1. Biochemistry and Chemical Biology
Download icon

Underground isoleucine biosynthesis pathways in E. coli

  1. Charles AR Cotton
  2. Iria Bernhardsgrütter
  3. Hai He
  4. Simon Burgener
  5. Luca Schulz
  6. Nicole Paczia
  7. Beau Dronsella
  8. Alexander Erban
  9. Stepan Toman
  10. Marian Dempfle
  11. Alberto De Maria
  12. Joachim Kopka
  13. Steffen N Lindner
  14. Tobias J Erb
  15. Arren Bar-Even  Is a corresponding author
  1. Max Planck Institute of Molecular Plant Physiology, Germany
  2. Max Planck Institute for Terrestrial Microbiology, Germany
  3. LOEWE Research Center for Synthetic Microbiology (SYNMIKRO), Germany
Research Article
Cite this article as: eLife 2020;9:e54207 doi: 10.7554/eLife.54207
9 figures, 3 tables and 3 additional files


Metabolic routes for 2-ketobutyrate biosynthesis.

A wide variety of pathways support the production of 2-ketobutyrate, an essential intermediate in the production of isoleucine. Outlined are five known, physiologically relevant 2-ketobutyrate synthesis pathways and two additional pathways we identified in this paper (marked in blue). The purple font indicates the key enzymes of the novel 2-ketobutyrate production pathways in vivo. Note, the pathways are represented in shorthand between identifiable intermediates and do not show each metabolic step.

Strains lacking threonine deaminases adapt to generate isoleucine under aerobic conditions.

Strains deleted in threonine deaminases (∆ilvA ∆tdcB) or threonine deaminases and serine deaminases (5, i.e. ∆ilvAtdcBsdaAsdaBtdcG) grew on minimal medium with 10 mM glucose after approx. 70 hr. Strains deleted in ketol-acid reductoisomerase (∆ilvC, an enzyme downstream of 2 KB in isoleucine biosynthesis) could not grow (valine supplied to relieve a second auxotrophy), indicating a metabolic leak at the level of 2 KB. All experiments were performed in technical triplicates in a 96-well plate reader. For the ∆ilvA ∆tdcB and 5 strains, cultivated without isoleucine (purple and green lines), replicates are shown individually. In all other cases, replicates showed an identical growth profile (±5%) and hence are represented by a single curve. Experiments were repeated on three separate occasions for all growth experiments shown. Isoleucine, 2-ketobutyrate, and valine were added at 2 mM where indicated.

13C-labeling of isoleucine rules out the citramalate pathway.

Labeling of proteinogenic amino acids was analyzed after the growth of E. coli on labeled glucose-1-13C or glucose-3-13C as a sole carbon source. (A) Schematic representation of predicted labeling of isoleucine with the natural isoleucine biosynthesis pathway of E. coli (left) or with the citramalate pathway (right). Blue coloring highlights predicted labeling when grown on glucose-1-13C and green coloring when grown on glucose-3-13C. (B) Observed labeling pattern. There is no difference in the labeling pattern between the mutant strains and the WT strain, ruling out the citramalate pathway for isoleucine biosynthesis. 5 represents ∆ilvAtdcBsdaAsdaBtdcG. Labeling experiments were repeated on at least two separate occasions.

Methionine biosynthesis is implicated in promiscuous 2-ketobutyrate production.

The ∆5 strain further deleted in homoserine O-succinyltransferase (∆5 ∆metA) did not grow when supplemented with 2 mM methionine (orange line) but mutants of the cystathionine β-lyase (∆5 ∆metC) grew rapidly and consistently on 10 mM glucose alone. A mutation in serine acetyltransferase (∆5 cysE*) resulted in near WT growth on glucose (purple line). Both isoleucine (Ile) and methionine (Met) were added at a concentration of 2 mM where indicated. All experiments were performed in replicates in a 96-well plate reader. Replicates showed an identical growth profile (±5%) and hence are represented by a single curve. Experiments were repeated at least three times for all growth experiments shown.

Figure 5 with 3 supplements
Possible routes to 2-ketobutyrate by methionine biosynthesis enzymes.

