Genomic Rearrangements Leading to Overexpression of Aldo-Keto Reductase YafB of Escherichia coli Confer Resistance to Glyoxal

Abstract

Glyoxal is toxic and mutagenic α-oxoaldehyde generated in vivo as an oxidation by-product of sugar metabolism. We selected glyoxal-resistant mutants from an Escherichia coli strain lacking major glyoxal-detoxifying genes, gloA and yqhD, by growing cells in medium containing a lethal concentration of glyoxal. The mutants carried diverse genomic rearrangements, such as multibase deletions and recombination, in the upstream region of the yafB gene, encoding an aldo-keto reductase. Since these genomic lesions create transcriptional fusions of the yafB gene to the upstream rrn regulon or eliminate a negative regulatory site, the mutants generally enhanced an expression of the yafB gene. Glyoxal resistances of the mutants are correlated with the levels of yafB transcripts as well as the activities of aldo-keto reductase. An overproduction of YafB in the glyoxal-resistant mutant lacking the putative NsrR-binding site provides evidence that the yafB gene is negatively regulated by this protein. We also observed that the expression of yafB is enhanced with an increased concentration of glyoxal as well as a mutation in the fnr gene, encoding a putative regulator. The bindings of NsrR and Fnr to the yafB promoter were also demonstrated by gel mobility shift assays.

INTRODUCTION

Glyoxal (GO) is a reactive α-oxoaldehyde formed by the oxidative degradation of glucose, lipid peroxidation, and autoxidation of ascorbic acid (5, 20). Due to its two reactive carbonyl groups, GO is capable of inducing cellular damage through multiple mechanisms. GO is converted to glycolaldehyde by aldehyde reductase YqhD (14) and presumably by aldo-keto reductases (AKRs) (Fig. 1) (13). Glycolaldehyde can be further oxidized or reduced to glycolic acid or 1,2-ethandiol. Glycolaldehyde oxidation is mediated by aldehyde dehydrogenase AldA and NAD⁺, while its reduction is accomplished by 1,2-propanediol oxidoreductase FucO using NADH as a cofactor (14). Glycolic acid is also believed to be generated directly from glyoxal by glutathione (GSH)-dependent glyoxalases (28). In contrast to GO, methylglyoxal (MG) detoxification is well known, involving the glyoxalase system and AKRs. The glyoxalase system consists of glyoxalase I and II (Glo I and Glo II; encoded by gloA and gloB, respectively), which require GSH as a cofactor, producing lactate from MG (29). The AKRs, including YafB (DkgB), YqhE, and YghZ, were also reported to detoxify MG by generating hydroxyacetone using NADPH (13). Although AKRs in pathogenic Escherichia coli were suggested as potential virulence factors (11), few AKRs of E. coli K-12 have been characterized in terms of their biological functions. Recently, we reported that Hsp31 (hchA) catalyzes the conversion of GO and MG to glycolic acid and d-lactate (Glo III in Fig. 1), respectively, without an additional cofactor (27).

Fig 1.

Metabolic pathway for glyoxal (GO) and methylglyoxal (MG). GO can be reduced to glycolaldehyde and further to 1,2-ethandiol with the involvement of AKRs, such as YafB, YqhE, and YghZ. MG is converted to d-lactate, lactaldehyde, and acetol by glyoxalase, MG reductase, and AKRs, respectively. Glo III (hchA) catalyzes conversions of GO and MG to glycolic acid and d-lactate without cofactors. AldA, aldehyde dehydrogenase; FucO, l-1,2-propanediol oxidoreductase; Glo I, glyoxalase I; Glo II, glyoxalase II; Glo III, glyoxalase III; GldA, glycerol dehydrogenase; YqhD, aldehyde reductase; GSH, glutathione; SHG, S-2-hydroxyethyl glutathione; SLG, S-lactoylglutathione.

The GO toxicity is caused by its lone pairs of electrons covalently modifying macromolecules. There are multiple targets for electrophile-mediated damages, including DNA, RNA, lipids, metabolites, and proteins (30). GO reacts primarily with guanine to form a tricyclic glyoxal-G adduct (gG) (21). Because gG can pair with all the bases in DNA (T, G, C, or protonated adenine), the formation of gG causes a mutation through mispairing of DNA strands during replication. Because GO rapidly reacts with guanine, this type of modification may be an earlier event, when cells are exposed to GO (30). The unpaired bases in single-stranded DNA (ssDNA) are less protective from GO and more mutable than stably paired bases. An in vitro analysis indicates that ssDNA reacts with GO more rapidly than double-stranded DNA (dsDNA) (12). This suggests that besides all kinds of RNAs, ssDNA regions, such as the replication forks and transcription bubbles, may be vulnerable to GO modification.

The promoters of rRNA and tRNA are known to be highly active during exponential growth such that over half of the total transcripts in E. coli are those of rRNA and tRNA, even though rRNA genes account for only 0.5% of the genome (22). The seven rRNA operons of E. coli K-12 typically consist of 16S, an internally transcribed spacer (ITS) containing at least one tRNA, 23S, and 5S rRNA genes. Some operons have one or two additional tRNAs following the 3′ distal region of the 5S gene. Exceptionally, the rrnD operon contains not only the conserved 5S rRNA gene (rrfD) but also an additional 5S gene (rrfF) in the distal end of the operon (see Fig. 4C). The rrn operons have two tandem promoters (P1 and P2; see Fig. 4C) on their 5′ ends, synthesizing transcripts that are processed to generate three mature rRNAs and tRNAs. The rrn operons are distributed around the replication origin (oriC) and transcribed in the same direction as the chromosome is replicated. The tRNA genes are in general cotranscribed with the rRNA genes (2), although there has been a report that the aspU gene, the distal tRNA of the rrnH operon located upstream of yafB, might be transcribed partially independently of the promoters upstream of the rrnH operon (33).

