Absolute quantification of translational regulation and burden using combined sequencing approaches

Abstract

Translation of mRNAs into proteins is a key cellular process. Ribosome binding sites and stop codons provide signals to initiate and terminate translation, while stable secondary mRNA structures can induce translational recoding events. Fluorescent proteins are commonly used to characterize such elements but require the modification of a part's natural context and allow only a few parameters to be monitored concurrently. Here, we combine Ribo-seq with quantitative RNA-seq to measure at nucleotide resolution and in absolute units the performance of elements controlling transcriptional and translational processes during protein synthesis. We simultaneously measure 779 translation initiation rates and 750 translation termination efficiencies across the Escherichia coli transcriptome, in addition to translational frameshifting induced at a stable RNA pseudoknot structure. By analyzing the transcriptional and translational response, we discover that sequestered ribosomes at the pseudoknot contribute to a σ³²-mediated stress response, codon-specific pausing, and a drop in translation initiation rates across the cell. Our work demonstrates the power of integrating global approaches toward a comprehensive and quantitative understanding of gene regulation and burden in living cells.

Keywords: RNA‐seq; Ribo‐seq; genetic circuits; transcription; translation.

Figure 1. Overview of the workflow

1. Major steps involved when quantifying transcription (RNA‐seq) and translation (Ribo‐seq) and the additional cellular features measured. Elements required for quantification in absolute units are highlighted in red.

2. Model for calculating the translation initiation rate of a ribosome binding site, see equation (2).

3. Model for calculating translation termination efficiency of a stop codon, see equation (3). Star denotes the location of the stop codon.

4. Model for calculating translational frameshifting efficiency between two coding regions “A” and “B” in zero and −1 reading frames, respectively, see equation (4).

Figure 2. Measuring translation initiation and translation termination signals across the E. coli transcriptome

1. Genetic design of the LacZ reporter construct whose expression is activated by the inducer IPTG.

2. Normalized RPF count profile averaged for all E. coli transcripts. Profiles generated for cells grown in the absence and presence of IPTG (1 mM). Start and stop codons are shaded.

3. Bar chart of all measured RBS initiation rates ranked by their strength. Strong RBSs with initiation rates > 1 ribosome/s are highlighted in red.

4. Bar chart of all measured translation termination efficiencies at stop codons ranked by their strength. Stop codons with translation termination efficiency > 0.99 are highlighted in red.

5. Distribution of initiation rates for cells grown in the absence and presence of IPTG (1 mM).

6. Distribution of translation termination efficiencies for cells grown in the absence and presence of IPTG (1 mM).

Figure 3. Simultaneous quantification of transcription and translation of endogenous genes and a synthetic genetic construct

1. Comparison of protein synthesis rate of endogenous E. coli genes measured using Ribo‐seq from this study (in molecules/s units) and from that by Li et al (2014) (in molecules/generation units). Each point corresponds to a single gene, and color denotes the ratio of transcription initiation rate to translation initiation rate (giving RNAP/ribosome) capturing whether transcription (light yellow) or translation (dark blue) is more dominant.

2. Transcription (bottom) and translation (top) profiles for uspA, ompA, and gapA, computed from the RNA‐seq and Ribo‐seq data without induction. Positions of the genetic parts and gene are shown below the profiles.

3. Promoter strengths in RNAP/s units and RBS initiation rates in ribosome/s units.

4. Transcription (bottom) and translation (top) profiles for lacZ. Profiles are shown for cells in the absence and presence of IPTG (1 mM). Position of genetic parts and gene is shown below the profiles. RBS is omitted from the genetic design due to its size.

5. Measured promoter strength in RNAP/s units, RBS initiation rate in ribosomes/s units, and the transcriptional terminator and translation termination efficiency for lacZ. Data shown for cells in the absence and presence of IPTG (1 mM).

Figure 4. Characterization of a synthetic pseudoknot construct that induces translational frameshifting

1. Genetic design of the PK‐LacZ construct. Expanded sequence shows the PK secondary structure with the slippery site underlined, as well as the two genes (gene10 and lacZ) in differing reading frames.

2. Translation profiles for the PK‐LacZ construct in cells cultured in the absence (bottom) and presence (top) of IPTG (1 mM). The gene10, middle, and lacZ regions are labeled above the profiles. Shaded region denotes the PK, and dashed lines denote the start codon and stop codons of gene10 and LacZ.

3. Fractions of the total RPFs and mRNA reads in each reading frame for the gene10, PK or middle, and lacZ regions. Data shown separately for cells cultured in the absence and presence of IPTG (1 mM).

4. Violin plots of the distributions of fractions of total RPFs and mRNA reads in each reading frame for all E. coli transcripts. Median values shown by horizontal bars. Data from two biological replicates. *P = 0.049; **P = 1.6 × 10⁻⁹ (Mann–Whitney U test).

Figure 5. Cellular response to the expression of a synthetic pseudoknot construct

1. Change in expression of chromosomal genes in E. coli cells following induction of PK‐lacZ expression (1 mM IPTG). Each point represents a transcript. Differentially expressed genes (mRNA count: P < 0.001 and absolute log₂ fold‐change > 1.37; translation efficiency: P < 0.01) are highlighted in color and by an alternative point shape (transcriptional regulation: purple cross; translational regulation: orange open circle).

2. Venn diagram of genes significantly regulated transcriptionally and translationally after induction of the PK‐LacZ construct. Colors match those in panel (A).

3. Change in codon occupancy for cells harboring the PK‐LacZ construct after induction by IPTG (1 mM) calculated from the Ribo‐seq data. Each point corresponds to a codon, which are ordered by amino acid identity and then by abundance in the genome (left most abundant, right least abundant). Dashed horizontal line denotes no change. Outliers are labeled and highlighted in red (Tukey test: 1.5 times the interquartile range below the first quartile or above the third quartile).

4. Translation initiation rates for all E. coli RBSs in cells harboring the LacZ and PK‐LacZ constructs in the absence and presence of IPTG (1 mM). Solid line shows the same initiation rate for both conditions. Dotted lines denote linear regressions for the data with no offset.

5. Fractions of mRNA reads and RPFs mapping to each synthetic expression construct (LacZ and PK‐LacZ) and E. coli transcripts, which are divided into three major categories: ribosomal, metabolic, and other functions. Data shown for cells cultured in the absence and presence of IPTG (1 mM).