The little elongation complex functions at initiation and elongation phases of snRNA gene transcription

Abstract

The small nuclear RNA (snRNA) genes have been widely used as a model system for understanding transcriptional regulation due to the unique aspects of their promoter structure, selectivity for either RNA polymerase (Pol) II or III, and because of their unique mechanism of termination that is tightly linked with the promoter. Recently, we identified the little elongation complex (LEC) in Drosophila that is required for the expression of Pol II-transcribed snRNA genes. Here, using Drosophila and mammalian systems, we provide genetic and molecular evidence that LEC functions in at least two phases of snRNA transcription: an initiation step requiring the ICE1 subunit, and an elongation step requiring ELL.

Figure 1. Functional conservation of the LEC complex in human cells

(A-C) The human LEC complex is enriched at the promoters of the snRNA genes. Genome browser track examples for the occupancy of human LEC subunits ICE1, ICE2, ELL, and ZC3H8 as well as Pol II in HCT116 cells at the U2 (RNU2), U11 (RNU11), and U12 (RNU12) snRNA genes. Nearby protein-coding genes such as STX5, SNHG1 and TAF12 have high levels of Pol II, but no LEC. (D) Scatter plot depicting the genome-wide occupancy levels of Pol II vs. ICE1, showing that the highest co-bound genes are snRNA genes (blue color). Pearson correlations are shown. (E-G) Scatter plots depicting the genome-wide occupancy levels of ELL (E), ICE2 (F), and ZC3H8 (G) versus ICE1. Occupancy levels were measured as the sum of reads per million (RPM) aligned within 10 bp of all ENSEMBL 69-annotated transcription start sites. (H-J) RNA-seq analysis shows that ICE1 regulates snRNA gene expression. (H-I) Genome browser tracks show examples of RNA-seq data surrounding the RNU11 and RNU12 genes. The expression of RNU11 and RNU12, but not nearby protein-coding genes, is decreased in ICE1 shRNA-treated samples. (J) The left panel shows a MA plot of the genome-wide expression changes as determined by RNA-seq for ICE1 shRNA. The x-axis shows the log10 normalized read abundance (baseMean) of knockdown and non-targeting samples as reported by DESeq (Anders and Huber, 2010). The y-axis shows the log2 fold changes of normalized read abundance of ICE1 knockdown (KD) divided by the wild-type (WT) condition. Annotated snRNA genes are shown in blue. Differentially expressed snRNA genes (FDR<0.1) as reported by DESeq are circled in red. The right panel shows a boxplot of the log2 fold change ICE1 KD divided by WT for expressed snRNA genes (blue box) and all expressed protein coding genes (gray box). A baseMean of 10 was chosen as the cutoff for determining whether a gene is expressed. Box plot colored portions indicate the middle quartiles; the whiskers indicate a maximum of 1.5 of the interquartile range; middle notches indicate a 95% confidence interval of the median. See Figure S1 for related information.

Figure 2. ICE1 is a scaffolding protein for LEC and is required for the targeting of LEC to subnuclear bodies

(A) The recruitment of ICE1 to coilin-positive subnuclear bodies requires its N-Terminal 500 amino acids, which includes a 163 amino acid coiled-coil domain. Different ICE1 truncations were expressed with an N-terminal GFP tag in HeLa cells and the localization of these ICE1 truncations was visualized by fluorescence microscopy. (B-E) Images (B-D) and their quantitation (E) show that HeLa cells depleted of ICE1 by two different shRNA are defective in the localization of ICE2 (B), ELL (C), and ZC3H8 (D) to subnuclear bodies. For (E), more than 100 GFP-positive cells were scored for whether bright nuclear dots of ICE2, ELL, or ZC3H8 were present, indicative of coilin body localization. Error bars represent the standard deviations. (F) Proposed molecular architecture of the LEC complex in human cells. The LEC complex might dimerize through the N-terminus of ICE1 (data not shown). ICE2 and ZC3H8 associate with ICE1 through its C-terminal portion while ELL binds to the N-terminal region (see Figure S2 for related information).