Fig 4.

Genomic rearrangements in the GO-resistant mutants and their consequences. (A) Extents and locations of the deletions in the upstream region of yafB. Multibase deletions in the mutants are shown with dashed (1), dotted (2), and solid (3) lines. The intergenic sequences (lowercase letters) are distinguished from genes (uppercase letters). The location of the 4-bp deletion (GCAT) in the NsrR-binding site of mutant 5 is presented with the filled triangle (▲) below the dotted line. The putative NsrR- and Fnr-binding sites are underlined and in boldface. The transcription initiation site was determined previously (17), with its putative −10 consensus for RpoS (shading; EcoCyc, http://ecocyc.org/). Terminator and anti-antiterminator of aspU are shown with arrows with solid and dashed lines, respectively. (B) Schematic representation of multibase deletion mutations. Because of the differences in deletion sizes, PCR products, amplified with yafBup-F and yafBup-R primers, from the GO-resistant mutants are smaller than those of the wild type (328 bp). Putative NsrR- and Fnr-binding sites are presented with white and gray boxes, respectively. The anti-antiterminator and terminator of aspU are shown with dotted and solid lines (loops), respectively. (C) Locations of the seven rRNA operons (rrn) on the E. coli chromosome. Arrows indicate directions of transcriptions. The origin of chromosomal replication, oriC, is positioned at about 84.5 min. The distance between rrnD and oriC is much shorter than that between rrnH and oriC. Schematic representation of the recombination event in mutant 4. The sequences from rrnD are shown in gray. Locations of primers for RT-PCR are indicated with short arrows: rrl-F, primer 1 and 1′; rrf-F, primer 2 and 2′; aspU-F, primer 3; thrV-F, primer 3′; and yafB-R, primer 4. (D) Confirmation of read-through messages by RT-PCR. The amplified read-through transcripts are separated on a 1.2% agarose gel. In the cases of mutants 2 and 3, no transcripts were observed in lanes 3 due to a removal of the aspU gene. The read-through transcripts from the rrnH promoter in the wild type were detected by additional rounds of PCR as shown in the left panel. Similar transcripts were seen with mutant 5 (data not shown).

NsrR is a nitrite-sensitive transcription repressor sensing nitric oxide (NO). NsrR requires the presence of an iron-sulfur (2Fe-2S) cluster for sensing redox change. NsrR has gained attention because of its possible role as a global regulator and belongs to a family of helix-turn-helix DNA-binding proteins (31). Although the NsrR-binding site was predicted to comprise two copies of an 11-bp motif arranged as an inverted repeat with 1-bp spacing, analysis of the chromatin immunoprecipitation with microarray technology (ChIP-chip) data suggested that a single half-motif (with the consensus sequence AANATGCATTT) can function as an NsrR-binding site in vivo (23).

The transcription factor fumarate nitrate reduction (Fnr) protein of E. coli is a global redox-responsive regulator that activates and represses a family of genes required for anaerobic and aerobic metabolism (3). Fnr recognizes an inverted repeat (TTGAT N4 ATCAA) separated by four nonconserved base pairs. It has been reported that a single functional half-site is sufficient to binding and activation of transcription (3, 18).

Previous studies indicate that cellular machineries coping with enhanced levels of glyoxals are diverse (13). They encompass enzymes as well as potential targets presumably associated with modifications by glyoxals (14) (C. Lee, unpublished data). The well-documented case involves a regulatory mutation in the transcription factor YqhC, overexpressing aldehyde reductase YqhD, which is responsible for detoxifying glyoxal (14). Here, we attempted to search for other genes associated with glyoxal by selecting glyoxal-resistant E. coli mutants deficient in yqhD and gloA and characterize their alterations genetically and biochemically.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Plasmids, strains, and phages used are listed in Table 1. All strains are derivatives of E. coli K-12. MG1655 was used as a wild-type strain for gene disruption and gene amplification. Other mutations were obtained from the CGSC (Yale University, New Haven, CT) and transferred to MG1655 using P1 phage. We introduced yqhD::kan and gloA::kan alleles (MJF388) (15) into the MG1655 strain to make the MG1655ΔgloA ΔyqhD mutant. The kanamycin cassette in yqhD::kan, fnr::kan, and nsrR::kan alleles were, if necessary, removed by using Flp recombinase (4). The BL21(DE3) strain (Novagen) was used for an overexpression and purification of proteins. Antibiotics were used at the following concentrations: tetracycline (Sigma), 17 μg/ml; kanamycin (Duchefa), 25 μg/ml; ampicillin (BioBasic), 100 μg/ml. M9 minimal medium (Amresco) containing 1 mM MgSO₄ and 0.1 mM CaCl₂ was used, with an addition of 0.2% glucose.

Table 1.

