Fig 1.
Dynamics of chromatin accessibility and gene expression are correlated during metamorphosis.
(A) Open chromatin regions (peaks) in wings were identified by FAIRE-seq on time points prior to metamorphosis (L3) and during pupal stages 6 h, 18 h, 24 h, 36 h, and 44 h APF. Heatmaps of Pearson correlation coefficients for each replicate across this time course reveal differences between the proliferative and postmitotic stages (red dotted line). (B) Dynamic open chromatin peaks were organized into 18 k-means clusters, displayed as a heatmap representing the fraction of the maximum FAIRE rpkm value. (C, D) Most chromatin accessibility changes are associated with gene activation rather than repression during metamorphosis. (C) We assigned dynamic FAIRE peaks to the nearest expressed gene and correlated peak changes (opening or closing) with observed gene expression changes (increasing or decreasing) measured by RNA-seq at each subsequent time point. This revealed four classes of FAIRE peak/RNA expression correlations: opening/increasing consistent with gene activation, closing/decreasing consistent with loss of activation, opening/decreasing consistent with binding of a repressor, and closing/activation consistent with a loss of repression. We show the number of dynamic FAIRE peaks that fall into each quadrant for comparisons at time point 1 (t1) (y-axis) and time point 2 (t2) (x-axis). (D) Genes were clustered based on RNA expression patterns across metamorphosis. Two clusters showing a high positive correlation between RNA signal (average log2 fold change from L3) and accessibility of their assigned FAIRE peaks (average maximum FAIRE rpkm value) are shown. The full dataset correlating RNA expression with accessibility of their assigned FAIRE peaks for all clusters is provided in the Supporting information. The underlying data in Fig 1A–1D can be found within S1 Data. APF, after puparium formation; Cdk, cyclin-dependent kinase; FAIRE, formaldehyde-assisted isolation of regulatory elements; FAIRE-seq, FAIRE sequencing; RNA-seq, RNA sequencing; rpkm, reads per kilobase, per million mapped reads.
Fig 2.
Temporal regulation of the wing differentiation program and cell cycle changes.
(A) The length (in bp) of introns, 5′ UTRs, and 3′ UTRs (left) and genes (right) for all protein coding genes, wing terminal differentiation genes, and cell cycle genes is shown. The majority of FAIRE peaks occur within introns (S2 Fig). Most cell cycle genes have a compact structure with few, short introns, whereas differentiation genes contain large introns, providing potential dynamic regulatory elements. (B, E) Heatmap of gene expression for differentiation genes (B) and cell cycle genes (E), plotted by the percent of maximum RNA rpkm value. Both groups of genes show dynamic expression during metamorphosis. (C, F) Line plots of average FAIRE signal of the six stages for differentiation genes (C) and cell cycle genes (F). Differentiation genes show an increase in FAIRE peak accessibility at time points when gene expression is high: 6 h (p-value = 0.0004088), 36 h (p-value = 1.36 × 10−7), and 44 h (p-value = 1.408 × 10−12), compared with L3, Mann-Whitney U Test). Cell cycle genes show an increase in accessibility at time points when gene expression is repressed: 24 h (p-value = 0.0209), 36 h (p-value = 1.655 × 10−5), and 44 h (p-value = 0.005469), Mann-Whitney U Test. (D) A Gal4 reporter containing the indicated (blue bar) portion of the Cpr51A regulatory region drives UAS-GFP in late wings (44 h) when the regulatory elements are accessible. (G) Motif discovery was performed on open regions for cell cycle genes using MEME and compared with known motifs using TOMTOM. Potential regulatory elements for cell cycle genes are highly enriched for E2F binding motifs, DRE promoter sequences, and the Pol II pausing-associated motif 1. (H) A GFP reporter containing the indicated regulatory element (blue bar) for the simple cell cycle gene pcna is silent at the postmitotic stage of 44 h but can be reactivated postmitotically when E2F or E2F + CycD/Cdk4 is expressed. (I) stg, e2f1, and cycE are complex cell cycle genes with large regulatory regions. Blue shading indicates regions of known regulatory