Pervasive and dynamic transcription initiation in Saccharomyces cerevisiae

Abstract

Transcription initiation is finely regulated to ensure proper expression and function of genes. The regulated transcription initiation in response to various environmental stimuli in a classic model organism Saccharomyces cerevisiae has not been systematically investigated. In this study, we generated quantitative maps of transcription start sites (TSSs) at a single-nucleotide resolution for S. cerevisiae grown in nine different conditions using no-amplification nontagging Cap analysis of gene expression (nAnT-iCAGE) sequencing. We mapped ∼1 million well-supported TSSs, suggesting highly pervasive transcription initiation in the compact genome of the budding yeast. The comprehensive TSS maps allowed us to identify core promoters for ∼96% verified protein-coding genes. We corrected misannotation of translation start codon for 122 genes and suggested an alternative start codon for 57 genes. We found that 56% of yeast genes are controlled by multiple core promoters, and alternative core promoter usage by a gene is widespread in response to changing environments. Most core promoter shifts are coupled with altered gene expression, indicating that alternative core promoter usage might play an important role in controlling gene transcriptional activities. Based on their activities in responding to environmental cues, we divided core promoters into constitutive class (55%) and inducible class (45%). The two classes of core promoters display distinctive patterns in transcriptional abundance, chromatin structure, promoter shape, and sequence context. In summary, our study improved the annotation of the yeast genome and demonstrated a much more pervasive and dynamic nature of transcription initiation in yeast than previously recognized.

Figure 1.

Pervasive and dynamic transcription initiation in S. cerevisiae. (A) Correlations between numbers of examined growth conditions and identified CTSSs. The x-axis indicates the number of examined conditions. Each dot in the box plots represents the number of identified CTSSs based on a combination of TSS data from N numbers (ranging from 1 to 9) of growth conditions. (B) Distribution of mapped CAGE tags in different genomic regions. (C) Distribution of distance between CTSSs and annotated translation start codon (ATG). (D) Proportions of CTSSs identified in different numbers of growth conditions.

Figure 2.

TSS maps improve yeast genome annotation. (A–C) Examples of CAGE signal distribution in Category I, II, and III genes. In each example, the top track illustrates the distributions of CTSS signals near the annotated ORF. The middle track (green box) represents the core promoter region. The vertical line represents the dominant TSS in each core promoter. The bottom track displays the locations of annotated ORF. The originally annotated ATG is at the far side of the black box (ORF). Revised start codon (A,B) or alternative start codon (C) are indicated by “ATG” and an orange triangle. (D) Multiple sequence alignment for orthologous protein sequences of YAL059W (ECM1). Only the first 60 alignment sites are shown. The first 24 amino acids in the N terminus of YAL059W are absent in its orthologous sequences. The red arrow indicates the revised start codon. (E) A histogram shows the number of observations for detected peptides in ECM1. No peptides have been detected by previous mass spectrometry studies in the section between two orange lines, which corresponds to the 24 amino acids of the misannotated region in YAL059W.

Figure 3.

Prevalent core promoter shift responding to changing environments. (A) An example of core promoter shift (CIK1) between two growth conditions, YPD and arrest. (B) Experimental validation displays the presence and shift of two CIK1 transcript isoforms in response to changing environments. (C) Distribution of “degree of shift” (D_s) values in “H₂O₂” growth condition, using “YPD” as a control. (D) Volcano plot displays the correlations between D_s and −log₁₀ FDR values (χ² test). Each dot represents one gene. Dots in red represent genes with significant promoter shift (FDR < 0.05 and D_s < −1 or D_s > 1). (E) Scatterplot of enriched GO terms with significant core promoter shift (FDR < 0.05 and D_s < −1 or D_s > 1) under H₂O₂ condition. (F) The proportions of genes with core promoter shift that also experienced differential gene expression in response to environmental cues (FDR < 0.05, DESeq2). The percentages are indicated in each bar. (G) Scatterplot of enriched GO terms with significant core promoter shift and altered gene expression.

Figure 4.

Distinct properties of constitutive and inducible core promoters in S. cerevisiae. (A) Pie chart shows the fractions of core promoters that can be detected in different numbers of growth conditions. Numbers in the pie chart represent the numbers of growth conditions. (B) Core promoter compositions in single- and multi-core-promoter genes. (C) Transcript abundance of constitutive core promoters and inducible core promoters in all examined growth conditions. (D) Patterns of nucleosome occupancy around dominant TSSs of different types of core promoters. The nucleotide occupancy data were obtained in rich medium (YPD).

Figure 5.

Classifications of core promoter shape. (A) Examples of sharp, intermediate, and broad core promoters in S. cerevisiae. Core promoters with PSS greater than −10 were classified as sharp core promoters (SP), smaller than −20 as broad core promoters (BP), and the others as intermediate core promoters (IP). (B) Histogram shows the distribution of PSS values in S. cerevisiae. (C) Relationships between the transcript abundance and PSS values of core promoters. The dot plot was generated using a sliding window analysis after sorting all core promoters by transcript abundance (TPM). Each window has 200 core promoters with a moving step of 40 core promoters. Each dot presents the median values of PSS and TPM of each window. (D) A box plot of PSS values of inducible core promoters and constitutive core promoters.

Figure 6.

Initiator motif and dinucleotide preference of core promoters in S. cerevisiae. Sequence logo demonstrating a consensus sequence of 20 nt surrounding the dominant TSS of inducible core promoters (A) and constitutive core promoters (B), which likely represents the Initiator element in yeast. (C) TSSs in S. cerevisiae have a strong preference of pyrimidine–purine dinucleotide at [+1,−1] positions in both constitutive and inducible core promoters. The constitutive core promoter has a significantly higher frequency of dinucleotide with “A” and the +1 position (CA, and TA), whereas the inducible core promoter has a significantly higher frequency of “G” at the −1 position. (*) P < 0.01, χ² test.

Figure 7.

Predicted core promoter motifs in S. cerevisiae. (A) The top enriched motifs present in the inducible core promoter sequences. (B) The top enriched motifs present in the constitutive core promoter sequences. These promoter motifs were predicted by de novo discovery approach for the 150-bp sequence surrounding the dominant TSS (−100 and +50 nt) in each core promoter.