Strains, phages, and plasmids used in this study^a

Strain, phage, or plasmid	Relevant characteristic(s)	Source
Strains
MG1655	F− λ−rph-1, wild-type E. coli K-12	Lab collection
ΔyafB strain	MG1655 ΔyafB::kan	CGSC
ΔQG strain	MG1655 ΔgloA::kan ΔyqhD::FRT	This work
ΔnsrR strain	MG1655 ΔnsrR::FRT	CGSC
Δfnr strain	MG1655 Δfnr::FRT	CGSC
ΔnsrR Δfnr strain	MG1655 ΔnsrR::FRT Δfnr::FRT	This work
ΔQ strain	MG1655 ΔyqhD::FRT	Lab collection
ΔQ ΔnsrR strain	MG1655 ΔyqhD::FRT ΔnsrR::FRT	This work
ΔQ Δfnr strain	MG1655 ΔyqhD::FRT Δfnr::FRT	This work
ΔQ ΔnsrR Δfnr strain	MG1655 ΔyqhD::FRT ΔnsrR::FRT Δfnr::FRT	This work
Mutant 1	ΔQG, Δ128 bp in yafB promoter	This work
Mutant 2	ΔQG, Δ268 bp in yafB promoter	This work
Mutant 3	ΔQG, Δ219 bp in yafB promoter	This work
Mutant 4	ΔQG, recombination between rrnD and rrnH	This work
Mutant 5	ΔQG, Δ4 bp in NsrR-binding site of yafB promoter	This work
BL21(DE3)	F−ompT hsdSB(rB−mB−) gal dcm(DE3)	Novagen
BL21(DE3) ΔnsrR Δfnr strain	BL21(DE3) ΔnsrR::FRT Δfnr::FRT	This work
Bacteriophages
λRZ5	Phage for recombination with pRS vectors	10
λnsrR-wt	λRZ5 with pRS-wt	This work
λnsrR-Δ	λRZ5 with pRS-nsrΔ	This work
TnphoA-132	λ::TnphoA-132, Tcr	32
Plasmids
pRS415	Apr, lacZ operon fusion plasmid	26
pRS-wt	pRS415, 163 bp, EcoRI+HindIII of wild-type yafB promoter	This work
pRS-nsrΔ	pRS415, 159 bp, Δ4 bp in the NsrR-binding site	This work
pET-YafB-His	pET21b::yafB	13
pNsrR	pET21a::nsrR	This work
pFnr	pET21a::fnr	This work
pBAD18	Apr, arabinose-inducible promoter	7
pBAD-YafB	pBAD18::yafB	This work
pCP20	FLP+ λ cI857+ λ pR Repts Apr Cmr	4

^aCGSC, E. coli Genetic Stock Center, Department of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, CT; Tc^r, tetracycline resistance; Ap^r, ampicillin resistance; Cm^r, chloramphenicol resistance; FRT, site for FLP recombinase (4).

Selection and mapping of GO-resistant mutations.

GO-resistant mutants were selected from the E. coli MG1655 ΔgloA ΔyqhD strain on LB plates containing 7 to 9 mM GO. Cells grown overnight at 37°C in LB medium were diluted 1:100 with fresh LB medium and cultured at 37°C until A₆₀₀ reached approximately 1.0, from which about 1 × 10⁷ cells were spread on LB plates containing GO. After 1 or 2 days of incubation, GO-resistant colonies appeared. The GO-resistant mutants were confirmed by replica plating and spotting onto LB plates containing different concentrations of GO (5, 7, and 9 mM). For mapping mutations, λ TnphoA-132 carrying the tet gene for tetracycline resistance was used for insertional mutagenesis (32). After infecting the GO-resistant mutant with phage, more than 10,000 independent clones with transposon insertions were obtained to generate a GO^r and TnphoA-132 pool. By infecting this pool with P1 phage, we obtained P1 phages containing both the GO^r mutation and TnphoA-132. These phages were used to identify insertions near the mutation of interest. In order to characterize the chromosomal region harboring the TnphoA-132 insertion, a fragment containing the TnphoA insertion junction was prepared by digesting with TaqI restriction enzyme. The self-ligated fragment made by T4 DNA ligase was amplified by TnpI (5′-GGGCTGCTCAGGGCGATATTACTGC-3′) and TnpO (5′-ACAGGGCAAAACGGGAAAGGTTCCG-3′) primers complementary to the sequence of TnphoA-132. DNA sequencing was performed with ABI 3100 (Perkin-Elmer). The results were analyzed with the BLAST program to search for the E. coli genome database. After determining the chromosomal locations of insertions, locations of the GO-resistant mutations were inferred from their cotransduction frequencies. The mutations were found by DNA sequencing. For amplifying and sequencing the upstream regions of yafB, we used yafBup-F (5′-CATGCGAGAGTAGGGAACT-3′) and yafBup-R (5′-ACCAAATGCAGGGATAGCCAT-3′), covering the 3′ end of rrfH to the 5′ end of the yafB coding sequence (CDS) (Fig. 4A).

Spotting assay and measurement of inhibitory concentration.

In order to compare levels of resistance, a spotting assay was carried out. Cells incubated overnight at 37°C in LB medium were diluted 1:100 with fresh LB medium and further cultured to cell densities of about 1.0 (A₆₀₀). Cells were diluted from 10⁻¹ to 10⁻⁶ with fresh LB medium, and 4 μl of each dilution was spotted onto LB plates containing different concentrations of GO. The plates were incubated for 16 h at 37°C and were photographed. To determine 50% inhibitory concentrations (IC₅₀), we measured optical densities (OD) of cells grown in media containing different concentrations of GO or MG. A 96-well culture plate containing 180 μl LB medium per well was added with various concentrations of GO or MG. Twenty microliters of culture at an A₆₀₀ of 1.0 was placed on each well. For efficient temperature equilibration, the 96-well plate contained in a sealed box was incubated for 8 h in a shaking water bath at 37°C. A₅₉₅ was measured using a microplate reader (model 680; Bio-Rad), from which IC₅₀ was estimated using Sigmaplot (SPSS Inc., Chicago, IL) by fitting to the sigmoidal dose-response (variable slope) equation.

