Influence of Operator Site Geometry on Transcriptional Control by the YefM-YoeB Toxin-Antitoxin Complex

Abstract

YefM-YoeB is among the most prevalent and well-characterized toxin-antitoxin complexes. YoeB toxin is an endoribonuclease whose activity is inhibited by YefM antitoxin. The regions 5′ of yefM-yoeB in diverse bacteria possess conserved sequence motifs that mediate transcriptional autorepression. The yefM-yoeB operator site arrangement is exemplified in Escherichia coli: a pair of palindromes with core hexamer motifs and a center-to-center distance of 12 bp overlap the yefM-yoeB promoter. YefM is an autorepressor that initially recognizes a long palindrome containing the core hexamer, followed by binding to a short repeat. YoeB corepressor greatly enhances the YefM-operator interaction. Scanning mutagenesis demonstrated that the short repeat is crucial for correct interaction of YefM-YoeB with the operator site in vivo and in vitro. Moreover, altering the relative positions of the two palindromes on the DNA helix abrogated YefM-YoeB cooperative interactions with the repeats: complex binding to the long repeat was maintained but was perturbed to the short repeat. Although YefM lacks a canonical DNA binding motif, dual conserved arginine residues embedded in a basic patch of the protein are crucial for operator recognition. Deciphering the molecular basis of toxin-antitoxin transcriptional control will provide key insights into toxin-antitoxin activation and function.

Toxin-antitoxin (TA) loci are widely distributed among eubacteria and archaea, with many species possessing tens of TA cassettes that can be grouped into distinct evolutionary families (14, 16, 17, 22, 44). The most prevalent TA complexes comprise two small proteins, a stable toxin and a labile antitoxin, that associate tightly so that the toxin remains inert. Liberation of the toxin from the complex permits it to target and impair a vital cellular function, at least under experimental conditions. Many toxins recently have been shown to be site-specific endoribonucleases. The RelE RNase toxin does not degrade free RNA but instead cleaves mRNA in the ribosomal A site with high codon specificity (46). Numerous other toxins are also site-specific RNases with a variety of recognition sites but act on free RNA (6-9, 21, 39, 57, 58, 60). Other toxins have distinct modes of action, either poisoning DNA gyrase (11, 18) or inhibiting translation elongation (30).

Whereas plasmid-encoded TAs assist in stable plasmid maintenance (17), the physiological role, if any, of chromosomally specified TA complexes has yet to be fully deciphered (33, 54). Competing evidence suggests that the systems induce either reversible cell cycle arrest or programmed cell death in response to nutrient deprivation or other adverse conditions (1, 45, 54). Roles for TAs as protection against invading mobile elements that specify related TA complexes (49), as regulators of multicellular development (41), and as part of a quorum-sensing relay (25) also have been proposed recently. The uncertainty in assigning biological roles to TAs stems in part from observations that deletion of chromosomal TA genes often elicits no obvious phenotype (54), although it has been shown that TA genes that originated from a superintegron of Vibrio sp. can stabilize Escherichia coli chromosomal regions into which the genes were introduced (52).

Pomerantsev et al. (47) first suggested that the yefM-yoeB locus of E. coli may encode a TA complex based on sequence similarity of YefM to the Phd antitoxin of plasmid P1. In a parallel study, ectopic expression of YoeB was shown to be toxic to E. coli, but YefM counteracted this toxicity (15). Accordingly, the two proteins form a complex (5, 19, 23), although the physiological trigger that releases YoeB from this complex remains elusive (54). YoeB is a purine-specific endoribonuclease that cleaves translated mRNA (6, 19). This cleavage is exacerbated by overproduction of the Lon protease, apparently by indirect inhibition of translation of further YefM antitoxin (6). YefM originally was described as a natively unstructured protein (4), but this conclusion was subsequently reevaluated as experimental, and modeling data have demonstrated that the protein is at least partially folded (23, 42) and dimeric (23).

The tertiary structures of the YoeB toxin and the YefM₂-YoeB complex have recently been described (19). One C terminus in the YefM homodimer is unfolded within the complex, whereas the second C terminus shows an α-helical conformation and conceals the endoribonuclease fold of YoeB. The N-terminal segments of YefM form a symmetrical dimer within the YefM₂-YoeB complex and do not contact YoeB directly. The three residues at the C-terminal end of YoeB that form part of the catalytic fold are distributed into a less favorable configuration when in complex with the YefM dimer, partly explaining the mechanism by which the antitoxin blocks the endoribonucleolytic activity of YoeB (19).

The genes for most TA complexes are autoregulated with the antitoxin acting as a transcriptional repressor and the toxin as a corepressor. This is also the case for YefM-YoeB: the antitoxin recognizes short (S) and long (L) palindromes, the latter covering the −10 promoter box of the yefM-yoeB cassette, and partially represses expression (23). The palindromes possess common 5′-TGTACA-3′ core motifs (Fig. 1A). YefM initially binds the L repeat in the operator site, followed by recognition of the S hexamer. YoeB enhances the interaction of YefM with the operator, both in vivo and in vitro, thereby reducing expression of a transcriptional fusion to a basal level. Paired 5′-TGTACA-3′ motifs are common in yefM-yoeB regulatory regions in the genomes of diverse bacteria, indicating that interaction of YefM-YoeB with these motifs is a conserved mechanism of operon autoregulation (23). Each N-terminal domain of YefM within the YefM₂-YoeB crystal structure possesses a conserved basic patch below the symmetrical dimer interface. This symmetrical underside surface of the antitoxin has been suggested to serve as the primary DNA anchor for DNA binding (19). Here, the assembly of the YefM-YoeB complex on its operator site is probed: the S hexamer fulfils a crucial role in this assembly. Synergistic interactions between YefM-YoeB complexes bound to the L and S repeats are highly dependent on their correct relative positioning on the DNA helix. Moreover, conserved residues in the YefM protein that mediate the YefM-YoeB-DNA interaction are identified. Understanding the molecular basis of yefM-yoeB transcriptional control ultimately may lead to the identification of artificial ligands that disrupt this interaction, resulting in transcriptional deregulation of the cassette. In this way, excess YoeB toxin may elicit bacterial cell death, a novel antibacterial strategy.

