(A-B) Principal component analysis based on log10(mean normalized peak area) separated by growth phase. For this figure, the combination of ionization mode and metabolite was treated as a feature of the data. (C) Pairwise Spearman’s correlations based on log2(fold–change) relative to the ancestor. The black boxes and points indicate the observed correlations, the grey boxes indicate correlations calculated after 100,000 randomizations of fold-changes within each line. p-values indicate the results of a two-tailed t-test between the observed and expected distributions. **** indicates a p-value ≤ 0.0001. (D-E) The observed correlations from C plotted in a network manner. Lines are clustered based on similarity and the color of the line connecting two points indicates the strength of the correlation.

Depiction of three pathways (bold-faced text) that contribute to NAD abundances in the cell. Graphics and pathway names are adapted from the EcoCyc database [36]. All data represents exponential phase measurements. Genes that code for enzymes are shown in purple and metabolites in green. Heatmaps positioned to the right of gene names show the fold-change in expression relative to the ancestor (data from [10]). Grey spaces in gene expression heatmaps represent evolved lines where that gene contains an indel or is deleted. Asterisks indicate genes that are transcriptionally regulated by NadR. Heatmaps positioned to the left of metabolite names show changes in metabolite abundance relative to the ancestor. PnuC transports compounds into the cell. Each heatmap represents one ionization mode, but a mixture of positive and negative ionization mode data is shown depending on which mode a compound was detected. See Fig. S7 for complete data.

Partial depiction of the EcoCyc pathway “superpathway of arginine and polyamine biosynthesis”. All data represents exponential phase measurements. Genes that code for enzymes are shown in purple, and metabolites in green. Heatmaps positioned to the right of gene names show the fold-change in expression relative to the ancestor (data from [10]). Grey spaces in gene expression heatmaps represent evolved lines where that gene contains an indel or is deleted. Asterisks indicate genes that are transcriptionally regulated by ArgR. Heatmaps positioned to the left of metabolite names show changes in metabolite abundance relative to the ancestor. Each heatmap represents one ionization mode, but a mixture of positive and negative ionization mode data is shown depending on which mode a compound was detected. See Fig. S8 for complete data.

The distribution of fold-changes relative to the ancestor for compounds involved in carbon metabolism. Red and black indicate detection in positive or negative ionization mode, respectively. Not all compounds were detected in both ionization modes. Compounds are ordered from top to bottom roughly as they occur in glycolysis or other reactions.

(A) Distributions of different pairwise correlations based on log10(normalized peak area), both (+) and (−) ionization mode data are considered. Data is from all samples with no averaging of replicates. Pairwise indicates all possible pairwise correlations. Replicates indicate comparisons of biological replicates. Intraphase and interphase are comparisons within or across growth phases, respectively. P-value indicates the result of a t-test testing if the intraphase distribution is greater than the interphase distribution. (B) Correlations of compounds across ionization mode. Each point is the correlation of (+) and (−) ionization modes within a single evolved line. (C) Distributions of normalized peak areas are similar across replicates and samples. (+) and (−) indicate ionization mode and colors indicate replicates.

Distributions of peak areas for compounds whose values were imputed using a QRILC method (see section 5.1.4 for a complete description). Imputed values are in orange. The combination of the growth phase (exponential, Ex; stationary, St) and the ionization mode (positive, (+); negative, (−)) is listed on the x-axis.

(A-B) The top 15 compounds contributing to PC1 and PC2 for the exponential phase metabolomes. The (+) and (−) next to a compound indicate the ionization mode of detection, not the charge of the molecule. Colors indicate the row-wise Z-scores based on normalized peak areas.

(A-B) The top 15 compounds contributing to PC1 and PC2 for the stationary phase metabolomes. The (+) and (−) next to a compound indicate the ionization mode of detection, not the charge of the molecule. Colors indicate the row-wise Z-scores based on normalized peak areas.

The theoretical and observed probabilities of finding features (the combination of metabolite and the ionization mode it was detected in) that are significantly altered (|log2(fold–change)| ≥ 1) in a given number of evolved lines (x-axis). Up and down refer to metabolites that are increased and decreased relative to the ancestral strain. Theoretical distributions were calculated using the Sum of Independent Non-Identical Binomial Random Variables (SINIB) method [32]. For more details on this method, see section 5.1.5.

The theoretical and observed number of shared, significantly altered (|log2(fold–change)| ≥ 1) metabolic features (the combination of metabolite and the ionization mode it was detected in) in a given number of evolved lines (x-axis). P-values represent two-tailed t-tests between the observed and theoretical distributions of the number of shared features. The predicted number of shared metabolic features is calculated based on the theoretical probabilities in Fig. S5. Up and down refer to metabolites that are increased and decreased relative to the ancestral strain.

(A) Complete data for Fig. 2. The distribution of fold-changes relative to the ancestor for each compound in each ionization mode is shown. Fold-change is calculated after averaging replicates normalized peak areas. The ancestors are averaged together. (B) Correlations between the various NAD derived compounds. Each axis represents the exponential phase log2(fold – change) relative to the ancestor for that compound, and each point is an evolved line. The dotted line is the 1:1 line.

(A) Complete data for Fig. 3. The distribution of fold-changes relative to the ancestor for each compound in each ionization mode is shown. Compounds are ordered from top to bottom roughly as they occur in the pathway. (B) Fold-change values for all amino acids. Amino acids are ordered by median fold-change in the exponential across both ionization modes.

The relationship between fold-changes in malate and α-ketoglutarate and NAD(H) are correlated within an individual evolved line. Because malate and α-ketoglutarate were only detected in negative ionization mode, only data from this mode is shown. The dotted line is the 1:1 line.