Figure 3. Overexpression of ICE1 in larval salivary glands re-localizes ELL and Pol II on polytene chromosomes

(A) Overlay of ICE1 (green) immunostaining and the phase contrast image in ICE1 overexpressing salivary glands. Many of the sites of ICE1 staining correspond to snRNA genes as annotated in Flybase (right panel). A few of the mapped sites correspond to genes annotated as snoRNAs. These particular snoRNAs are LEC targets, as they were previously reported to have high levels of LEC in S2 cells and their levels are reduced after RNAi of ICE1 and ELL in S2 cells (Smith et al., 2011). (B) In ICE1 overexpressing glands, some of the strongest sites of ELL co-localize with the ectopically expressed ICE1. (C) ICE1 overexpression leads to strong occupancy of Pol II at the ectopic sites of ICE1 when visualized with the 8WG16 monoclonal recognizing unmodified Pol II. Phase contrast images for B-C are shown in the upper left panels. See Figure S3 for related information.

Figure 4. ICE1 specifically regulates the occupancy of Pol II and LEC subunits at the snRNA genes

(A) Genome browser track examples of Pol II occupancy in non-targeting (Non T) and ICE1-depleted Drosophila S2 cells. Pol II occupancy is specifically down-regulated at U11 and U12 snRNA genes, but does not change at their neighboring protein-coding genes, such as Baldspot, Sc2, and Abl. (B) The left panel shows a MA plot of the genome-wide Pol II occupancy after ICE1 depletion in Drosophila S2 cells. The x-axis shows the log2 geometric average of Pol II occupancy as measured as the maximum read sums per million sequenced (RPM) within the TSS and 100 bp downstream of the TSS for all annotated transcription start sites. Only Pol II-bound transcripts are shown that had enriched Pol II regions (MACS p < 0.05) in the wild-type sample within 100 bp of an annotated start site. The y-axis shows the log2 fold change of the Pol II occupancy levels measured by taking the occupancy levels in the knockdown divided by the non-targeting control. Pol II-bound snRNA genes are shown in blue. Some transcriptionally silent snRNA genes are called bound by Pol II using our peak finding criteria due to their proximity to other transcribed genes, so we defined a high confidence LEC target set for start sites that showed a two-fold or greater loss of ICE1 occupancy with 10 bp of the transcription start site in the ICE1 RNAi condition compared to wild-type ICE1 ChIP-seq levels. These genes are highlighted with a red circle. The right panel is a box plot analysis of the log2 fold change in Pol II occupancy after ICE1 RNAi. (C) Genome browser track examples of Pol II occupancy in non-targeting and ICE1-depleted human HCT116 cells. Pol II occupancy is specifically reduced at RNU11 and RNU12 snRNA genes, but does not change at their neighboring protein-coding genes. (D) The left panel shows a MA plot of the genome-wide Pol II occupancy after ICE1 depletion in HCT116 cells, as plotted in Panel B. LEC-bound snRNAs are highlighted with red circles. The snRNA genes with LEC bound in the wild-type condition show a general loss of Pol II after RNAi. The right panel is a box plot representation of the log2 fold change in Pol II after ICE1 knockdown. (E-F) ICE1 is required for recruitment of LEC subunits ICE2, ELL, and ZC3H8 to the RNU11 and RNU12 snRNA genes. ChIP-qPCR was performed with ICE1, ICE2, ELL, ZC3H8, ELL, and Pol II antibodies in control and ICE1-depleted HCT116 cells. Error bars represent the standard deviations. See Figure S4 for related information.