FIG. 1.

Organization of the yefM-yoeB regulatory region and characterization of operator site mutations in vivo. (A) Nucleotide sequence of the yefM-yoeB regulatory region. The transcription start site (+1) is indicated by the filled dot, with −10 and −35 promoter motifs boxed (23). The L and S palindromes are denoted by open, horizontal arrows. The translation start codon of yefM is in boldface. Insertions of 1, 2, 3, and 6 bp between the L and S palindromes are shown above the sequence; single-base-pair substitution mutations in the S repeat are shown below the sequence. (B) Effect of single-base-pair substitution mutations in the S repeat on transcriptional repression by YefM-YoeB. Transcriptional fusions of the yefM-yoeB regulatory region to the lacZYA operon in plasmid pRS415 were transformed into E. coli SC301467, which is deleted of five chromosomal TA cassettes including yefM-yoeB. β-Galactosidase levels were determined with (open bars) and without (shaded bars) YefM-YoeB supplied in trans from the pBAD33 arabinose-inducible vector (pBADyefMyoeB). The transcriptional fusions contained either the wild-type (WT) sequence or single-base-pair substitution mutations in the S repeat shown in panel A. The mutations are numbered relative to the +1 transcription start site. The upper panel shows repression levels relative to wild-type. (C) Effect of insertion mutations between the L and S repeats on transcriptional repression by YefM-YoeB. Experiments were performed as described for panel B, except that transcriptional fusions of the yefM-yoeB regulatory region to the lacZYA operon in plasmid pRS415 contained either the wild-type (WT) sequence or the insertions of 1, 2, 3 or 6 bp between the L and S palindromes shown in panel A.

MATERIALS AND METHODS

Bacterial strains.

E. coli DH5α (endA1 hsdR17 (r_K⁻ m_K⁻) supE44 thi1 recA1 gyrA96 (Nal^r) relA1 Δ(lacZYA-argF)U169 deoR [φ80dLacΔ(lacZ)] (56) was used for plasmid propagation and cloning, BL21(DE3) (F⁻ ompT gal dcm lon hsdSB(r_B⁻ m_B⁻) λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5]) was used for recombinant YefM and YefM-YoeB overproduction, and SC301467 (MG1655 ΔmazF ΔchpB ΔrelBE Δ(dinJ-yafQ) Δ(yefM-yoeB)) (6) was used for β-galactosidase assays. Bacteria were grown in LB medium at 37°C. Antibiotics were added at final concentrations of 100 μg/ml (ampicillin) and 10 μg/ml (chloramphenicol).

Oligonucleotides.

Oligonucleotides used in the present study are listed in Table 1.

TABLE 1.

Oligonucleotides used in this study

Oligonucleotide	Sequence (5′-3′)a
1	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAAcCTGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
2	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACAGgTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
3	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAActCTGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
4	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACAGagTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
5	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAActgCTGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
6	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACAGcagTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
7	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAActgatgCTGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
8	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACAGcatcagTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
9	TAACGCTCATCATTGTACAATGAACTGTACAAAAGAGGAGATTGACATG
10	CATGTCAATCTCCTCTTTTGTACAGTTCATTGTACAATGATGAGCGTTA
11	TCGATGCATCAGAATAGATAGATAGCTAAGTTATCAAGATCATACAGGC
12	GCCTGTATGATCTTGATAACTTAGCTATCTATCTATTCTGATGCATCGA
13	GCTACAGCGAAGCGgcgCAGAATTTGTCG
14	CGACAAATTCTGcgcCGCTTCGCTGTAGC
15	CGATCCTTATTACTgcgCAGAATGGAGAGGC
16	GCCTCTCCATTCTGcgcAGTAATAAGGATCG
17	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACgGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
18	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACcGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
19	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTaTACAAAAGAGGAGATTGACATGGGATCCAGTGC
20	GCACTGGATCCCATGTCAATCTCCTCTTTTGTAtAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
21	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTGgACAAAAGAGGAGATTGACATGGGATCCAGTGC
22	GCACTGGATCCCATGTCAATCTCCTCTTTTGTcCAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
23	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTGTcCAAAAGAGGAGATTGACATGGGATCCAGTGC
24	GCACTGGATCCCATGTCAATCTCCTCTTTTGgACAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
25	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTGTAtAAAAGAGGAGATTGACATGGGATCCAGTGC
26	GCACTGGATCCCATGTCAATCTCCTCTTTTaTACAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
27	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTGTACcAAAGAGGAGATTGACATGGGATCCAGTGC
28	GCACTGGATCCCATGTCAATCTCCTCTTTgGTACAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
29	GTCGAGAATTCTCTACAAACTAATTAATAAATAGTTAATTAACGCTCATCATTGTACAATGAACTGTACAAAAGAGGAGATTGACATGGGATCCAGTGC
30	GCACTGGATCCCATGTCAATCTCCTCTTTTGTACAGTTCATTGTACAATGATGAGCGTTAATTAACTATTTATTAATTAGTTTGTAGAGAATTCTCGAC
31	TAACGCTCATCATTGTACAATGAACTGTACAAAAGAGGAGATTGACATG
32	CATGTCAATCTCCTCTTTTGTACAGTTCATTGTACAATGATGAGCGTTA
33	TAACGCTCATCATTGTACAATGAAcCTGTACAAAAGAGGAGATTGACATG
34	CATGTCAATCTCCTCTTTTGTACAGgTTCATTGTACAATGATGAGCGTTA
35	TAACGCTCATCATTGTACAATGAActgCTGTACAAAAGAGGAGATTGACATG
36	CATGTCAATCTCCTCTTTTGTACAGcagTTCATTGTACAATGATGAGCGTTA
37	TCGATGCATCAGAATAGATAGATAGCTAAGTTATCAAGATCATACAGGC
38	GCCTGTATGATCTTGATAACTTAGCTATCTATCTATTCTGATGCATCGA
39	TCCTGAGTAGGACAAATCCG
40	GGTCATAGCTGTTTCCTGTG
41	TAACGCTCATCATTGTACAATGAAcCTGTACAAAAGAGGAGATTGACATG
42	CATGTCAATCTCCTCTTTTGTACAGgTTCATTGTACAATGATGAGCGTTA
43	TAACGCTCATCATTGTACAATGAActgCTGTACAAAAGAGGAGATTGACATG
44	CATGTCAATCTCCTCTTTTGTACAGcagTTCATTGTACAATGATGAGCGTTA