Purification of YafB protein.

The BL21(DE3) strain transformed with pET21b-YafBhis (13) was grown in LB medium at 37°C with ampicillin (100 μg/ml) until OD₆₀₀ reached 0.5. The recombinant protein was expressed by the addition of 0.2 mM IPTG (isopropyl-d-thiogalactopyranoside; Duchefa) for 3 h. Cells were harvested by centrifugation and resuspended in a binding buffer (1 mg/ml lysozyme, 20 mM Tris-Cl [pH 7.9], 5 mM imidazole, and 500 mM NaCl). After disruption by sonication, cell debris was eliminated by centrifugation at 15,000 × g for 15 min, and the protein was purified by the standard procedure with His-Bind resin (Novagen).

Generation of polyclonal antibodies against YafB.

Five- to seven-week-old female BALB/c mice were immunized intraperitoneally with an emulsion of 50 μg YafB and adjuvant in phosphate-buffered saline (PBS). For 4 weeks after the first immunization, booster injections were made weekly. After 5 injections, polyclonal antiserum was obtained. Mice were bled by the retro-orbital plexus 1 week after final immunization, and the blood was incubated for 30 min at 37°C. The blood clot was removed by centrifugation, and the pooled sera collected from supernatants were stored at −70°C until use.

qRT-PCR.

The parent (MG1655 ΔgloA ΔyqhD) and mutant strains were cultured in LB medium at 37°C with agitation to an OD₆₀₀ of 1.0. RNAs were prepared by TRIzol reagent (Invitrogen), and 0.5 μg of total RNA from each mutant was used for making cDNA by reverse transcription. cDNA synthesis was carried out using ImPromII reverse transcriptase (Promega). Two microliters of each cDNA synthesized was used for quantitative real-time PCR (qRT-PCR). Reaction mixtures (20 μl) included SYBR green I, 1 unit of HS Taq DNA polymerase, 5 mM MgCl₂, 2 mM deoxynucleoside triphosphate (dNTPs) (Prime Q-Master mix), and 0.5 pmol of specific primers, which were used for amplification of all target and reference transcripts. The primers for yafB are yafB-F (5′-TGACGCGTGAGATCGGTATTTCCA-3′) and yafB-R (5′-ATATGGATGCCGTGCTGTTTAGCC-3′). For confirming read-through transcripts from the rrnH operon to yafB, we used the following forward primers with the yafBup-R primer as a set; rrl-F (5′-CTCGATGTCGGCTCATCACATCC-3′), rrf-F (5′-CAGAACGCAGAAGCGGT-3′), aspU-F (5′-GGTAGTTCAGTCGGTTAGAATACCTG-3′), and thrV-F (5′-GCTGATATGGCTCAGTTGGTAGAGCG-3′). For the reference gene, ompC-F (5′-CAGCTCTGCTGGTAGCAGGC-3′) and ompC-R (5′-GACCTGCGAATGCCACACGG-3′) were used. PCRs of 18 and 23 cycles were normally carried out for detecting transcripts, while in some cases more than 10 additional cycles were run for detecting smaller amounts of transcripts. CFX-96 (Bio-Rad) was used for PCR and for detection of fluorescence change. The values for the ratio of cycle threshold (C_T), as determined from the Bio-Rad CFX manager software, were compared between samples (24).

Construction of yafB and lacZ reporter plasmids and β-galactosidase assay.

The XbaI-XhoI fragment obtained from pET21b-YafBhis (13) was ligated to pBAD18 (7), digested with NheI and SalI with compatible ends. The plasmid constructed was used for complementation of yafB deletion from the wild-type strain. Expression of YafB in the presence of 0.2 mM l-arabinose was confirmed by 12% SDS-PAGE. The pRS415 plasmid was used to construct a transcriptional fusion for reporter assay (26). DNA fragments containing the putative yafB promoter (bp 229004 to 229167) of the parent strain and mutant 5 (4-bp deletion in the putative NsrR-binding site) were amplified and inserted into the EcoRI-BamHI site of the lacZ 5′ untranslated region (5′-UTR) of pRS415, carrying the complete lacZYA operon without the lac promoter. The reporter plasmids were introduced into MG1655 and other mutant strains with lacZ deletions.

To construct a single-copy version of the multicopy reporter, we recombined the reporter plasmid with λRZ5, followed by a lysogenization into the appropriate host as previously described (10). The bacteriophage λRZ5 contains a deletion of the promoter-proximal two-third region of lacZ (ΔlacZ_SC), the wild-type lacY gene, and the 5′-truncated bla gene (′bla) oriented opposite of lacZ. Thus, λRZ5 forms white plaques (Lac⁻) in the presence of X-Gal (5-bromo-4-chloro-3-indolyl galactoside) and exhibits the ampicillin-sensitive (Ap^s) phenotype. Because the lac-bla sequence of λRZ5 is homologous to the sequence in pRS415, a recombination can occur between the yafB promoter-LacZ fusion plasmid and bacteriophage DNA using the bla and lac homology, yielding a bacteriophage λ carrying a gene fusion (see Fig. 6A). To obtain such a recombinant, a host strain harboring the fusion reporter on a plasmid was infected with λRZ5 to prepare a lysate. Bacterial hosts infected with this lysate were plated on an LB plate containing X-Gal and ampicillin. All the lysogens containing transcriptional fusions showing the Lac⁺ Ap⁺ phenotype were restreaked and verified with PCR as a monolysogen (25). The established lysogens were kept for further analysis.

