Scheme of the isoleucine/methionine auxotroph strain (IMaux) used in this study as a 2KB biosensor, showing the four previously identified isoleucine biosynthesis pathways in E. coli.

Red arrows indicate gene deletions, grey arrows indicate reactions that are inactive in the presence of oxygen (in addition, glmES is present in some E. coli strains but not in E. coli MG1655). ilvG* indicates that this gene is truncated in E. coli MG1655 due to a frameshift and hence inactive

IMaux strain characterization.

(A) The strain can serve as a sensor for isoleucine, 2KB, or compounds that can be converted to 2KB. Methylaspartate can be converted to 2KB in vivo via promiscuous reactions of native E. coli enzymes (B) Supplementation of branched chain amino acids valine, leucine, and isoleucine (BCAA) at 1 mM each relieves 2KB toxicity, indicating that the mechanism involves dysregulation of BCAA biosynthesis. All experiments have been performed in M9 minimal medium with 20 mM glycerol as a main carbon source and 1 mM methionine. Growth curves are the average of at least three technical replicates. DT, doubling time.

The proposed recursive isoleucine biosynthesis pathway. Pyruvate is condensed with glyoxylate by AHAS II encoded by ilvG, then the product of this reaction is reduced and undergoes dehydration to 2KB.

The resulting 2KB undergoes the same sequence of reactions to produce the isoleucine precursor (3S)-3-methyl-2-oxopentanoate (MOP), which is further transaminated to isoleucine. AHAS II is the only AHAS enzyme in E. coli that can catalyze the first reaction with glyoxylate and pyruvate in vivo, though the condensation of 2-KB with pyruvate can also mediated by AHAS I and AHAS III. Colored arrows indicates that the same enzyme promiscuously catalyzed both reactions. HMOP, 2-hydroxy-2-methyl-3-oxopropanoate; DHB, (R)-2.3-dihydroxybutanoate; 2KB, 2-ketobutyrate; AHB, (S)-2-aceto-2-hydroxybutanoate; DHMP, (R)-2,3-dihydroxy-3-methylpentanoate; MOP, (3S)-3-methyl-2-oxopentanoate.

(A) Growth of the IMaux ΔilvG strain with or without plasmid-based overexpression of ilvGM. All strains were grown in M9 with 20 mM glycerol and 1 mM methionine, plus supplemental carbon sources as indicated in the legend. Growth curves are the average of at least three technical replicates. DT, doubling time. (B) Max OD reached with different glyoxylate concentrations. Strains were grown in M9 with 20 mM glycerol, 1 mM methionine, and 10 µM cuminic acid as inducer. Blue dots indicate a range of glyoxylate concentrations, orange represents isoleucine (positive control).

(A) Expected isoleucine labeling patterns based on 13C glucose and unlabeled 2KB or unlabeled glyoxylate. Orange and white circles represent labeled and unlabeled carbons, respectively. Dashed arrows indicate multiple reactions. (B) Observed isoleucine labeling patterns for strains grown on 13C glucose supplemented with 12C carbon sources as indicated in the figure. The GC-MS derivatization method removes the C1 carbon, leaving only the C2-C6 fragment for analysis. Bars represent the average of three technical replicates.

FBA predictions of theoretical biomass and isoleucine yield for the discovered route and other isoleucine biosynthesis routes in E. coli.

a this route requires an anaerobic environment. NA indicates this route is not applicable as anaerobic growth (without electron acceptor) is not feasible.

Summary of the demonstrated isoleucine biosynthesis routes in E. coli.

Strains and plasmid used in this study.

Summary of hypothetical 2KB biosynthesis pathways.

Question marks indicate that the indicated reaction has not been demonstrated.