^aRestriction enzyme recognition sites are in boldface, mutations from the wild-type sequence are in lowercase, the 5′-TGTACA-3′ hexamers in the L and S repeats are underlined, and the translation start codon of yefM is in italics.

Plasmids.

Plasmid pET22yefMyoeB consists of the yefM-yoeB cassette cloned between NdeI and XhoI restriction enzyme sites in the pET-22b(+) overexpression vector (Novagen) and was used to produce the YefM-YoeB complex in which YoeB is C-terminally tagged with a His₆ motif (23). Derivatives of pET22yefMyoeB producing YefM-YoeB complexes in which R10 or R31 of YefM were mutated to alanine were generated by site-directed mutagenesis using oligonucleotides 13 and 14 (13/14) and 15/16, respectively (Table 1). Plasmid pRSyy_wt consists of the yefM-yoeB promoter-operator region cloned as annealed oligonucleotides between BamHI and EcoRI restriction sites of pRS415. The lacZ operon in pRS415yy_wt is expressed from the yefM-yoeB promoter-operator region and is repressed by YefM-YoeB supplied in trans (23). Versions of pRSyy_wt that contain 1-, 2-, 3-, or 6-bp insertions between the L and S repeats in the yefM-yoeB operator were produced similarly using oligonucleotides 1/2, 3/4, 5/6, and 7/8, respectively, or which contain the −1T→G, +1G→A, +2T→G, +3A→C, +4C→T, or +5A→C substitution mutations in the S repeat (see Fig. 1A) using oligonucleotides 17/18, 19/20, 21/22, 23/24, 25/26, and 27/28, respectively (Table 1). Plasmid pBADyefMyoeB consists of yefM-yoeB amplified by PCR, digested with XhoI-HindII, and cloned between the equivalent sites in the arabinose-inducible vector, pBAD33 (23). Derivatives of pBADyefMyoeB producing YefM-YoeB complexes in which R10 or R31 of YefM were mutated to alanine were generated by site-directed mutagenesis using oligonucleotides 13/14 and 15/16, respectively (Table 1). The nucleotide sequences of all mutated genes or promoter-operator regions were verified.

Protein production and purification.

The YefM-YoeB complex was overproduced and purified as detailed previously (23).

EMSA.

Electrophoretic mobility shift assay (EMSA) was performed as outlined elsewhere (23). Briefly, wild-type DNA substrates consisted of 5′-biotinylated, double-stranded oligonucleotides 29/30 (Table 1) that included the yefM start codon and the yefM-yoeB regulatory region. Oligonucleotides 1/2, 3/4, 5/6, and 7/8 consist of the same sequence, but with 1-, 2-, 3-, or 6-bp insertions, respectively, between the L and S repeats in the yefM-yoeB operator. Oligonucleotides 17/18, 19/20, 21/22, 23/24, 25/26, and 27/28 harbor the −1T→G, +1G→A, +2T→G, +3A→C, +4C→T, or +5A→C substitution mutations, respectively, in the S repeat (see Fig. 1A). Reactions containing 0.1 nM biotin-labeled DNA and the protein concentrations indicated in the figure legends were assembled in binding buffer [10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM dithiothreitol, 5 mM MgCl₂, 1 μg of poly(dI-dC), 2.5% glycerol] in final volumes of 20 μl and incubated for 20 min at 22°C. For competition experiments, binding reactions containing 0.8 μM YefM-YoeB and 0.3 nM concentrations of the 99-bp biotinylated oligonucleotides 29/30 harboring the wild-type sequence (Table 1) were incubated at 22°C for 15 min. Increasing amounts (up to 2.4 μM) of unlabeled 49-bp competitor oligonucleotides were subsequently added, and incubation continued for an additional 15 min. The competitor DNAs contained the wild-type sequence (oligonucleotides 31/32) or the same sequence with 1- or 3-bp insertions between the L and S repeats (oligonucleotides 33/34 and 35/36, respectively). A nonspecific competitor oligonucleotide had the same length and base composition as wild-type competitor DNA, but the sequence was randomized (oligonucleotides 37/38) (Table 1). Samples were electrophoresed on 8% native polyacrylamide gels in 0.5× Tris-borate-EDTA buffer for 90 min at 70 V at 22°C. DNA was transferred by capillary action or electroblotting to positively charged nylon membranes (Roche), and the transferred DNA fragments were immobilized onto the membrane by UV cross-linking. Detection of the biotin end-labeled DNA was performed using a LightShift chemiluminescent EMSA kit (Pierce) (2).