Fig 6.

Transcriptional repressions of yafB by NsrR and Fnr. (A) Construction of reporter phage λ for single-copy expression of lacZ. The plasmid-borne fusions were made first, and then they were crossed onto λRZ5 and inserted into the chromosome at the λ attachment site (attP) to generate strains carrying single-copy promoter fusions. The mutated version of the putative NsrR-binding site is indicated with an asterisk. Details are described in Materials and Methods. (B) The reporter phage containing nsrR-wt or nsrR-Δ were introduced into lacZ deletion strains; wild-type, ΔnsrR, Δfnr, or ΔnsrR Δfnr strains. (C) Transcription levels of yafB in Fnr and NsrR deletion mutants are elevated relative to that of the wild type. Derepression of yafB was higher in the double mutant (ΔnsrR Δfnr strain) than in single mutants. Exponentially growing cells in LB (OD₆₀₀ = 1.0) were treated for 15 min with 0, 1, and 3 mM GO. Total RNAs were used as templates for reverse transcription. The products were separated by 1.5% agarose gel electrophoresis, in which ompC was used as an internal reference. The RT-PCR data (top) were consistent with those of real-time qRT-PCR analysis (bottom).

The β-galactosidase assay was performed according to Miller (19). Cells were grown in LB to an OD₆₀₀ of 1.0. A total of 0.2 ml of each cell culture was suspended in Z-buffer (40 mM NaH₂PO₄, 60 mM Na₂HPO₄, 10 mM KCl, 1 mM MgSO₄, and 50 mM mercaptoethanol, pH 7.0) to a final 1 ml. One-hundred microliters of chloroform and 50 μl of 0.1% sodium dodecyl sulfate (SDS) were added and vortexed for 10 s. A total of 0.2 ml of o-nitrophenyl-β-d-galactoside (ONPG; 4 mg/ml) was added to the mixture, and the sample was vortexed again. Enzyme reactions were stopped by adding 0.5 ml of 1 M Na₂CO₃ when yellow color developed. After measuring OD₄₂₀ and OD₅₅₀, β-galactosidase activity was calculated according to Miller (19).

Measurement of YafB activities.

YafB activities of the GO-resistant mutants were measured at 25°C using a Beckman Coulter DU800 spectrophotometer by monitoring the initial rate at 340 nm with oxidation of NADPH. A total of 10 mM of MG and 200 mM of GO were used as the substrates for assay, with 10 μg of dialyzed soluble cell extracts. The standard assay for reducing aldehyde was carried out in 1 ml of 100 mM potassium phosphate buffer (pH 7.0) with 0.1 mM NADPH as a cofactor. Glyoxal was purchased from Sigma-Aldrich.

EMSA.

The nsrR and fnr genes were amplified from E. coli genomic DNA and subcloned into pET21a (Novagen) to construct pNsrR and pFnr plasmids, respectively. E. coli BL21(DE3) (Novagen) strains carrying nsrR and fnr mutations were transformed with pNsrR and pFnr, which were inoculated into LB medium with ampicillin (100 μg/ml) until the OD₆₀₀ reached 0.5. The recombinant protein was expressed by an addition of 0.2 mM IPTG (Duchefa) for 16 h at 20°C. Cells were harvested by centrifugation and resuspended in buffer (50 mM Tris-Cl [pH 7.9], 5 mM dithiothreitol [DTT], 0.5% Triton X-100, 10% glycerol, and 500 mM NaCl). After disruption of cells by several rounds of freeze and thaw, cell debris was eliminated by centrifugation at 15,000 × g for 15 min. The soluble fractions of crude extract were concentrated and used for electrophoretic mobility shift assay (EMSA). The probe DNA, the yafB promoter region (bp 229004 to 229167) prepared by PCR, was end labeled with radioactive [γ-³²P]ATP (GE Healthcare) and mixed with cell extract from the BL21(DE3) ΔnsrR Δfnr strain expressing NsrR (pNsrR) or Fnr (pFnr). The extract from the BL21(DE3) ΔnsrR Δfnr strain transformed with empty vector was used as a negative control. The reaction mixtures were incubated at 37°C for 30 min, run on an 8% low-ionic-strength polyacrylamide gel, and electrophoresed at 4°C in Tris-borate-EDTA buffer for 16 h.

RESULTS

Selection and localization of GO-resistant mutations.

Previously, we isolated a number of GO-resistant mutants overexpressing YqhD from the E. coli MG1655 strain on LB plates containing lethal concentrations of GO (14). In this study, we used the strain deficient in GO-associated genes (gloA, yqhD) to screen for novel genes. The mutants were isolated from this strain (MG1655 ΔgloA ΔyqhD strain) by growing on LB plates containing 7 to 9 mM GO. The GO-resistant phenotypes were confirmed by replica plating and by spotting onto fresh plates containing GO. By doing so, 40 isolates reproducibly showing GO-resistant phenotypes were obtained. When we examined total cellular proteins on a 12% SDS-PAGE gel, 29 isolates (72.5%) showed protein overproduction at ca. the 30-kDa position, as shown in Fig. 2A, which was identified as YafB (aldo-keto reductase of 29.5 kDa) by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) fingerprinting. Based on their levels of expressions and patterns of genomic rearrangement probed by PCR (see Materials and Methods for details), we categorized them into five different subtypes, represented by mutants 1 to 5 (Table 2). The rest were uncharacterized.

