Optimization of carbon and energy utilization through differential translational efficiency

Abstract

Control of translation is vital to all species. Here we employ a multi-omics approach to decipher condition-dependent translational regulation in the model acetogen Clostridium ljungdahlii. Integration of data from cells grown autotrophically or heterotrophically revealed that pathways critical to carbon and energy metabolism are under strong translational regulation. Major pathways involved in carbon and energy metabolism are not only differentially transcribed and translated, but their translational efficiencies are differentially elevated in response to resource availability under different growth conditions. We show that translational efficiency is not static and that it changes dynamically in response to mRNA expression levels. mRNAs harboring optimized 5'-untranslated region and coding region features, have higher translational efficiencies and are significantly enriched in genes encoding carbon and energy metabolism. In contrast, mRNAs enriched in housekeeping functions harbor sub-optimal features and have lower translational efficiencies. We propose that regulation of translational efficiency is crucial for effectively controlling resource allocation in energy-deprived microorganisms.

Fig. 1

Overview of omic experiments carried out and the correlation between RNA-seq and Ribo-seq in all growth conditions. a Correlations between RNA-seq and Ribo-seq in CO, H₂:CO₂, and fructose. Pearson’s (P) and Spearman’s (S) correlation coefficients are shown inside each subfigure. Colors in the scatter plot represent the translational efficiency values. The color bars show TE log₂ values. b An example of Ribo-seq, RNA-seq, TE, and TSS profiles mapped onto genomic region between 4,535,800 to 4,564,000 and showing genes Clju_RS20670 to Clju_RS20800. RNA-seq and Ribo-seq profiles were normalized to reads per million (RPM). TE of each gene is a ratio Ribo-seq to RNA-seq level. Arrows indicate transcription start site (TSS) positions

Fig. 2

Differential translation and differential TE of subsystems in fructose, CO and H₂:CO₂ growth cultures. Genes were grouped into RAST subsystems and translation and transcription were both percent-normalized per each experiment. Each column represents the level of translation or transcription (% normalized) per each experiment and the bubble color reflects the intensity (% normalized). Each row represents a comparison between the three datasets and the size of bubbles represents the level of translation or transcription per each subsystem (% normalized). TE is compared per each subsystem and depicted by triangles. Subsystems with differentially increased TE are depicted by up-pointing triangle with thick edges and the colors represent percent-normalized values across the three conditions. The number of genes is shown per each subsystem. The top panel manifests the top 20 subsystems that are differentially upregulated (P < 0.01) at the translational level in fructose (heterotrophic) relative to both autotrophic conditions and the data are sorted according to percent translation in fructose. The bottom panel manifests subsystems that are differentially upregulated (P < 0.01) at the translational level in both autotrophic conditions with data sorted in descending order according to percent translation levels in CO. All values are shown as log₂

Fig. 3

Metabolic map of major carbon and energy pathways exhibiting differential translation and differential TE. Differential fold change is calculated as the log₂ CO/fructose or H₂:CO₂/fructose translation ratios. Heterotrophically induced (red arrows), autotrophically induced (blue arrows), insignificant (gray arrows) and condition-specific (green arrows) translation is depicted in all pathways. Glycolysis and Gluconeogenesis: fructose phosphotransferase system (PTS); fructokinase/fructose-6-phosphate isomerase (G1); 1-phosphofructokinase (G2); 6-phosphofructokinase (G3); ketose-bisphosphate aldolase (G4); triose-phosphate isomerase (G5); glyceraldehyde-3-phosphate dehydrogenase (G6); phosphoglycerate kinase (G7); phosphoglycerate mutase (G8); enolase phosphopyruvate hydratase (G9); pyruvate:ferredoxin oxidoreductase (P1); pyruvate, phosphate dikinase (P2); pyruvate kinase (P3); pyruvate carboxylase (P4); PEP carboxykinase (P5). Fermentation: phosphotransacetylase (Ac1), acetate kinase (Ac2), bifunctional aldehyde/alcohol dehydrogenase (E1), aldehyde:ferredoxin oxidoreductase (E2), additional alcohol dehydrogenases (E3), acetolactate synthase (B1), acetolactate decarboxylase (B2), 2,3-butanediol dehydrogenase (B3); lactate dehydrogenase (L). Incomplete TCA cycle: citrate synthase (T1); citrate lyase (T2); aconitase (T3); isocitrate dehydrogenase (T4); malate dehydrogenase (T5); fumarase (T6); fumarate reductase (T7). Wood−Ljungdahl pathway: electron-bifurcating [FeFe] hydrogenase (H1); other [FeFe] hydrogenases (H2); [NiFe] hydrogenase (H3); hydrogenase maturation factor (H4); bifunctional CO dehydrogenase/ acetyl-CoA synthase (CODH/ACS) (W1); seleno formate dehydrogenase (W2); non-seleno formate dehydrogenase (W3); formyl-THF ligase (W4); methenyl-THF cyclohydrolase (W5); methylene-THF dehydrogenase (W6); methylene-THF reductase (W7). Rnf complex and ATPase: RnfC (RC); RnfD (RD); RnfG (RG); RnfE (RE); RnfA (RA); RnfB (RB); ATPase (A). Fructose (Fruc); fructose 1-phosphate/6-phosphate (Fruc-1P/-6P); fructose 1,6-bisphosphate (Fru-1,6P); dihydroxyacetone phosphate (DHAP); glycerol 3-phosphate (Gly-3P); 1,3-bisphosphoglycerate (1,3-DPG); 3-phosphoglycerate (3-PG); 2-phosphoglycerate (2-PG); phosphoenolpyruvate (PEP); oxaloacetate (OXACO); citrate (Cit); isocitrate (Cit-Ac); a-ketoglutarate (a-KetoGlu); malate (Mal); fumarate (Fum); succinate (Suc); acetolactate (AcLac); acetoin (Acn); acetaldehyde (AcHO); acetyl-phosphate (Acl-p); tetrahydrofolate (THF); reduced ferredoxin (Fdr); oxidized ferredoxin (Fdo)