DNase I footprinting.

A 300-bp PCR fragment in which the top strand (Fig. 1A) was 5′ biotinylated was generated with oligonucleotides 39/40 using pRSyy_wt as a template for the wild-type promoter-operator region. The same plasmid, but with 1-, 2-, 3-, or 6-bp insertions between the L and S repeats in the yefM-yoeB operator, was used for amplification of mutated fragments. Footprinting reactions and Maxam-Gilbert sequencing reactions were performed and processed as described by Kȩdzierska et al. (23).

Assays of β-galactosidase activity.

Strain SC301467 harboring the pRS415 plasmid with a lacZ gene under transcriptional control of the yefM-yoeB promoter (pRSyy_wt), with pRSyy_wt possessing 1-, 2-, 3-, or 6-bp insertions between the L and S repeats, or with single base-pair substitution mutations in the S repeat, was cotransformed with pBAD33 encoding yefM-yoeB under the control of an arabinose-inducible promoter (pBADyefMyoeB). At an optical density at 600 nm of ∼0.2, toxin-antitoxin expression was induced by addition of 0.2% arabinose for 1 h. β-Galactosidase assays were performed with cells permeabilized with chloroform and sodium dodecyl sulfate as described by Miller (37).

SPR.

Surface plasmon resonance (SPR) measurements used a Biacore 3000 instrument (Biacore AB) primed with immobilization buffer (10 mM Tris-HCl [pH 7.5], 200 mM NaCl, 0.1 mM EDTA). Streptavidin (SA) chips (Biacore AB) were activated with three consecutive injections (100 μl at 50 μl/min) of activation buffer (1 M NaCl, 50 mM NaOH) before immobilization of biotinylated DNA (2 nM injected at 10 μl/min). Double-stranded 49-bp oligonucleotides possessing either the wild-type operator sequence (oligonucleotides 9/10) or the same sequence but with 1- or 3-bp insertions between the L and S repeats (oligonucleotides 41/42 and 43/44, respectively) (Table 1) were bound to the chip surface with similar response units to allow comparison of results with different substrates. A double-stranded oligonucleotide with an identical base composition to the wild type, but with an unrelated sequence (oligonucleotides 11/12), was immobilized in one flow cell as a reference. SA chips were pretreated with running buffer (10 mM Tris-HCl [pH 7.5], 50 mM NaCl, 5 mM MgCl₂) prior to injection of YefM-YoeB (50 nM) in running buffer at a flow rate of 50 μl/min at 25°C. The chip surface was regenerated after each sample with a 5-sec pulse of 10 mM Tris-HCl (pH 7.5), 0.5 M NaCl, 0.05% SDS, and 1 mM EDTA (40 μl/min), followed by equilibration with 150 μl of running buffer (40 μl/min). The data were reference subtracted using the oligonucleotide of unrelated sequence and analyzed with BIAevaluation 3.1 software (Biacore AB). None of the kinetic models proposed by the software gave a satisfactory close curve-fitting for the interactions, therefore excluding them from derivation of binding constants for the YefM-YoeB-DNA associations.

CD spectroscopy.

YefM-YoeB was buffer-exchanged using Microcon 3-kDa cutoff filters (Millipore) into 20 mM Tris-HCl (pH 8.5)-50 mM NaCl. For circular dichroism (CD), YefM-YoeB or derivatives in which R10 or R31 of YefM were mutated to alanine were used at concentrations of 20 to 30 μM. CD scans were performed in a Jasco J-810 spectropolarimeter at 20 nm/min with a 0.2-nm data pitch and a 1-s response using a 1-mm bandwidth for eight accumulations at 20°C.

RESULTS

Interaction of YefM-YoeB with the S repeat.

A transcriptional fusion of the yefM-yoeB regulatory region to a promoterless lac operon is repressed ∼6-fold by YefM and ∼30-fold by YefM-YoeB in vivo. The promoter-operator region is bound weakly by YefM in EMSA but is shifted into a discrete nucleoprotein complex by YefM-YoeB at <200 times the protein concentration at which YefM induces a partial shift (23). Previously, five of the six positions in the S repeat in the yefM-yoeB regulatory region were mutated simultaneously. The mutations abolished detectable DNA binding by both YefM and YefM-YoeB in vitro. Nevertheless, YefM repressed transcriptional fusions of the mutated and wild-type yefM-yoeB promoter-operator regions to promoterless lac operons equally efficiently: the protein-DNA complexes may have different stabilities in vitro and in vivo (see below). However, the YefM-YoeB complex failed to exert the additional repression of the disrupted promoter-operator region that it achieves with the wild-type sequence (23). Thus, the S palindrome is required for transcriptional repression by YefM-YoeB, but not by YefM, in vivo.

The contribution of the S hexamer to repression and DNA binding by YefM-YoeB was dissected further by mutating each nucleotide in the repeat separately (Fig. 1A). The effects of these substitutions on repression of the lac operon transcriptional fusion by YefM-YoeB in vivo and on binding by the protein complex in vitro were assessed. None of the six mutations, including the +1G→A alteration of the transcription start site, changed basal expression from the yefM-yoeB promoter (Fig. 1B). Purine-to-purine mutations of transcription start points have been noted previously to have no discernible effects on transcription initiation (26, 29, 55). The +1G→A transition affected repression by YefM-YoeB most dramatically: the complex repressed expression from the mutated promoter-operator region ∼2-fold compared to the 30-fold repression of the wild-type site. The −1T→G, +3A→C, and +5A→C transversions were the most well-tolerated changes, whereas the +2T→G and +4C→T mutations reduced repression by YefM-YoeB, but less markedly than the +1G→A alteration. In summary, mutation of any of the positions in the S hexamer perturbed transcriptional repression by YefM-YoeB, albeit to differing extents.