Fig 2.

YafB expressions in the GO-resistant mutants. (A) The GO-resistant mutants show an expression of a 30-kDa protein, which was identified as YafB (*, 29.5 kDa, an aldo-keto reductase) by MALDI-TOF MS fingerprinting. Expression of YafB is highest in mutant 1. (B) Western blotting with specific antiserum for YafB; (C) results of a quantitative real-time PCR experiment showing yafB transcription in various mutants; (D) specific activities of YafB enzymes in the mutants. The data were standardized with the wild-type value as 1. Rough correlation between the data in panels C and D exists. Detailed procedures are described in Materials and Methods.

Table 2.

Summary of the glyoxal-resistant mutations

Strain	Location	Size of PCR product (bp)a	Mutation type	No. of isolates (%)
Wild type	228846-229173	328
Mutant 1	229018-229145	200	Multibase deletion (Δ128 bp)	18 (62.1)
Mutant 2	228856-229123	60	Multibase deletion (Δ268 bp)	4 (13.8)
Mutant 3	228928-229146	111	Multibase deletion (Δ219 bp)	5 (17.2)
Mutant 4	–b	318	Recombination between rrnD and rrnH	1 (3.5)
Mutant 5	229086-229089	324	Multibase deletion (Δ4 bp)	1 (3.5)

^aFor amplifying and sequencing the upstream regions of yafB, the primer set covering the 3′ end of rrfH to the 5′ end of the yafB CDS were used (see Fig. 4A).

^b–, see Fig. 4 for details.

Phenotypes of YafB-overexpressing mutants were tested by spotting cells onto an LB plate containing GO (Fig. 3A). All the mutants exhibit higher resistances to glyoxal than the wild-type MG1655, except for mutant 5, which is apparently more resistant than the parent strain (MG1655 ΔgloA ΔyqhD strain) but not as good as the wild type (Fig. 3B). Degrees of GO resistance were further compared by measuring IC₅₀s of the mutant strains, which were monitored by growing cells in media containing different concentrations of GO and determining their optical densities (Fig. 3B). After phenotypic characterization, the mutations were mapped to the yafB region by chromosomal mapping with transposon tagging, followed by phage P1 transduction. TnphoA-132 (Tc^r) was used to isolate insertions near the GO-resistant mutations (see Materials and Methods), whose locations were determined by inverse PCR, followed by an analysis of cotransductional linkages. The candidate locus was found at approximately 4.9 min on the E. coli linkage map, which was deduced from cotransduction frequencies to insertions in degP (31%), yaeF (71%), gloB (95%), and yafP (46%), indicating that the mutations are closely linked to yafB.

Fig 3.

Phenotypes of GO-resistant mutants. (A) Results of spotting assays on LB plates containing various concentrations of GO. Mutant 1 exhibits normal growth on medium containing 7 mM GO, which is lethal to not only the MG1655 ΔgloA ΔyqhD strain but also the wild-type MG1655 strain. (B) Growth inhibition in various concentrations of GO. IC₅₀s (mM) are 3.23 (MG1655, wild type), 1.42 (ΔQG strain, parent), 8.55 (mutant 1), 4.41 (mutant 2), 4.99 (mutant 3), 6.39 (mutant 4), and 1.96 (mutant 5).

We further analyzed the relationship between GO resistances and expression levels of yafB by carrying out a Western blotting experiment with antiserum raised against YafB protein (Fig. 2B) as well as qRT-PCR for yafB transcription (Fig. 2C). The result of protein expression on SDS-PAGE was consistent with the levels of proteins/transcripts by Western blotting and qRT-PCR. Further analysis of YafB enzyme activity in crude extracts again confirms this relationship (Fig. 2D).

The mutations were found in the promoter region of the yafB gene.

In order to identify the mutational changes responsible for yafB expressions, we first carried out PCR amplifications and sequenced yafB, including its promoter with the primer sets shown in Fig. 4A (covering the 3′ end of rrfH to the 5′ end of the yafB CDS, bp 229004 to 229167), from which we found that the PCR products of YafB overproducers were smaller than those of the parent strain (328 bp; Table 2). By sequencing the amplified DNAs, we were able to reveal that multibase deletions and recombination occurred in the promoter region of yafB (Fig. 4A and B; Table 2). No sequence alteration was found in the coding region of yafB. In mutant 1, the yafB promoter was removed, along with the putative terminator of aspU (128 bp; bp 229018 to 229145), leaving the anti-antiterminator intact. The 3′ deletion ends 21 bp upstream of the initiator ATG of yafB. Therefore, the transcriptional fusion created by deleting the yafB promoter and the putative aspU terminator (Fig. 4A) is expected to enhance yafB transcription, presumably as part of the rRNA and aspU operon. In the case of mutant 2, the 3′ end of deletion locates at 43 bp upstream of the initiator ATG of yafB (268 bp; bp 228856 to 229123). Because the 5′ end of deletion starts in the rrfH gene, a transcriptional fusion is also created (Fig. 4A and B; Table 2), leading to an expression of yafB under the upstream rrlH promoter. Mutant 3 has a deletion of 219 bp, starting from the first base of the aspU gene to 20 bp upstream of the yafB ATG (bp 228928 to 229146), shifting a single base pair from that of mutant 1 (Fig. 4A and B; Table 2). Here, the aspU gene and its terminator (bp 229004 to 229055; Fig. 4A) with anti-antiterminator (p 228967 to 229003; Fig. 4A), possibly attenuating the transcription of yafB (http://cmgm.stanford.edu/∼merino/), are completely removed.

