Kevin L. Griffith

Proteolytic degradation of Escherichia coli transcription activators SoxS and MarA as the mechanism for reversing the induction of the superoxide (SoxRS) and multiple antibiotic resistance (Mar) regulons

In Escherichia coli, the SoxRS regulon confers resistance to redox‐cycling compounds, and the Mar regulon provides a defence against multiple antibiotics. The response regulators, SoxS and MarA, are synthesized de novo in response to their inducing signals and directly activate transcription of a common set of target genes. Although the mechanisms of transcription activation by SoxS and MarA have been well studied, little is known about how the systems are shut‐off once the inducing stress has subsided, except that de novo synthesis of the regulators is known to cease almost immediately. Here, we induced the SoxRS regulon and determined that, upon removal of the inducer, expression of the regulons genes quickly returns to the preinduced level. This rapid shut‐off indicates that the system is reset by an active process. We found that SoxS is unstable and infer that SoxS degradation is responsible for the rapid return of the system to the ground state upon removal of the inducing signal. We also found that MarA is unstable and that the instability of both proteins is intrinsic and unregulated. We used null mutations of protease genes to identify the proteases involved in the degradation of SoxS and MarA. Among single protease mutations, only lon mutations increased the half‐life of SoxS and MarA. In addition, SoxS appeared to be nearly completely stable in a lon ftsH double mutant. Using hexahistidine tags placed at the respective ends of the activators, we found that access to the amino‐terminus is essential for the proteolytic degradation.

Interdependence of the position and orientation of SoxS binding sites in the transcriptional activation of the class I subset of Escherichia coli superoxide-inducible promoters

SoxS is the direct transcriptional activator of the member genes of the Escherichia coli superoxide regulon. At class I SoxS‐dependent promoters, e.g. zwf and fpr, whose SoxS binding sites (‘soxbox’) lie upstream of the −35 region of the promoter, activation requires the C‐terminal domain of the RNA polymerase α‐subunit, while at class II SoxS‐dependent promoters, e.g. fumC and micF, whose binding sites overlap the −35 region, activation is independent of the α‐CTD. To determine whether SoxS activation of its class I promoters shows the same helical phase‐dependent spacing requirement as class I promoters activated by catabolite gene activator protein, we increased the 7 bp distance between the 20 bp zwf soxbox and the zwf−35 promoter hexamer by 5 bp and 11 bp, and we decreased the 15 bp distance between the 20 bp fpr soxbox and the fpr−35 promoter hexamer by the same amounts. In both cases, displacement of the binding site by a half or full turn of the DNA helix prevented transcriptional activation. With constructs containing the binding site of one gene fused to the promoter of the other, we demonstrated that the positional requirements are a function of the specific binding site, not the promoter. Supposing that opposite orientation of the SoxS binding site at the two promoters might account for the positional requirements, we placed the zwf and fpr soxboxes in the reverse orientation at the various positions upstream of the promoters and determined the effect of orientation on transcription activation. We found that reversing the orientation of the zwf binding site converts its positional requirement to that of the fpr binding site in its normal orientation, and vice versa. Analysis by molecular information theory of DNA sequences known to bind SoxS in vitro is consistent with the opposite orientation of the zwf and fpr soxboxes.

Systematic mutagenesis of the DNA binding sites for SoxS in the Escherichia coli zwf and fpr promoters: identifying nucleotides required for DNA binding and transcription activation

SoxS is the direct transcriptional activator of at least 15 genes of the Escherichia coli superoxide regulon. SoxS is small (107 amino acids), binds DNA as a monomer and recognizes a highly degenerate DNA binding site, termed ‘soxbox’. Like other members of the AraC/XylS family, SoxS has two putative helix–turn–helix (HTH) DNA‐binding motifs, and it has been proposed that each HTH motif recognizes a highly conserved recognition element of the soxbox. To determine which nucleotides are important for SoxS binding, we conducted a systematic mutagenesis of the DNA binding sites for SoxS in the zwf and fpr promoters and determined the effect of the soxbox mutations on SoxS DNA binding and transcription activation in vivo by measuring β‐galactosidase activity in strains with fusions to lacZ. We found that the sequences GCAC and CAAA, termed recognition elements 1 and 2 (RE 1 and RE 2), respectively, are critical for SoxS binding, as mutations within these elements severely hinder or eliminate SoxS‐dependent transcription activation; substitutions within RE 2 (CAAA), however, are tolerated better than changes within RE 1 (GCAC). Although substitutions at the seven positions separating the two REs had only a modest effect on SoxS binding, AT basepairs were favoured within this ‘spacer’ region, presumably because, by facilitating DNA bending, they help bring the two recognition elements into proper juxtaposition. We also found that the ‘invariant A’ present at position 1 of 14/15 functional soxboxes identified thus far is important for SoxS binding, as a change to any other nucleotide at this position reduced SoxS‐dependent transcription by ≈ 50%. In addition, positions surrounding the REs seem to show a context effect, in that certain substitutions there have little or no effect when the RE has the optimal binding sequence, but produce a pronounced effect when the RE has a suboptimal sequence. We propose that these nucleotides play an important role in effecting differential expression from the various promoters. Lastly, we used gel retardation assays to show that alterations in transcription activation in vivo are caused by effects on DNA binding. Based on this exhaustive mutagenesis, we propose the following optimal sequence for SoxS binding: AnVGCACWWWnKRHCAAAHn (n = A, C, G, T; V = A, C, G; W = A, T; K = G, T; R = A, G; H = A, C, T).

Characterization of TetD as a transcriptional activator of a subset of genes of the Escherichia coli SoxS/MarA/Rob regulon

In Escherichia coli, SoxS, MarA and Rob form a closely related subset of the AraC/XylS family of positive regulators, sharing ∼42% amino acid sequence identity over the length of SoxS and the ability to activate transcription of a common set of target genes that provide resistance to redox‐cycling compounds and antibiotics. On the basis of its ∼43% amino acid sequence identity with SoxS, MarA and Rob, TetD, encoded by transposon Tn10, appears to be a fourth member of the subset. However, although its expression has been shown to be negatively regulated by TetC and not inducible by tetracycline, the physiological function of TetD is unknown. Accordingly, in the work presented here, we initiate a molecular characterization of TetD. We show that expression of TetD activates transcription of a subset of the SoxS/MarA/Rob regulon genes and confers resistance to redox‐cycling compounds and antibiotics. We show that mutations in the putative TetD binding site of a TetD‐activatable promoter and a mutation in the proteins N‐terminal DNA recognition helix interfere with transcription activation, thereby indicating that TetD directly activates target gene transcription. Finally, we show that TetD, like SoxS and MarA, is intrinsically unstable; however, unlike SoxS and MarA, TetD is not degraded by Lon or any of the cells known cytoplasmic ATP‐dependent proteases. Thus, we conclude that TetD is a bona fide member of the SoxS/MarA/Rob subfamily of positive regulators.

A back-propagation neural network predicts absorption maxima of chimeric human red/green visual pigments.

The absorption spectra of human red and green visual pigments have peak wavelengths, lambda max, that differ by 31 nm, yet the opsins differ in only 15 amino acids. Mutagenesis studies have demonstrated that seven of the 15 amino acids determine the spectral shift. We trained neural networks to predict the lambda max of any red/green chimeric protein. Seven mutants were excluded from the original training set. The trained networks were able to predict the lambda max for the excluded mutants. As an additional test, five new chimeric pigments were constructed and lambda max determined. The neural networks correctly predicted the lambda max of all five mutants. The use of neural networks is a novel approach to the problem of wavelength modulation in visual pigments.