The wild-type yefM-yoeB regulatory region is bound efficiently in EMSA into a single nucleoprotein complex at ∼1.0 μM YefM-YoeB (23) (Fig. 2). Each of the six nucleotide substitutions in the S repeat greatly reduced binding by YefM-YoeB, even when more than three times this protein concentration was tested (Fig. 2). In accord with the in vivo data (Fig. 1B), the +1G→A, +2T→G, and +4C→T changes exerted the most profound effects. Binding of YefM-YoeB to substrates harboring any of the three other mutations was affected less severely. Thus, substitution of any position in the S palindrome markedly reduces YefM-YoeB binding in EMSA.

FIG. 2.

EMSA analysis of single-base-pair substitution mutations in the S repeat of the yefM-yoeB operator site. Double-stranded 99-bp oligonucleotides (0.1 nM) that were 5′ biotinylated on one strand and that included the yefM translation start codon and 74 bp upstream were subjected to EMSA using increasing amounts of YefM-YoeB. Reactions were processed as outlined in Materials and Methods. Open and filled arrows denote positions of unbound oligonucleotide and YefM-YoeB-DNA complexes, respectively. YefM-YoeB concentrations (μM) used in each panel were (left to right): 0, 0.2, 0.8, 1.6, and 3.2. To illustrate the weak binding induced by the mutations, long exposures of the gels on which the EMSA reactions were electrophoresed are shown. Shorter exposures were used for quantification. As >75% of the wild-type (WT) DNA is bound into protein-DNA complex at the highest concentration of YefM-YoeB used, the quantification for this gel is shown in Fig. 3.

The relative positions of the L and S palindromes determine correct YefM-YoeB interaction with the operator.

The L and S repeats have a center-to-center distance of 12 bp in E. coli K-12 (Fig. 1A). This distance is maintained precisely in the regulatory regions of yefM-yoeB in genomes of diverse bacteria, suggesting that it is a fundamental feature of yefM-yoeB transcriptional control (23). Separation of the binding sites by slightly more than one turn of the DNA helix may permit the cooperative assembly of YefM-YoeB complexes at the two sites. To assess the significance of this spacing, the distance between the palindromes was increased by 1, 2, 3, and 6 bp (Fig. 1A), and the effect of these changes on the interaction of YefM-YoeB with the modified operator sites was assessed. As well as increasing the spacing between the L and S repeats, the insertions extend the distance from the −10 promoter box to the transcription start point. However, transcriptional fusions of the mutated yefM-yoeB promoter-operator regions to lacZYA showed that basal expression was unchanged in all cases compared to the wild-type region, revealing that none of the insertions affected intrinsic promoter activity (Fig. 1C). RNA polymerase demonstrates considerable flexibility in choosing alternative transcription start sites when the original start point is mutated (26). Moreover, the 3- and 6-bp insertions between the L and S repeats restore a G nucleotide to the original +1 position, suggesting that the polymerase may use the same transcription start point in these mutants as in the wild-type site.

The 1-, 2-, 3-, and 6-bp insertions between the L and S repeats each impaired YefM-YoeB repression of a lac operon transcriptional fusion to the yefM-yoeB regulatory region in vivo (Fig. 1C). The decreased repression was very similar with all insertions: 2- to 3-fold regulation with the insertion mutants compared to 30-fold repression with the wild-type site. The extent to which the insertions modulated repression by YefM-YoeB supplied in trans was analogous to that observed when five positions in the S hexamer were mutated simultaneously, thereby abolishing the interaction of the complex with the hexamer (23). This suggests that the residual two- to threefold regulation associated with the mutated sites corresponds to the repression achieved by the interaction of YefM-YoeB with the intact L repeat only, and that a 1-bp insertion alone between the palindromes is sufficient to perturb any cooperative interactions between YefM-YoeB complexes bound to the two repeats.

The interactions between YefM-YoeB and operator sites with modified spacing between the L and S repeats were examined further in vitro. EMSA revealed that the 1-, 2-, 3-, and 6-bp insertions each greatly reduced binding by the YefM-YoeB complex (Fig. 3), affirming the importance of correct spacing between the palindromes. YefM-YoeB protects both the L and S repeats from DNase I digestion in footprinting studies (23). In contrast, YefM-YoeB protected the L repeat, but not the S repeat, in operator DNA possessing the 1-bp insertion (Fig. 4A). Protection of the wild-type and mutated sites occurred at approximately the same protein concentration. Analogously, the L palindromes in operator sites bearing longer insertions were also protected in DNase I footprints, but the S repeats were not (Fig. 4B). Thus, YefM-YoeB recognizes the L repeat more efficiently than the S repeat, which correlates with previous observations that the former is bound at lower YefM concentrations than the latter (23).

FIG. 3.