Figure 5. ELL knockdown affects the Pol II distribution across snRNA genes

(A) Genome browser track examples of Pol II occupancy at snRNA and neighboring genes in non-targeting (Non T) and ELL-depleted Drosophila S2 cells. Pol II occupancy at snRNA genes is less affected after ELL knockdown compared to ICE1 depletion (compare to Figure 5A and Figures 4A and S4A), even though ELL itself is greatly reduced at the snRNA genes. (B) Overlay and expanded view of the NonT and ELL knockdown Pol II occupancies at U11 (top panels) and U12 (bottom panels). The peak Pol II signal in the ELL knockdown is shifted in the direction of the transcription start site (TSS) for each gene. Left panels compare the Pol II signals in NonT and knockdown using the same fixed scales. The right panels are scaled to the maximum peak signal of each condition to allow better visualization of the shift in the overall distribution of Pol II after ELL knockdown. (C) Track examples showing a shift in Pol II towards the TSS of two snoRNA genes that are regulated by LEC (Smith et al., 2011). Fixed and percent-of-maximum versions are plotted as in (B). (D) Quantitation of the degree of shift as seen by two distinct measurements. The left panel shows a box plot representation of the distance, in base pairs, from the Pol II peak summit in the wild-type condition to the Pol II peak summit in the ELL RNAi knockdown. The box plot in the right panel instead uses the distance from the center of read density of the Pol II enriched region at each gene. All genes are oriented 5’ to 3’ such that negative values indicate a shift toward the 5’ end of the gene. The set of LEC-bound snRNAs and all Pol II-bound genes are the same as in Figure 4B. See Supplemental Figure S5 for related information.

Figure 6. Human LEC regulates RPPH1 gene expression

(A) Human LEC subunits are present with Pol II at the RPPH1 gene, a gene that is transcribed by both Pol II and Pol III (James Faresse et al., 2012). Genome browser tracks show the occupancy of human LEC subunits ICE1, ICE2, ELL, and ZC3H8 as well as Pol II at the RPPH1 gene in HCT116 cells. Pol II and Pol III ChIP-seq tracks, shown in light blue, are from (James Faresse et al., 2012). (B) ICE1 regulates RPPH1 gene expression. Genome browser track of RPPH1 gene expression determined by RNA-seq. (C) ICE1 is required for Pol II occupancy at the RPPH1 gene. Genome browser profiles of Pol II occupancy at the RPPH1 gene in non-targeting and ICE1-depleted HCT116 cells are shown. (D) ICE1 is not required for the recruitment of Pol III to the RPPH1 genes. Pol II, but not Pol III, is reduced at the RPPH1 promoter after ICE1 RNAi. Another Pol III transcribed non-coding RNA gene, RNY1, with little or no associated Pol II, is shown for comparison. Error bars represent the standard deviations. See Figure S6 for related information.

Figure 7

LEC has roles in establishing Pol II occupancy and productive transcription of Pol II at snRNA genes. (A) A typical snRNA gene is composed of a distal sequence element (DSE), a core promoter element (PSE), the transcription unit, and a 3’ box which signals for 3’ RNA processing. The PSE is bound by the small nuclear RNA-activating protein complex (SNAPc), which is recruited to snRNA promoters in early G1 of the cell cycle before Pol II is recruited (James Faresse et al., 2012). LEC, represented as a multi-tool pocketknife, is composed of the scaffolding protein ICE1, the transcription elongation factor ELL, and proteins of unknown function, ICE2 and ZC3H8. How LEC specifically recognizes Pol II transcribed snRNA genes is currently unknown (represented by a dotted blue line encircling a question mark). (B) Knockdown of ICE1 (dashed line) leads not only to the release of LEC components from snRNA genes, but also to the loss of Pol II occupancy, leading to reduced snRNA transcription. (C) Knockdown of ELL (dashed line) leaves LEC largely intact and has a minimal effect on Pol II levels. Instead, knockdown of ELL results in an overall Pol II redistribution towards the 5’ end of the snRNA gene, suggestive of a defect in transcription elongation. Although the role of ICE2 and ZC3H8 are currently unknown, their presence with other LEC components in coilin-stained subnuclear bodies, suggests that LEC could regulate other post-transcriptional steps, such as snRNP maturation or recycling.