Schematic representation of methionine biosynthesis from homoserine as operating in E. coli (black arrows). Promiscuous reactions of cystathionine γ-synthase (MetB) and cystathionine β-synthase (MetC) which have been suggested to produce 2-ketobutyrate in various organisms are shown as red arrows.

Figure 5—figure supplement 1
SDS-PAGE of all purified enzymes used in this study.

Expected monomeric molecular masses are: MetB 41.6 kDa, MetC 43.12 kDa, CysE 30.3 kDa, CysE Ala33Thr 30.3 kDa, PFL-AE 30.5 kDa, PFL 87.7 kDa, TdcE 88.3 kDa.

Figure 5—figure supplement 2
MetB catalyzes O-succinyl-L-homoserine (OSHS) cleavage to generate 2-ketobutyrate (2 KB).

(A) Kinetics of OSHS cleavage by MetB. (B) Formation of 2 KB is suppressed by the addition of cysteine (Cys, 300 µM) and (C) to a much lesser extent by the addition of homocysteine (Homo-Cys, 300 µM). The same phenomenon is observed at much higher concentrations of cysteine (Cys, 3 mM – Panel D) and full homocysteine-dependent repression can be observed at concentrations of this magnitude (Homo-Cys, 6 mM – Panel E). Error bars correspond to standard deviations. Note: MetB can perform two separate replacement reactions: either with cysteine as co-substrate (black arrow, Figure 5) or with homocysteine as co-substrate (reaction 2, Figure 5). To visualize the elimination (reaction one in Figure 5) and replacement reactions, the succinate concentrations shown here account for the succinate generated exclusively from the replacement reaction and are calculated by subtracting the concentration of 2 KB (which is stoichiometrically 1:1 with the concentration of succinate generated by the elimination reaction) from the total concentration of succinate. Effectively, succinate shown in panels B–E is exclusively the product of the replacement reaction 2 in Figure 5. The bar charts in panels D and E represent the extent to which the elimination and replacement reactions contribute to the overall catalytic turnover of MetB.

Figure 5—figure supplement 3
Kinetic characterization of CysE variants.

(A) Michaelis-Menten plots of CysE variants as a function of acetyl-CoA concentration at a saturating L-serine concentration of 20 mM and (B) Michaelis-Menten plots of CysE variants as a function of serine concentration at a non-saturating acetyl-CoA concentration of 0.6 mM. All points were measured in triplicate, showing the mean and standard deviation.

Deletion of metC or mutation in cysE substantially affects metabolite concentrations and gene expression.

(A) Quantitative determination of the concentrations of methionine and O-succinyl-L-homoserine as performed using an HRES-LC-MS. The concentration of methionine in the Δ5 ΔmetC and Δ5 cysE* strains was 2- to 3-fold lower than in the WT strain and the Δ5 strain (p-value <0.05, rank sum test). Isoleucine was added only to the Δ5 strain (as it is required for its growth). Relative to the WT strain and the Δ5 strain, the concentration of O-succinyl-L-homoserine was 2- to 3-fold higher in the Δ5 ΔmetC strain (p-value <0.05, rank sum test) and 30- to 60-fold higher in the Δ5 ΔcysE* strain (p-value <0.05, rank sum test). Error bars correspond to standard deviations. (B) Quantitative determination of transcript levels of metA and metB as measured by reverse transcriptase quantitative PCR. The transcript levels of metA and metB were more than 3-fold higher in the Δ5 cysE* strain than in the WT strain and the Δ5 strain (p-value <0.05, rank sum test). Error bars correspond to standard deviations.

Figure 6—source data 1

Measured concentrations of methionine and O-succinyl-L-homoserine.

Figure 6—source data 2

Measured transcript levels of metA and metB.

A mutation in serine acetyltransferase (cysE*) enables 2-ketobutyrate production by MetB.