Interestingly, mutant 4 underwent a genomic recombination involving the rrnD (73.8 min) and rrnH (4.8 min) operons, located far apart on the E. coli chromosome in opposite orientation with respect to the position of oriC (84.5 min). As shown in Fig. 4C, the rrnH operon is homologous to rrnD, so that a recombination may occur. A set of oligonucleotides recognizing the rrn and yafB genes was used to amplify and sequence the recombined region to reveal its boundaries, which are located in the rrsH and rrsD genes on the 5′ side and in the upstream regions of the yafB and yhdZ genes (Fig. 4C). Mutant 5 has a 4-bp (GCAT) deletion near the transcriptional start (+1) site of yafB (17). The sequence of this region was predicted as an NsrR-binding site, located 86 bp upstream of the yafB open reading frame. It is likely that a removal of this site results in derepression of yafB (Fig. 2A and B). Since the deletion has shortened the distance between the promoter and the transcription initiation site, a shift in transcription initiation might occur.

yafB transcriptions in the GO-resistant mutants are regulated by upstream promoters.

In order to confirm the transcriptional fusions involved with yafB in GO-resistant mutants, we carried out RT-PCR using primers recognizing yafB and its upstream genes (Fig. 4C and D). In the wild type, small amounts of read-through transcripts synthesized from the upstream promoter exist, as detected with the primers recognizing rrlH (primer 1), rrfH (primer 2), and aspU (primer 3), which were visualized by additional cycles of PCR (Fig. 4D, left). Although the transcription of the rrn operon is known to be terminated at the aspU terminator, some read-through messages extended to yafB appear to be allowed. In the cases of GO-resistant mutants, fusion transcripts from the upstream genes were produced (Fig. 4D). However, mutants 2 and 3 did not synthesize transcripts from the aspU gene (lanes 2-3 and 3-3), presumably due to a deletion of aspU (Fig. 4D). The data indicate that all the GO-resistant mutants, except for mutant 5 (containing an NsrR-binding site mutation), produce the fusion transcripts by the upstream promoters of the rrn operon, which are responsible for the GO-resistant phenotypes. Because the transcription of the rRNA operon is reduced during the stationary phase (6), we expected that the transcriptions of yafB fusions might be decreased when cells enter the stationary phase. As a matter of fact, we observed growth-rate-dependent changes in yafB expressions in the GO-resistant mutants, which were monitored by 12% SDS-PAGE (data not shown).

In order to further assess physiological relevance of YafB in GO detoxification, we constructed strains either lacking or overexpressing YafB from the wild-type MG1655. As shown in Fig. 5, the yafB deletion renders the wild-type E. coli more sensitive to glyoxal. When we introduced pBAD18 encoding C-terminally His-tagged YafB under the control of arabinose-inducible promoter (P_ara), the strain became highly resistant to GO in the presence of 0.2% arabinose. We also observed that mutation 1 in the MG1655 background exhibits the highest resistance to glyoxal.

Fig 5.

Glyoxal sensitivities of the strains with various levels of YafB expression. Strains tested were transformed with pBAD18 (vector control) or pBAD-YafB (+, arabinose-inducible yafB) and were spotted onto LB plates containing 0.2 mM arabinose. Various levels of glyoxal resistances were observed with the deletion and overproduction of the ΔyafB strain in the wild-type MG1655 background. Mutation 1 shows the highest GO resistance when transferred to the wild-type MG1655 strain.

Regulations of yafB expression by NsrR and Fnr.

We found that mutant 5 has a deletion of 4 bp in the putative half-site for NsrR binding (Fig. 4A and B). In order to investigate whether the change in the NsrR-binding site affects expression of yafB, a series of mutants were constructed in bacteriophage λ carrying a transcriptional fusion with reporter, from which a β-galactosidase assay was carried out (Fig. 6B). Compared to the wild type, yafB expression was increased about 1.5-fold in the NsrR-binding site mutant. The fact that transcription from the wild-type promoter is increased about 2-fold in the nsrR mutant indicates that NsrR may not act as a sole regulator interacting with the yafB promoter. It was also found that an Fnr-binding site exists upstream of yafB, overlapping with the transcriptional start site (+1; Fig. 6A) (17). The expression of yafB was increased, relative to that of the wild type, about 1.3-fold in the fnr mutant and more than 2-fold when both repressors were deleted. Because Fnr is likely to be involved in the transcription of yafB, we performed RT-PCR to examine yafB transcriptions in the strains lacking Fnr, NsrR, or both. As shown in Fig. 6C, levels of yafB transcriptions in the Fnr and NsrR mutants were elevated compared to that of the wild type, which is more apparent in the double mutant. The level of yafB transcript in the Fnr mutant is higher than that of the wild type but lower than that of the NsrR mutant (Fig. 6C, top). We also tested whether transcriptional enhancement of yafB by a mutation in the repressor gene confers GO resistance. We observed susceptibility of the mutant strains to GO by spotting and measuring IC₅₀ values, which are higher than that of the wild type (data not shown). The values of IC₅₀ (mM) are 1.49 (ΔyqhD strain, parent), 2.31 (ΔnsrR strain), 2.46 (Δfnr strain), and 2.41 (ΔnsrR Δfnr strain). These data support the involvements of nsrR and fnr as negative regulators of yafB.

