The critical role of S-lactoylglutathione formation during methylglyoxal detoxification in Escherichia coli

Survival of exposure to methylglyoxal (MG) in Gram-negative pathogens is largely dependent upon the operation of the glutathione-dependent glyoxalase system, consisting of two enzymes, GlxI (gloA) and GlxII (gloB). In addition, the activation of the KefGB potassium efflux system is maintained closed by glutathione (GSH) and is activated by S-lactoylGSH (SLG), the intermediate formed by GlxI and destroyed by GlxII. Escherichia coli mutants lacking GlxI are known to be extremely sensitive to MG. In this study we demonstrate that a ΔgloB mutant is as tolerant of MG as the parent, despite having the same degree of inhibition of MG detoxification as a ΔgloA strain. Increased expression of GlxII from a multicopy plasmid sensitizes E. coli to MG. Measurement of SLG pools, KefGB activity and cytoplasmic pH shows these parameters to be linked and to be very sensitive to changes in the activity of GlxI and GlxII. The SLG pool determines the activity of KefGB and the degree of acidification of the cytoplasm, which is a major determinant of the sensitivity to electrophiles. The data are discussed in terms of how cell fate is determined by the relative abundance of the enzymes and KefGB.


Promoter predictions for mltD and yafS
To guide our experimental approach in creating a gloB null mutant, an assessment of promoter elements for the respective genes was undertaken.
In the first instance, promoter predictions were accessed on RegulonDB, a curated database containing information on the transcriptional regulatory network of E. coli K-12 (Gama-Castro et al., 2008). At the time of preparation of this paper RegulonDB stated four potential σ 70 promoters for mltD with the furthest predicted -35 element being 184 bp upstream of the start codon (Table S2) and thereby within the gloB gene. No promoter predictions were stated for yafS on the RegulonDB database.
This analysis was complemented using the web-based tool BPROM that can predict σ 70 promoters (www.softberry.com). The upstream sequences (500 bp from protein encoding sequence) of mltD and yafS were analysed with BPROM using the default settings and potential -10 and -35 promoter elements identified. BPROM predicted one promoter region for mltD, located up to 180 bp upstream (between positions 234109 and 234135 on the chromosome, Table S3) thereby overlapping with the furthest promoter prediction stated in RegulonDB. Two putative promoter regions were predicted for yafS by BPROM (Table S4). The -35 element of the first region was located 68 bp (end of -35 box) upstream from the start codon and thereby also within the gloB gene. A second promoter was predicted further away (~450 bp from start codon), however, this was not considered in our strategy to inactivate gloB since this may have resulted in the expression of a considerable GlxII fragment.  Explanations for BPROM outputs: 1) The length of presented sequence.
3) The number of predicted promoters.
4) The positions of predicted promoters and their scores with 'weights' of two conserved promoter boxes. Promoter position is assigned to the first nucleotide of the transcript (transcription start site position).

Estimation of the number of GlxI & GlxII enzymes per cell
The number of GlxI and GlxII molecules in a single cell can be estimated from the knowledge of the enzyme activity of the purified proteins, the activity in cell extracts and the weight of total cellular protein.
The purified GlxI enzyme exhibits a maximal activity of ~676 µmol · min -1 · mg -1 protein. The dimeric protein has molecular mass of ~30 kDa (Clugston et al., 1998), thus 1 mol of dimeric GlxI can convert ~20280 mol of substrate per minute. The specific GlxI activity in E. coli extracts is ~0.016 µmol · min -1 · mg -1 total cell protein (MacLean et al., 1998). Given the activity of the pure protein, this equates to ~7.89 x 10 -13 mol GlxI per mg of total cell protein. We can approximate that 1 mg total cell protein is equivalent to ~3.6 x 10 9 cells based the following assumptions: OD 650nm of 1 = 1 x 10 9 cells/ml, OD 650nm of 1 = 0.5 mg cell dry weight/ml (Elmore et al., 1990), protein content of cell dry weight: 55 % (Neidhardt & Umbarger, 1996). Therefore a single cell will have 2.17 x 10 -22 mol of GlxI, which after consideration of the Avogadro constant (6.022 x 10 23 ) equates to an estimated number of GlxI molecules of ~130 dimers per cell.
The same calculations can be performed for the GlxII enzyme. The maximal activity of the purified enzyme is ~112 µmol · min -1 · mg -1 protein and the molecular mass is ~ 28.4 kDa (O'Young et al., 2007  Cells that lack either YeiG or FrmB in addition to GlxII are not more sensitive to MG stress than the single mutant lacking GlxII. Cells from the parent ( ), ∆gloB ( ), MJF625 (∆gloB, ∆yeiG; ) and MJF626 (∆gloB, ∆frmB; ) were grown in K0.2 minimal media, exposed to 0.7 mM MG and viable cells enumerated exactly as for experiments presented in Fig. 2. The mean and standard deviation of three independent experiments are shown.

The relationship between intracellular pH and MG
The intracellular pH was determined for the parent (A) and ∆gloB (B) cells treated with a range of MG concentrations (0 -0.8 mM) in K0.2. The intracellular pH of parent cells was measured over 15 min after addition of 0.8 mM MG (C). * MG was added at t15 min thus the t15 min time point has been added as a duplicate of t14 min to illustrate the rapid kinetics of cytoplasm acidification. The data are mean ± s.e.m.