Fig. 4

Transcriptional and translational regulation of the Rnf (white), formate dehydrogenase (black), and hydrogenase (gray) complexes in all growth conditions. Results are manifested for CO, H₂:CO₂, and fructose in blue, green and red, respectively. a Transcription and translation of the Rnf complex (normalized to reads per million (RPM)). The rnf complex (Clju_c11350-Clju_c11410) has one major TSS upstream of rnfC. The rnfEAB genes are transcribed from an internal promoter that is positioned at the 3′-end of rnfG. The rseC gene is transcribed from one TSS and transcription is comparable across all conditions; however, it is poorly translated in fructose. b Comparison between TE of the Rnf genes in each condition. The rnfC gene has the lowest TE in fructose. c Shotgun proteomics showing the relative abundance of the Rnf complex proteins (RnfA was not detected) in H₂:CO₂ and fructose. d Expression of the formate dehydrogenase and the hydrogenase genes (Clju_c06990-Clju_c07080, normalized to reads per million (RPM)). Both clusters are expressed from upstream TSSs. The hydN and the fdhA genes are translated at a much lower efficiency compared to the hydrogenase B and D genes despite having higher transcription. e Comparison between the TE of the hydrogenase and the formate dehydrogenase genes in each condition. The hydrogenase E2 gene is transcribed at a higher level from an internal promoter

Fig. 5

Influence of 5´UTR features on TE. a Comparison between genes with low and high TE in all conditions. Low-TE genes are below 20th percentile, whereas high-TE genes are above 80th percentile in all conditions. P < 1e⁻¹⁰ are signified with “***”, 1e⁻¹⁰ ≤ P < 0.01 are signified with “*” and P ≥ 0.01 are signified with “n.s”. b Low- and high-TE genes in all conditions have visible differences in their RBS sequence and the AU content in their upper RBS region (URR). The RBS and the URR regions are both highlighted with dotted gray boxes. c RBS affinity towards the anti-Shine Dalgarno (aSD) sequence (AAGGAGGU) positively affects TE in all conditions. The affinities of RBS towards the aSD region were categorized into 11 groups ranging from ∆G of 0 to −12.7. d The RBS motif per each category in CO was determined using MEME. e Positive effect of the 15 bp AU% content in the URR on TE in all conditions. TSS data were used to ensure that TSS is upstream of the URR. f The distance of the RBS 5′ end from the start codon is most optimum at 13 bp. Deviation of the RBS position in either direction negatively influences TE

Fig. 6

mRNA features and transcription levels determine the dynamics of differential TE. a Scatter plot between log₂ TE as four quartiles on the Y axis and RNA-seq (FPKM) values grouped into ten bins on the X axis. TE_max cutoffs are shown next to each regression with the same color codes. The dots represent the mean of TE in each RNA-seq category. The vertical lines over each dot represents 95% confidence intervals of each bin. The linear regression lines are fitted with shadowed regions representing 95% confidence intervals. b Stacked barplot demonstrating the enrichment of genes in RAST categories within each TE_max quartile (shown as percent) with the same color codes as in (a). The white numbers represent gene counts for each TE_max quartile in a given RAST category. Significantly enriched quartiles (Fisher exact test) are indicated with circled gene count numbers