sequence that exhibit dynamic accessibility. Gal4 reporters containing the indicated portions (blue bars) of their regulatory regions drive UAS-degradable GFP to capture their regulatory functions in the pupal wing. Expression correlates with accessibility for these regions. p-values were determined by Mann-Whitney U Test; ****< 0.0001, **< 0.01, *< 0.05. The underlying data in Fig 2A, 2B and 2E can be found within S2 Data. Cdk, cyclin-dependent kinase; Cpr51A, Cuticular protein 51A; ctr, control; CycD, Cyclin D; CycE, Cyclin E; DRE, DNA replication-related element; E2F, E2F transcription factor; FAIRE, formaldehyde-assisted isolation of regulatory elements; GFP, green fluorescent protein; MEME, Multiple Em for Motif Elicitation; PCNA, proliferating cell nuclear antigen; Pol II, RNA polymerase II; RNA-seq, RNA sequencing; rpkm, reads per kilobase of transcript, per million mapped reads; stg, string; TOMTOM, Motif Comparison Tool; UAS, upstream activation sequence.
Fig 3.
Global impacts of cell cycle exit disruption on gene expression and open chromatin.
(A) G0 can be delayed to 36 h or bypassed beyond 50 h through short-term expression of E2F or E2F + CycD/Cdk4. Transgenes were overexpressed in the dorsal layer of wing epithelia under the control of Apterous-Gal4/Gal80TS from 12 h APF. Twenty-four-hour and 44-h wings were immunostained for ph3. (B, C) Genotype and scheme of RNA-seq and FAIRE-seq experiments to disrupt cell cycle exit during metamorphosis. (D, E) MA plots of RNA and FAIRE changes comparing bypassed exit (E2F + CycD/cdk4) between Ctr (D) and delayed cell cycle exit (E2F) (E) at 44 h. Abundant changes in expression of cell cycle genes, ribosome biogenesis, and cuticle formation genes are observed, while chromatin accessibility is nearly identical between conditions in which cells enter a delayed G0 versus continue cycling. The underlying data in Fig 3D and 3E can be found within S3 Data. APF, after puparium formation; Cdk, cyclin-dependent kinase; Ctr, control; CycD, Cyclin D; CyO, Curly O balancer; DP, dimerization partner; E2F, E2F transcription factor; FAIRE, formaldehyde-assisted isolation of regulatory elements; FAIRE-seq, FAIRE sequencing; Gal80TS, temperature-sensitive Gal80; GO, Gene Ontology; hsflp, heatshock-flippase; MA plot, scatter plot onto M (log ratio) and A (mean average) scales; ph3, phosphohistone H3; RNA-seq, RNA sequencing; TM6B, TM6B balancer; UAS, upstream activation sequence.
Fig 4.
Distal enhancer accessibility of complex cell cycle genes is developmentally controlled and independent of cell cycling status.
(A) Expression of cell cycle genes is increased when we delay or bypass cell cycle exit (log2 fold change for cell cycle genes versus ctrs expressing GFP). (B) Line plots of average FAIRE signal for cell cycle genes. Accessibility at most cell cycle genes’ TSS is slightly decreased when cell cycle exit is delayed (44 h E2F expression, p-value = 1.004 × 10−5, Mann-Whitney U Test). (C) Regulatory elements for simple cell cycle genes (orc6, pcna) remain accessible independent of cycling status. Complex cell cycle genes (cycE, stg) lose accessibility at regulatory regions independent of cycling status. (D) Expression of cycE and stg during metamorphosis (gray line, compared with 24-h wings) and genetic manipulations (colored dots, compared with 24-h ctr wings). stg possesses higher barrier for activation compared with CycE. Closed stg regulatory elements likely prevent stg expression in the late robust E2F-expressing wings, whereas E2F + CycD/Cdk4 can overcome this proliferation barrier. (E) Dynamic distal wing enhancers and TSS proximal regions for cycE and stg contain sites that bind E2F1, Su(H), and Yorkie based on ChIP in cycling tissues. The underlying data in Fig 4D can be found within S4 Data. Cdk, cyclin-dependent kinase; ctr, control; ChIP, chromatin immunoprecipitation; CycD, Cyclin D; CycE, Cyclin E; E2F, E2F transcription factor; FAIRE, formaldehyde-assisted isolation of regulatory elements; FC, fold change; GFP, green fluorescent protein; orc6, origin recognition complex subunit 6; pcna, proliferating cell nuclear antigen; stg, string; Su(H), Suppressor of Hairless; TSS, transcription start site; wt, wild type.