EMSA analysis of insertion mutations between the L and S repeats of the yefM-yoeB operator site. For the wild-type (WT) DNA, a 99-bp double-stranded oligonucleotide (0.1 nM) that was 5′ biotinylated on one strand and that included the yefM translation start codon and 74-bp upstream was subjected to EMSA using increasing amounts of YefM-YoeB. Other substrates comprised the same sequence except that they harbored 1-, 2-, 3-, or 6-bp insertions illustrated in Fig. 1A. Reactions were processed as outlined in Materials and Methods. Open and filled arrows denote positions of unbound oligonucleotide and YefM-YoeB-DNA complexes, respectively. The YefM-YoeB concentrations used in each panel were (left to right): 0, 0.2, 0.8, 1.6, and 3.2 μM. Quantification of the reactions for wild-type DNA is derived from the EMSA reactions shown in Fig. 2.

FIG. 4.

DNase I footprinting reactions of 1-, 2-, 3-, or 6-bp insertion mutations between the L and S repeats of the yefM-yoeB operator site. Reactions were performed as described in Materials and Methods using PCR fragments biotinylated at the 5′ end of the top strand shown in Fig. 1A. A. Titration reactions in which increasing amounts of YefM-YoeB were used in footprinting reactions with the wild-type (WT) and 1-bp insertion DNAs. YefM-YoeB concentrations (μM, left to right): 0, 0.007, 0.018, 0.036, 0.072, 0.18, 0.36, 0.72, and 1.8. The locations of the L and S repeats are marked by inverted arrows. Shaded boxes denote the regions protected from DNase I digestion by YefM-YoeB. A+G, Maxam-Gilbert sequencing reactions. (B) Single point footprinting reactions with the wild-type (WT) and 1-, 2-, 3-, or 6-bp insertion DNAs. The footprinting reactions either contained no YefM-YoeB (−) or 3.5 μM YefM-YoeB (+). The locations of the L and S repeats are marked by inverted arrows. Shaded boxes denote the regions protected from DNase I digestion by YefM-YoeB.

The preceding transcriptional fusion and footprinting data demonstrate that YefM-YoeB interacts efficiently with a variant operator site in which the S hexamer has been displaced from its normal position on the face of the helix relative to the L repeat. In contrast, the data from EMSA suggest that the complex formed by YefM-YoeB at the mutated site may be unstable and dissociates during electrophoresis (Fig. 3). To test this hypothesis, competition experiments were performed in which preassembled complexes between YefM-YoeB and biotinylated wild-type operator site DNA were incubated with increasing concentrations of unlabeled competitor oligonucleotides. Competitor fragments contained the wild-type operator DNA, sites with 1- or 3-bp insertions, or possessed the same base composition as the fragment bearing the wild-type site, but with a randomized sequence. A competitor oligonucleotide carrying the wild-type site displaced YefM-YoeB efficiently from the labeled fragment, whereas the oligonucleotide with the randomized sequence did not compete at up to 2.4 μM (Fig. 5). Significantly, oligonucleotides harboring either 1- or 3-bp insertions between the L and S palindromes competed equally well, albeit less strongly than the wild-type DNA. These results confirm that YefM-YoeB can bind the mutated operator sites specifically, but that the resulting complexes are unstable.

FIG. 5.

Competition EMSA of YefM-YoeB binding to the wild-type (WT) yefM-yoeB operator site. Binding reactions containing 0.8 μM YefM-YoeB and a 0.3 nM concentration of a 99-bp biotinylated oligonucleotide harboring the wild-type sequence were incubated at 22°C for 15 min. Increasing amounts (up to 2.4 μM) of unlabeled 49-bp competitor oligonucleotides were subsequently added, and incubation continued for an additional 15 min. Reactions were analyzed further as described in Materials and Methods. The competitor DNAs contained the wild-type sequence or the same sequence with 1- or 3-bp insertions between the L and S repeats (see Fig. 1A). A control oligonucleotide had the same length and base composition as other competitor DNAs, but the sequence was randomized.

In view of the instability of YefM-YoeB complexes with substrates possessing displaced S and L palindromes, SPR was used to investigate the YefM-YoeB-operator interactions in real time. Using a Biacore 3000 instrument, SA sensor chips (Biacore) were derivatized with biotinylated 49-bp oligonucleotides bearing the wild-type operator site, with the same site but containing 1- or 3-bp insertions between the L and S repeats, and with an unrelated DNA of the same length and base composition as a reference in one flow cell. Optimal conditions for the experiments were first identified using a range of YefM-YoeB concentrations and flow rates. Subsquently, YefM-YoeB was passed over the immobilized DNAs at 50 nM for 20 s (association), during which time equilibrium was reached, and allowed to wash off subsequently for 125 s (dissociation). The complex bound strongly to the wild-type sequence under these conditions, generating ∼80 response units (Fig. 6). Intriguingly, YefM-YoeB reproducibly bound with higher response units to oligonucleotides bearing the 1- or 3-bp insertions than to the wild-type site. In contrast, a substrate in which five of the six positions in the S repeat were disrupted (Smut) (23) was bound very poorly by the complex. The dissociation patterns of YefM-YoeB from the wild-type site and from the sites containing the insertion mutations were indistinguishable (Fig. 6). The different binding patterns observed with sites possessing a disrupted S repeat and an intact S repeat that is displaced from its position relative to the L repeat indicate that the S hexamer in the latter can still be bound by YefM-YoeB. Thus, the altered interaction patterns in SPR of YefM-YoeB with DNA fragments containing insertions between the L and S repeats may reflect independent binding of YefM-YoeB complexes to the repeats, compared to the synergistic binding to the wild-type site. If so, the interaction with the displaced S hexamer must be sufficiently weak, or transient, that YefM-YoeB can be dislodged by DNase I (Fig. 4) and that the complexes produced do not readily survive gel electrophoresis (Fig. 3). In summary, SPR data revealed that the YefM-YoeB complex interacts with operator sites possessing elongated inter-repeat spacers at least as efficiently as with the site bearing the conventional spacing but that the stabilities of the nucleoprotein complexes differ at wild-type and mutated sites.

