Effects of cycloheximide on the interpretation of ribosome profiling experiments in Schizosaccharomyces pombe

Abstract

Stress conditions lead to global and gene-specific changes in RNA translation. Ribosome profiling experiments have identified genome-wide alterations in the distribution of ribosomes along mRNAs. However, it is contentious whether these changes reflect real responses, or whether they are artefacts caused by the use of inhibitors of translation (notably cycloheximide). To address this issue we performed ribosome profiling with the fission yeast Schizosaccharomyces pombe under conditions of exponential growth (unstressed) and nitrogen starvation (nutritional stress), and both in the presence and absence of cycloheximide. We examined several aspects of the translational response, including density of ribosomal footprints on coding sequences, 5' leader ribosomal densities, distribution of ribosomes along coding sequences, and ribosome codon occupancies. Cycloheximide had minor effects on overall ribosome density, which affected mostly mRNAs encoding ribosomal proteins. Nitrogen starvation caused an accumulation of ribosomes on 5' leaders in both cycloheximide-treated and untreated cells. By contrast, stress-induced ribosome accumulation on the 5' side of coding sequences was cycloheximide-dependent. Finally, codon occupancy showed strong positive correlations in cycloheximide-treated and untreated cells. Our results demonstrate that cycloheximide does influence some of the results of ribosome profiling experiments, although it is not clear if this effect is always artefactual.

Figure 1

Effects of CHX on overall ribosome densities on coding sequences. (a) Scatter plots comparing mRNA levels (top) and ribosome densities (bottom) between untreated and CHX-treated cells. Data are presented for cells growing in the presence of a nitrogen source (+N) or starved for nitrogen (−N). Genes encoding ribosomal proteins are shown in green. All data have been normalized to RPKMs (Reads Per Kilobase per Million mapped reads). (b) Left: Boxplots comparing ribosomal densities between untreated and CHX-treated cells for groups of genes of the indicated average lengths of coding sequences. Right: similar data for indicated lengths of 5′ leader sequences. The red boxes display the behaviour of mRNAs encoding ribosomal proteins. (c) Comparison of changes in mRNA levels and ribosomal densities between cells starved for nitrogen (−N) and unstressed cells (+N). Data are presented for CHX-treated cells (left) and for untreated cells (left). Genes encoding ribosomal proteins are shown in green.

Figure 2

Effects of CHX on ribosome densities on 5′ leader sequences. (a) Experimental design: RPFs in coding sequences (CDS) and 5′ leader sequences are quantified in different experimental conditions. (b) Ratio of total reads mapping to 5′ leader sequences to total reads mapping to coding sequences (CDS) for unstressed cells (+N) and nitrogen starved cells (−N), and for cells treated or untreated with CHX (±CHX). The numbers indicate the fold-difference between pairs of −N and +N samples. Data are presented for two biological replicates. (c) Scatter plot comparing the ratio of reads mapping to 5′ leader sequences to reads mapping to coding sequences (CDS) for individual genes; each plot compares unstressed cells (+N) and nitrogen starved cells (−N). Data are presented for cells treated with CHX (left) or untreated (right). The red lines correspond to a ratio of 1. (d) Average values for the ratios presented in (c). The numbers indicate the fold-difference between pairs of −N and +N samples. Data are displayed for two biological replicates.

Figure 3

Effects of CHX on ribosome distribution across coding sequences. (a) Experimental design: RPFs on nucleotides 10 to 400 and on nucleotides 401–800 are quantified in different experimental conditions, and the ratio between both numbers is calculated. (b) Mean ratios calculated as described in A for all coding sequences, in the presence and absence of a nitrogen source (±N) and in the presence and absence of CHX treatment (±CHX). The numbers indicate the fold-difference between paired −N and +N samples. Data are presented for two biological replicates. (c) Scatter plot comparing the ratios obtained as defined in A for individual genes; each plot compares unstressed cells (+N) and nitrogen starved cells (−N). Data are presented for cells treated with CHX (left) or untreated (right). The red lines correspond to a ratio of 1. (d) Metagene displaying average distributions of RPFs along coding sequences in four experimental conditions. A running window of 60 nucleotides was used to smoothen the plotted lines.

Figure 4

Effects of CHX on relative codon occupancies. Scatter plots displaying relative codon occupancies obtained as described in Methods. Each dot corresponds to a single codon. Termination codons are not displayed. The positions of rare codons CCG and CGG are indicated. The dotted lines correspond to 1.5-fold differences. The Pearson correlations between datasets are indicated. (a) Comparison of the effects of CHX treatment in nitrogen-starved cells. (b) As in (a), for cells grown with a nitrogen source. (c) Comparison of the effects of nitrogen starvation in the presence of CHX. (d) As in (c), in the absence of CHX.