Fig 5.
Compromising cell cycle exit impacts chromatin accessibility and gene expression at a subset of wing terminal differentiation genes.
(A) log2 fold changes in RNA and (B) line plots of average FAIRE signal for genes involved in cuticle formation and differentiation. Preventing cell cycle exit reduces their expression and chromatin accessibility (E2F + CycD/Cdk4 at 44 h, p-value = 0.0046, Mann-Whitney U Test). (C) Selected cuticle protein genes exhibiting a failure to open potential regulatory elements at 44 h when cell cycle exit is delayed or bypassed. (D, E) Representative TEM (D) and chitin staining (E) of 64-h wings that delayed or bypass cell cycle exit in the dorsal wing epithelium using Apterous-Gal4/Gal80TS to activate E2F or E2F + CycD/Cdk4 expression during the final cell cycle. Extracellular matrix formation and chitin deposition are disrupted when cell cycle exit is compromised. The underlying data in Fig 5A can be found within S5 Data. Cdk, cyclin-dependent kinase; Cpr, Cuticular protein; Ctr, control; CycD, Cyclin D; E2F, E2F transcription factor; FAIRE, formaldehyde-assisted isolation of regulatory elements; FC, fold change; Gal80TS, temperature-sensitive Gal80; TEM, transmission electron microscopy.
Fig 6.
Bypassing cell cycle exit disrupts chromatin dynamics at ecdysone target genes and alters their expression.
(A) Scatterplot of ecdysone-responsive genes in 44-h wings under conditions that bypass cell cycle exit versus normal exit. Genes with significant changes in expression are labeled in red. (B, C) Chromatin regions of Blimp-1, Hr3, E74, and E75 fail to close or open at 44 h when cell cycle exit is compromised. (D) Blimp-1 antibody staining of wings at 36 h and 40–42-h wings with normal cell cycle exit (Ctr) or bypassed cell cycle exit in the posterior (using engrailed-Gal4/Gal80TS). Compromising cell cycle exit delays the activation of Blimp-1 in a compartment-autonomous manner. (E) Peaks that fail to open at 44 h from cuticle-development genes harbor high-scoring Blimp-1 binding sites. The underlying data in Fig 6A can be found within S6 Data. Cda5, Chitin deacetylase-like 5; Cpr, Cuticular protein; Ctr, control; E2F, E2F transcription factor; Hr3, Hormone receptor 3; Gal80TS, temperature-sensitive Gal80; RPKM, reads per kilobase of transcript, per million mapped reads; TEM, transmission electron microscopy; wt, wild type.
Fig 7.
A model for the developmental coordination of cell cycle exit and chromatin accessibility.
Potential regulatory elements at complex cell cycle genes such as stg become inaccessible in a developmentally controlled manner during robust G0. This may limit their activation in response to proliferative signals. Delaying or disrupting cell cycle exit impacts the subsequent opening of chromatin at genes in the wing terminal differentiation program that are potentially controlled via transcription factors downstream of ecdysone signaling. Cpr, Cuticular protein; cycE, cyclin E; E2F, E2F transcription factor; stg, string.