(A) A schematic representation of endogenous cysteine, methionine, and isoleucine biosynthesis in E. coli. Thickness of arrows indicates flux intensity. (B) Schematic representation of cysteine, methionine, and isoleucine biosynthesis when threonine deaminases are deleted and with the cysE* mutation. Thickness of the arrows indicates higher flux via O-succinyl-L-homoserine cleavage to 2-ketobutyrate. As indicated by the red arrows, the transcript levels of metA and metB as well the concentration of O-succinyl-L-homoserine increase, while the concentrations of cysteine and methionine decrease. Numbers next to the arrows correspond to the factor by which the metabolite concentration or transcript level increased or decreased (as derived from Figure 6). Cysteine concentrations in the Δ5 ΔcysE* strain were ≈50% of that in WT and Δ5 strains.

Figure 8 with 3 supplements
Propionate-dependent isoleucine production under anaerobic conditions.

(A) Addition of 5 mM propionate enabled growth of the Δ5 strain under anaerobic conditions. Further addition of 30 mM formate improved growth substantially. (B) A Δ5 ΔpflB strain could not use propionate as precursor of isoleucine. 1 mM of acetate was added for the cultivation of this strain, as the deletion of pflB disrupted the endogenous acetyl-CoA biosynthesis. (C) Overexpression of tdcE in the Δ5 ΔpflB strain enabled growth with propionate as isoleucine precursor. All experiments were performed in replicates in a 96-well plate reader. Replicates showed an identical growth profile (±5%) and hence are represented by a single curve. Experiments were repeated three times for all growth experiments shown.

Figure 8—figure supplement 1
Growth via the MetB-dependent pathway under anaerobic conditions.

Both the isolated ΔilvA ΔtdcB CysE* (solid line) and MAGE-generated Δ5 CysE* (dotted line) strains grew under anaerobic conditions without the medium addition of isoleucine (or propionate). The biphasic growth of the engineered Δ5 CysE* strain is unusual, but we note the same growth is observed for the positive control (green dotted). The timeframe of this growth and the lack of propionate in the media rule out PFL-dependent growth.

Figure 8—figure supplement 2
In vitro activation and validation of PFL and TdcE activity (A) UV-Vis absorbance spectrum of 100 µM PFL-AE before and after reconstitution.

The increase in intensity between 400 and 500 nm indicates increased iron-sulfur cluster content (Byer et al., 2018). (B) SDS-PAGE analysis of the activation mix with (+) and without (−) PFL-AE. The activation mix was exposed to oxygen and then SDS loading dye was added. The double band in the control without PFL-AE is probably due to partial glycyl radical activation during overexpression in E. coli. (C) Glycyl radical stability after activation. The activation mix was stored on ice and the activity of PFL and TdcE was measured with the pyruvate cleavage assay at the time points indicated. This experiment was repeated on three different days.

Figure 8—figure supplement 3
Michaelis-Menten graphs of PFL and TdcE for propionyl-CoA and formate (see Materials and methods section for exact conditions).

Error bars show standard deviation of three replicates. The data was fit to the Michaelis-Menten equation, except for PFL with propionyl-CoA, for which the data was fit to a substrate inhibition equation. The kinetic parameters are listed in Table 1.

13C-labeling of isoleucine reveals the activity of the KBFL route under anaerobic conditions.

(A) Schematic representation of 2-ketobutyrate production via PFL/TdcE-catalyzed reaction. Compounds in gray are added to the medium (external) while those in black represent intracellular metabolites. The red circle indicates the labeled carbon. (B) 5 mM of propionate-1-13C were added to the medium of different strains cultivated on 10 mM glucose. 10 mM acetate, 30 mM formate, and/or 2 mM methionine were added to the media of some strains, as indicated at the bottom of the figure. Alanine and threonine were not labeled in all cases, confirming that the labeling of isoleucine is derived from direct propionate conversion to 2 KB.

Figure 9—source data 1

Retention time, relative peak, and m/z for the carbon labeling experiment.



Table 1
Kinetic parameters of PFL and TdcE for catalyzing the condensation of propionyl-CoA with formate to generate 2-ketobutyrate.

See Materials and methods for assay conditions and Figure 8—figure supplements 2 and 3. Note that the kcat values were calculated assuming that the enzymes are fully activated.

