MYC regulates the non-coding transcriptome

Abstract

Using RNA-seq (RNA sequencing) of ribosome-depleted RNA, we have identified 1,273 lncRNAs (long non-coding RNAs) in P493-6 human B-cells. Of these, 534 are either up- or downregulated in response to MYC overexpression. An increase in MYC occupancy near their TSS (transcription start sites) was observed for MYC-responsive lncRNAs suggesting these are direct MYC targets. MYC binds to the same TSS across several cell lines, but the number of TSS bound depends on cellular MYC levels and increases with higher MYC concentrations. Despite this concordance in promoter binding, a majority of expressed lncRNAs are specific for one cell line, suggesting a determinant role of additional, possibly differentiation-specific factors in the activity of MYC-bound lncRNA promoters. A significant fraction of the lncRNA transcripts lack polyadenylation. The RNA-seq data were confirmed on eight selected lncRNAs by NRO (nuclear run-on) and RT-qPCR (quantitative reverse transcription PCR).

Figure 1. MYC regulates the expression of ncRNAs

(a) A heatmap of ncRNAs detected in P493-6 cells. P493-6 cells were analyzed by RNA-seq in triplicate under conditions of high and low MYC expression. Transcripts are ordered by the change in expression upon MYC upregulation with those most upregulated at the top and those most downregulated at the bottom. ${\log }_{2}F{C}_{i,\mathit{gene}}={\log }_{2}CP{M}_{i,\mathit{gene}}-\frac{1}{N}{\displaystyle {\sum }_{i=1}^N{\log }_2}CP{M}_{i,\mathit{gene}}$ where i refers to the sample (column), gene refers to the gene (row) and N refers to the total number of samples (6). (b) A volcano plot of observed fold changes in ncRNA expression vs significance.

Figure 2. Total and nuclear RNA content at high and low MYC levels

Total RNA was prepared using an RNA extraction robot from (a) ~2×10⁵ P493-6 cells with high or low expression of MYC, or (b) ~4×10⁶ purified nuclei in triplicate. Samples were eluted in a fixed volume of water and RNA concentration determined by Nanodrop. Values are mean + SEM, **P<0.01.

Figure 3. Binding of MYC to the promoters of lncRNAs in the presence of high (right) and low (left) levels of MYC

The MYC binding to the proximal promoters of the lncRNAs with detectable transcription is shown in red. Genes are ranked according to their level of expression.

Figure 4. MYC stimulates the transcription of distinct sets of lncRNAs in different cell lines

In the extreme right and left panels, lncRNA promoters are ranked according to MYC occupancy for P493-6 cells (high MYC levels) and MM.1S cells respectively. The central panel depicts an overlap of an enlarged segment from the side panels and shows that MYC promoter occupancy is virtually the same in both cell lines (purple color). However, the two cell lines differ starkly in the expression of the corresponding lncRNAs represented in the bar strips flanking the central panel (with blue signifying expression in P493-6 cells and red indicating expression in MM.1S cells).

Figure 5. Transcriptional activity from bidirectional promoters

MYC binds to bidirectional promoters and drives transcription. ncRNAs are frequently produced from bidirectional promoters. Shown here are the number of MYC-bound bidirectional promoters producing 0 (no transcription), 1 (unidirectional) or 2 (bidirectional) transcripts and whether MYC regulates the genes involved (red, blue and green colored are MYC-regulated, black and white are not MYC-regulated). Despite the symmetric nature of the E-box motif, MYC frequently acts asymmetrically at bidirectional promoters. Numbers indicate the number of bidirectional promoters acting in a specific way.

Figure 6. Validation data for two representative MYC-regulated lncRNAs

(a) MYC ChIP-seq data were visualized in the UCSC genome browser for the MIR17HG locus. The locations of miRNA precursors are indicated. (b) Expression of MIR17HG in high MYC and low MYC conditions was determined by RNA-seq, NRO and steady-state RT-qPCR. (c) Mature miRNAs from the miR-17-92 cluster were also affected by MYC levels. Addition of doxycycline (DOX) reduced miRNA levels. This effect of the repression increased over 72 hours. (d) MYC ChIP-seq data were visualized at the VPS9D1-AS1 locus. The VPS9D1-AS1 gene is antisense to, and resides completely within, its corresponding sense coding gene VPS9D1. Expression of (e) VPS9D1-AS1 and (f) VPS9D1 were determined by RNA-seq, NRO and RT-qPCR. This sense-antisense pair was discordantly regulated such that VPS9D1-AS1 was transcriptionally upregulated under high MYC expression, whereas VPS9D1 was correspondingly downregulated suggesting possible regulatory relationship. The change in VPS9D1 expression observed by RNA-seq and steady-state RT-qPCR was not recapitulated in the NRO data indicating that VPS9D1 is regulated at the post-transcriptional level. Values are mean +/− SEM, n=3, **P<0.01, ***P<0.001, ND not detected.