FIG. 6.

SPR analysis of YefM-YoeB binding to oligonucleotides containing the wild-type (WT) yefM-yoeB operator site, or the operator site bearing 1- or 3-bp insertions between the L and S repeats (see Fig. 1A). The Smut oligonucleotide harbors an operator in which five of the six positions in the S repeat were disrupted (23).

Twin arginine residues critical for YefM-YoeB binding to DNA.

The tertiary structure of the YefM₂-YoeB complex was recently described (19). The N-terminal domains of both YefM monomers possess conserved basic patches below the symmetrical dimer interface at the base of the structure illustrated in Fig. 7. Although a conventional DNA-binding motif is not evident, this symmetrical underside surface of the antitoxin has been suggested to serve as the primary DNA anchor for operator site binding. In particular, two arginine residues (R10 and R31) that are highly conserved among YefM homologues from diverse sources are prime candidates for direct interactions with the operator site (19). Site-specific mutation of the arginine residues separately to alanine did not appreciably disrupt the gross structure of the YefM-YoeB complex based on CD analysis (Fig. 7), although definitive information awaits comparative analysis of the tertiary structures of wild-type and mutated complexes.

FIG. 7.

R10A and R31A mutations in YefM within the YefM-YoeB complex. On the left, the tertiary structure of the YefM₂-YoeB heterotrimer is shown (19). YoeB is colored green, and the two YefM monomers are depicted in light and dark blue. The locations of the R10 and R31 residues in each YefM monomer are highlighted in red. These residues were previously denoted R19 and R40, respectively (19) but are renumbered here. A proposed nine-amino-acid extension at the N-terminal end of E. coli YefM is missing from the purified protein (19), and the transcription start for yefM-yoeB is located within the DNA sequence for this hypothetical extension (23), revealing that the first codon for YefM is nine codons downstream of that suggested by Kamada and Hanaoka (19) and Cherny and Gazit (4) and is in agreement with that proposed by Grady and Hayes (15). The image was prepared using PyMOL (http://www.pymol.org). On the right side of the figure, far UV CD spectra of YefM-YoeB (WT) and YefM-YoeB bearing either R10A or R31A mutations in YefM are shown.

The contribution of the R10 and R31 residues to autoregulation and DNA binding by YefM-YoeB was assessed by testing the effects of the alanine mutations on repression of a lac operon under the control of the yefM-yoeB promoter-operator region in vivo and for binding to the operator site in vitro. The R10A and R31A mutations in YefM both entirely ablated the 30-fold transcriptional repression observed with the wild-type complex in vivo (Fig. 8A). Moreover, the mutated complexes did not bind detectably to the operator DNA either in EMSA analysis (data not shown) or in SPR studies (Fig. 8B). Thus, both the R10 and R31 residues in YefM are absolutely necessary for DNA binding by the YefM-YoeB complex.

FIG. 8.

DNA-binding properties of YefM-YoeB complexes bearing R10A or R31A mutations in YefM. (A) Effect of the R10A and R31A mutations on transcriptional repression by YefM-YoeB in vivo. Transcriptional fusions of the yefM-yoeB regulatory region to the lacZYA operon in plasmid pRS415 were transformed into E. coli SC301467, which is deleted of five chromosomal TA cassettes including yefM-yoeB. β-galactosidase levels were determined with wild-type (WT) or mutated YefM-YoeB supplied in trans from the pBAD33 arabinose-inducible vector (pBADyefMyoeB). (B) SPR analysis of wild-type or mutated YefM-YoeB binding to a 49-bp oligonucleotide containing the yefM-yoeB operator site.

DISCUSSION

The activity of TA complexes is controlled at numerous levels: intracellular toxin availability is influenced by the strength of the interaction between toxin and antitoxin and the relative susceptibilities of the labile antitoxin and resistant toxin to protease degradation. The kinetics with which the toxin binds to and is released from its target determine the timespan that the cell is exposed to the toxin, as does the rate at which antitoxin can later be resynthesized, if at all. The latter is determined in part by the autoregulatory circuit that controls TA gene expression, which also dictates the level of TA complex that is produced during steady-state conditions. All TA complexes that have been tested are subject to negative regulation at the transcriptional level (17). With some notable exceptions (12, 13), the antitoxin is the principal repressor, with the toxin acting as a cofactor that can dramatically enhance DNA binding by the antitoxin, most likely via conformational changes induced in the antitoxin. The toxin alone exhibits no DNA-binding activity. Repression invariably involves the interaction of the proteins with palindromic sequences situated in the vicinity of the TA promoter. For example, the MazEF complex binds three degenerate 12-bp repeats, which can also be viewed as two overlapping repeats that could be recognized alternately by the complex (34). As with YefM-YoeB, the operator site bound by the RelBE TA complex comprises a pair of juxtaposed inverted repeats that overlap the −10 promoter box and extend toward the antitoxin open reading frame (28). A perfect palindrome is bound more strongly by RelBE than an imperfect second repeat. This partially mirrors the interaction of YefM-YoeB with its operator site where the L repeat is recognized first by the antitoxin, before coverage of the adjacent S repeat. The additional palindromic nucleotides at the outer ends of the L repeat likely promote preferential binding to this repeat compared to the S hexamer. The L and S palindromes have a center-to-center distance of 12-bp (23), which is also the case with the repeats bound by RelBE (28). The spacing was shown here to be crucial for the correct stable positioning of YefM-YoeB at the two repeats. This may reflect synergistic contacts between YefM-YoeB assembled at the sites. Altering the relative positions of the binding sites may perturb these contacts. Although the yefM-yoeB and relBE operators share some organizational similarities, the 5′-TT(G/A)(T/C)AA-3′ motif recognized by RelBE differs from the 5′-TGTACA-3′ core sequence that is bound by YefM-YoeB. Moreover, the YefM and RelB DNA-binding proteins have disparate tertiary structures (see below). The operator region involved in transcriptional autoregulation of the phd-doc TA cassette of bacteriophage plasmid P1 includes two palindromic binding sites for the Phd antitoxin. These sites are separated by a center-to-center distance of 13 bp, and each site consists of core 5′-GTGTACAC-3′ sequences, similar to the repeats within the yefM-yoeB operator region. Moreover, the N-terminal regions of YefM and Phd are homologous (59), and the Phd protein possesses arginine residues at equivalent positions to R10 and R31 of YefM. These observations suggest that the regulatory patterns of the two systems may be very similar. Other TA complexes bind to operator sites that have a variety of palindrome arrangements. For example, the FitAB complex binds a ∼60-bp DNA stretch centered on the palindromic sequence 5′-TGCTATCA-N₁₂-TGATAGCA-3′ (35), and the Kis-Kid proteins bind to two imperfect 18-bp inverted repeats that are separated by 33 bp (38). In contrast, the CcdA-CcdB TA complex encoded by the F plasmid spirals around a ∼110-bp sequence upstream of ccdAB that extends into the 5′ end of the ccdA gene. The complex interacts discontinuously with the DNA, with alternating stretches of occupation and absence (10).