Vmax (U/mg)kcat (s−1)KM (µM)Ki (µM)Vmax (U/mg)kcat (s−1)KM (mM)
TdcE2.1 ± 0.13.2 ± 0.2520 ± 81-1.8 ± 0.072.7 ± 0.120 ± 3
PFL (PflB)1.1 ± 0.61.7 ± 0.9830 ± 580690 ± 5000.59 ± 0.030.9 ± 0.0569 ± 11
Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Escherichia coli)E. coli strainsThis studyTable 2
Sequence-based reagentThis studyPCR primersSupplementary file 2
Recombinant DNA reagentpZASS-tdcE (plasmid)This studyOverexpression vector of TdcE.
Recombinant DNA reagentpCA24N-metB (plasmid)This studyOverexpression vector of MetB.
Recombinant DNA reagentpCA24N-metC (plasmid)This studyOverexpression vector of MetC.
Recombinant DNA reagentpCA24N-cysE (plasmid)This studyOverexpression vector of CysE.
Recombinant DNA reagentpCA24N-cysE_A33T (plasmid)This studyOverexpression vector of CysE with a mutation A33T.
Recombinant DNA reagentpORTMAGENyerges et al., 2016Addgene catalog no. 72680
Chemical compound, drug13C-1-glucoseSigma Aldrich
Chemical compound, drug13C-3-glucoseSigma Aldrich
Chemical compound, drugSodium propionate-1-13CSigma Aldrich
Commercial assay or kitRNeasy Mini KitQiagen
Commercial assay or kitqScript cDNA Synthesis KitQuantaBio
Commercial assay or kitMaxima SYBR Green/ROX qPCR Master MixThermo Fisher Scientific
Software, algorithmbreseqDeatherage and Barrick, 2014RRID:SCR_010810
Table 2
Strains used in this study.
Strain nameDeletions/GenotypeDescriptionSource
BL21 (DE3)E. coli fhuA2 [lon] ompT gal (λ DE3) [dcm] ∆hsdS
λ DE3 = λ sBamHIo ∆EcoRI-B int::(lacI::PlacUV5::T7 gene1) i21 ∆nin5
Strain for protein productionLab collections
SIJ488E. coli
K-12 MG1655Tn7:: pAra-exo-beta-gam; pRha-FLP; xylSpm-IsceI
MG1655 derivative with genome integrated recombinase and flippase genes57
ilvAtdcBSIJ488 ∆ilvAtdcB::kanThreonine deaminase deletion strainThis study
∆5SIJ488 ∆ilvAtdcBsdaAsdaBtdcG::kanThreonine deaminase and serine deaminase deletion strainThis study
∆5 ∆metASIJ488 ∆ilvAtdcBsdaAsdaBtdcGmetA::kanThreonine deaminase, serine deaminase and homoserine O-succinyltransferase deletion strainThis study
∆5 ∆metCSIJ488 ∆ilvAtdcBsdaAsdaBtdcGmetC::kanThreonine deaminase, serine deaminase and cystathionine β-lyase deletion strainThis study
∆5 cysE*SIJ488 ∆ilvAtdcBsdaAsdaBtdcGmetC::kan cysEThreonine deaminase, serine deaminase and homoserineO-succinyltransferase deletion strain with cysE point mutation G97AThis study
ilvCSIJ488 ∆ilvC::capKetol-acid reductoisomerase deletion strainThis study
∆5 ∆pflBtdcEilvAtdcBsdaAsdaBtdcGpflBtdcE::capThreonine deaminase, serine deaminase, pyruvate formate-lyase and 2-ketobutyrate formate-lyase deletion strainThis study
pflBSIJ488 ∆pflB::kanPyruvate formate-lyase deletion strainThis study
tdcESIJ488 ∆tdcE::cap2-Ketobutyrate formate-lyase deletion strainThis study

Additional files

Supplementary file 1

Summary of identified mutations in 16 strains lacking threonine deaminases (∆ilvA ∆tdcB) or threonine and serine deaminases (∆ilvA ∆tdcB ∆sdaA ∆sdaB ∆tdcG) that were evolved to grow on 10 mM glucose after 70 hr without isoleucine or 2 KB.

Supplementary file 2

List of all primers used in this study.

Transparent reporting form

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)