clusters of genes up-regulated in response to starvation (table S4). Thus, RNA Pol II ChIP-Seq from whole worms provides effective measurement of instantaneous transcriptional activity.

Genome-wide analyses of Pol II binding reveal that Pol II is concentrated in the promoters of many human, Drosophila, and yeast genes (812). We hypothesized that Pol II is recruited to the promoters of growth and development genes during L1 arrest. We performed ChIP-Seq, using three different antibodies against the C-terminal domain of Pol II, using animals in the arrested L1 state for 12 hours and after 1 hour of recovery by feeding (figs. S10 and S11). We defined a 5′ bias index for Pol II binding as the ratio of Pol II enrichment in a 200–base pair (bp) window spanning the transcription start site to enrichment over the entire coding region. To focus on genes with the most reliable Pol II distribution, we included only genes of >300 bp with Pol II enriched at least threefold over the coding region (1238 genes). We consistently saw an overall decrease in 5′ bias after 1 hour of recovery (Fig. 4A). On average, 5′ bias decreased by about 20%, and for the 178 genes (14%) with 5′ bias >2, average bias decreased by 37%. 8WG16 is the most specific of the three antibodies for nonphosphorylated Pol II (13), and it detected the largest difference in 5′ bias between L1 arrest and recovery.

Fig. 4.

(A) Histograms of log2 5′ bias are plotted for three different Pol II antibodies after L1 arrest for 12 hours (red) and 1 hour of recovery (green). In the S2 graph, L1 arrest after 6 hours is plotted as a red dotted line. Biological replicates are plotted in the 4H8 graph. The vertical dashed line on each graph marks a 5′ bias of 2. P values resulted from a paired-sample t test for the difference between L1 arrest and recovery for genes with 5′ bias >2 during arrest. (B) Fold-enrichment of Pol II over the coding region during L1 arrest and after 1 hour of recovery is plotted for each ChIP. Starting with 1238 genes plotted in (A), 178 genes (14.4%) with average 5′ bias during L1 arrest of at least 2 are plotted in the left box plot, and the remaining 1060 are plotted in the right box plot. The box plots show the median and two quartiles of each distribution, with bars reflecting 5th and 95th percentiles and a red bar indicating the average. (C) Normalized read counts for each condition are plotted over the hsp-3 gene. (D) The three most significantly enriched gene ontology (GO) terms and associated hypergeometric P values are shown for 178 genes with 5′ bias during L1 arrest. (E) Histograms of Pol II-binding peak positions relative to the transcription start site are plotted for each L1 arrest ChIP. Bins of 40 bp and strand-corrected peak locations were used. (F) Average of log2 mean normalized transcript abundance is plotted for two sets of genes: the set of 1238 plotted in (A) are in black, and a subset of them (178 genes) with 5′ bias greater than 2 are in red. (Left) Expression during L1 arrest and recovery; (right) expression during L1 growth. The bars on the black curves represent the standard deviations of 1000 averages computed for a set of 178 genes randomly selected from the 1238 genes.

In spite of cell-type heterogeneity, RNA Pol II accumulation was readily detected from whole worms, which indicated its prevalence. Notably, 5′ bias was detected with each antibody in spite of the different specificities of the antibodies (Fig. 4A). Promoter accumulation of Pol II has been detected in C. elegans embryos with the use of antibody 8WG16 (14), but the S2 antibody was raised against a Ser2 phosphorylated polypeptide from the C-terminal domain of Pol II, which is correlated with elongation in other systems (1517). However, we found that S2 also bound nonphosphorylated Pol II in C. elegans (fig. S10).

The subset of genes with 5′ biased Pol II was particularly responsive to feeding. Pol II enrichment over the entire coding region, which reflects processive elongation, increased substantially during recovery for this set of genes (Fig. 4B). The average increase in Pol II enrichment per gene was 67% (P < 10–5; paired-sample t test), with S2 showing the largest increase (110%) and 8WG16 the smallest (36%). The heat shock protein hsp70 homolog hsp-3 illustrates nutritional control of promoter accumulation (Fig. 4C). Finally, genes affecting development and positively regulating growth rate were most significantly enriched among those with 5′ bias (Fig. 4D, and tables S3 and S5).

There are multiple potential points of regulation between Pol II recruitment and uninhibited elongation, and there is unlikely to be a single mechanism underlying Pol II promoter accumulation during L1 arrest. Nevertheless, we speculate that in many of the observed instances, Pol II had initiated elongation and was paused proximally. About half of the 178 genes with 5′ bias had peak Pol II positions just inside the transcription start site (Fig. 4E), consistent with pausing, although the distribution was not as tight as with bona fide pausing (18). Including all 1209 genes with 5′ bias >3, rather than only those with Pol II enriched over the coding region (and therefore expressed), accentuated the bimodality in peak Pol II positions, with the majority of peaks upstream of the start site (fig. S12). The negative elongation factor (NELF) complex promotes pausing in D. melanogaster (10, 1820), but NELF homologs have not been identified in C. elegans, which suggests that the postrecruitment accumulation we observe is mechanistically distinct (21).

Accumulation of Pol II at promoters anticipates nutritionally induced gene expression. We examined mRNA expression during L1 arrest, recovery, and growth for genes with and without 5′ biased Pol II (Fig. 4F). Consistent with pausing not necessarily being repressive (20, 21), we saw that genes with Pol II accumulation were expressed during L1 arrest. However, their average mRNA expression increased significantly during recovery, demonstrating an anticipatory role of promoter accumulation. Genes with Pol II accumulation during L1 arrest were also up-regulated during L1 growth after hatching in the presence of food (Fig. 4F), which underscored their characterization as larval growth and development genes. These results support our Pol II ChIP-Seq results and demonstrate that Pol II promoter accumulation is nutritionally controlled and that a decrease in accumulation is correlated with an increase in elongation and mRNA expression.

This work demonstrates the profound influence of energy homeostasis on gene regulation during development. Analyses of growth, gene expression, and RNA Pol II binding indicate that C. elegans responds rapidly to feeding during recovery from developmental arrest. We suggest that this rapid response is facilitated by RNA Pol II accumulation on growth and development genes, poising them for expression. Nutritional control of Pol II promoter accumulation may be a general feature of gene regulation involving energy homeostasis.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1169628/DC1

Materials and Methods

SOM Text

Figs. S1 to S14

Tables S1 to S7

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article