The tertiary structures of numerous antitoxin proteins have been elucidated recently, either as part of a complex with the cognate toxin and/or in the unbound form or in association with operator site DNA (19, 20, 22, 24, 28, 31, 32, 35, 36, 43, 53). These structures have revealed significant diversity in the folds that antitoxins use to bind to their regulatory sites. The N-terminal domain of the MazE antitoxin adopts a β-barrel conformation from which extended C termini interact with the MazF toxin. The protein forms a heterohexamer with MazF (MazF₂-MazE₂-MazF₂) in the crystal structure (20). Although MazE lacks a canonical DNA-binding motif, the protein is a member of a diverse superfamily of DNA-binding proteins (3). Basic residues in the N-terminal region of each MazE monomer likely contact two overlapping palindromes that comprise the mazEF operator site (20, 31). Although the MazF and CcdB toxins share structural similarities, the organizations of the MazE and CcdA antitoxins differ markedly (32). The CcdA dimer possesses a ribbon-helix-helix (RHH) structure in which basic residues located in antiparallel β-strands interact with numerous 6-bp palindromic sequences as mentioned above (10, 32). Similarly, the FitA, RelB, and ParD antitoxins are RHH dimers that bind their cognate operator sites (28, 35, 43), as is the ω protein which is a discrete factor that represses transcription of the ɛζ TA complex (40). An archaeal RelB homologue lacks the RHH fold found in bacterial RelB proteins but instead was proposed to recognize DNA via a cluster of basic residues at its N-terminal end (53). In contrast, the N-terminal domain of the PezA antitoxin of Streptococcus pneumoniae exhibits a DNA-binding domain fold with a helix-turn-helix motif that is characteristic of the Xre and Cro/CI family (24). Like the MazE and archaeal RelB antitoxins, YefM does not possess a canonical DNA-binding motif, but instead a pair of basic residues conserved in many YefM homologues were proposed to be involved in operator recognition (19). The R10 and R31 residues are situated far from the YoeB catalytic site in the YefM-YoeB heterotrimer, being located near to the “base” of the structure in a region that is predicted to be readily accessible by DNA (Fig. 7A). Alanine substitutions at these positions entirely abolished operator site binding by YefM-YoeB, affirming that a novel fold mediates palindrome recognition by YefM and related antitoxins. Solution of the YefM and YefM-YoeB structures bound to DNA will further reveal the mode of YefM-DNA interaction, as will complementary mutagenesis studies of other conserved amino acids in the vicinity of R10 and R31.

The physiological role of the YefM-YoeB complex remains unknown. Deletion of five TA cassettes, including yefM-yoeB, in E. coli elicited no obvious phenotype, did not affect the deletant's ability to compete with the wild-type strain, and did not alter the bacterium's response to a range of nutritional and other stresses (54). It is possible that the deletion strain harbored other TA systems (50) that may have confounded these experiments. However, intriguing hints at a possible role for YefM-YoeB come from observations that the yefM gene is upregulated during growth of E. coli in biofilms (48) and that both yefM and yoeB are overexpressed in persister cells (51). Persisters are a subpopulation of dormant cells that naturally display transient multidrug tolerance. Persister cells are produced in slow-growing biofilms, for example, but increased persistence can also be achieved by ectopic overproduction of the toxin components of TA complexes (27). The endoribonucleolytic activity of YoeB and other RNA cleaving toxins may be important in inducing the dormant state. Nevertheless, the impact of YoeB overproduction on persister cell formation has not been tested to our knowledge, and deletion of yoeB had no effect on persistence (51). The latter may reflect redundancy among TA systems for persister cell formation. In view of the upregulation of yefM that has been reported in biofilms (48) and of yefM-yoeB in persister cells (51), it is tempting to speculate that derepression of yefM-yoeB autoregulation occurs in these circumstances in response to an as-yet-unknown environmental or cell cycle signal(s) that interferes with the YefM-YoeB-operator interaction. Further dissection of this interaction, along with the identification of small molecules that intervene in it, will provide key insights into TA control, activation, and function.