Interestingly, the transcription of yafB is increased proportionally to GO concentrations. Although these data were consistent with the results of the β-galactosidase assay, it should be noted that the differences observed at a high GO concentration must be underestimated, since the activity of β-galactosidase is considerably inhibited by increased GO concentration, presumably due to its active-site residues reacting with GO. To further confirm these results, we carried out a qRT-PCR experiment, in which the difference was more apparent (Fig. 6C, bottom).

Interaction of NsrR and Fnr with the yafB promoter.

Based on the above-described results in which we found a mutation in the NsrR-binding site of the yafB promoter and that the transcription of yafB was enhanced in the nsrR-null mutant, we speculated that the NsrR repressor can directly bind to the putative binding site of the yafB promoter. In addition, an increase in yafB expression was also observed in the fnr deletion mutant, which was supported by the presence of a putative regulatory sequence. In order to obtain more direct evidence, we carried out EMSA with the yafB promoter as a probe. As shown in Fig. 7, complexes of DNA with NsrR and Fnr were formed proportional to the amounts of extract obtained from the strains expressing NsrR (pNsrR) or Fnr (pFnr), whereas the negative control (ΔnsrR Δfnr strain) does not indicate any binding.

Fig 7.

Binding of NsrR and Fnr to the yafB promoter. Electrophoretic mobility shift assay was carried out with C-terminally His-tagged NsrR and Fnr. The probe DNA labeled with [γ-³²P]ATP was incubated with freshly prepared cell extracts from BL21(DE3) ΔnsrR Δfnr strains expressing NsrR, Fnr, or the empty vector pET21a. The yafB promoter region obtained by PCR from genomic DNA was used as a probe. The amount of cell extract is shown with the graded triangle on top. The reaction mixtures were incubated at 37°C for 30 min and run on polyacrylamide electrophoresis image gels for 16 h at 4°C. The NsrR-DNA complex and the free DNA probe are marked with an asterisk and an arrow, respectively.

DISCUSSION

Role of YafB in scavenging glyoxal.

Previously, we reported that AKRs participate in stress tolerance by detoxifying reactive aldehyde species (13), in which four E. coli AKRs (YafB, YqhE, YeaE, and YghZ) were shown to have activity to methylglyoxal. Although YqhE and YafB exhibited considerably higher enzyme activities to methylglyoxal than the others, their levels of expressions are varied. The expression of YafB is moderate but considerably lower than that of YghZ (13). Thus, it is likely that under normal conditions, the YafB expression is to some extent repressed, which can be relieved by a mutation to confer glyoxal resistance. We obtained cis-acting changes in the upstream region of yafB for its expression. In the case of the yqhC activator, mutations occurred in the trans-acting regulator constitutively expressing YqhD reductase, thereby conferring glyoxal resistance (14). For some reason, we were unable to isolate such a mutation for the yafB gene.

Our data indicate that the transcription of yafB in the wild-type strain is independent of the transcription of the rrsH operon (Fig. 4D). However, some amounts of read-through messages were detected, presumably from the promoter of aspU (Fig. 4D). This may explain why an overexpression of yafB is achieved by deleting the putative rho-independent terminator located between the aspU and yafB genes. The rrnD operon has a transcription termination site located at about 25 bp distal to the 3′ end of the second 5S rRNA gene (rrfF). This terminator was excluded during the recombination between the rrnH and rrnD operons to create mutant 4 (Fig. 4C), resulting in an overexpression of yafB.

The uncommon recombination of rrnD and rrnH occurred.

It has been reported that rrnD can recombine with the rrnB or rrnE operon, located in opposite orientation with respect to oriC (Fig. 4C). Likewise, rrnH recombines with rrnG (8). Because of an asymmetrical rearrangement of the genomic regions involved, a recombination between the rrnD and rrnH operons was not reported. Therefore, the change in mutant 4 appears to involve a duplication of rrnD. Since an inactivation of one or two rRNA operons did not affect cell proliferation due to compensatory transcriptions from other rRNAs (1), a significant difference in doubling time between wild-type and mutant strains was not observed in LB or M9 minimal media (data not shown). In addition, the tRNA genes contained in the rrnH operon (ileV, alaV, and aspU) have identical copies in rrnD (ileU and alaU) or in other chromosomal regions (aspT and aspV), so there would not be any effect on cell growth. An occurrence of rare recombination in mutant 4 might partly be explained by the presence of homology between rrsD and rrsH. However, the recombination event at the 3′ end is not readily understood. Since this region is rich in repeat sequences, recombination may frequently occur (9). As a matter of fact, the recombination breakpoint is located in the stem and loop structure for transcription termination (yhdZ), which is also the case in mutant 1 (Fig. 4C).

Regulators for yafB expression.

Mutant 5 has a 4-bp deletion in the putative NsrR-binding site, leading to a derepression of yafB (Fig. 6B). This suggests that the NsrR-binding site may serve as a regulatory sequence for NsrR. By sequence analysis, we found a putative Fnr-binding site located adjacent to the putative NsrR-binding site (cTctT N4 cTCAA; lowercase letters represent nonconserved bases) (Fig. 4A). The reporter assay and RT-PCR data show considerable derepression of yafB in both NsrR and Fnr mutants, supporting the role of the Fnr-binding site in the expression of yafB (Fig. 6B and C). Although we were unable to isolate mutations in the repressor genes, it is likely that these repressors are indeed functional. There has been evidence that a redox imbalance caused by reactive oxygen species (ROS) or reactive nitrogen species (RNS) generates toxic aldehydes, including glyoxal (16). The global regulators NsrR and Fnr, sensing ROS and RNS, respectively, may provide an organism with an ability to remove reactive electrophilic species (RES) by transcriptionally derepressing the yafB gene.