Overview of research in the Wolf Lab. We study the mechanisms of gene regulation in the bacterium Escherichia coli. We focus on a closely related group of four paralogous proteins, SoxS, MarA, Rob, and TetD, which are members of the AraC/XylS family of proteins, many of which are transcription activators that mediate stress responses and virulence in pathogenic bacteria. These four proteins activate transcription of the same set of defense response genes, although the extent of activation of a given gene varies from protein to protein. For historical reasons, most of our work has been on SoxS but recently we have invested considerable effort in studying Rob.
Summary of why we study SoxS and the significant discoveries we have made with it. Through our studies of SoxS we have discovered a new mechanism of transcription activation, which we call “pre-recruitment”. This discovery is rather amazing given that investigators have studied mechanisms of gene regulation in bacteria for more than 50 years. We also discovered that activation of transcription by SoxS requires that it make a heretofore unknown protein-protein contact with RNA polymerase, one that enables SoxS to “appropriate” RNAP and divert it from “strong”, e.g., rRNA promoters, to SoxS-dependent promoters. We also determined that the binding site for SoxS is highly degenerate, that it activates transcription from two types of promoters (class I and class II), that it functions as a monomer, and that it is highly unstable. All of these properties are uncommon among bacterial transcriptional regulators. Thus, SoxS has many interesting and unusual characteristics that have made it an important protein for in-depth study. But, much more remains to be learned.
The genes of the SoxRS regulon are expressed by an unusual two-gene, two-stage system. SoxS is the direct transcriptional activator of the genes of the SoxRS regulon. Expression of these SoxS-dependent genes mediates the cell’s defense against reactive oxygen species, e.g., superoxide radical. The first unusual property of SoxS, determined by the labs of B. Demple and B. Weiss, is that it is synthesized de novo in response to oxidative stress as part of a two-gene, two-stage process. In most regulatory systems, the protein components are continually expressed and the system becomes activated by small molecule inducers that interact directly with the regulator or by a signal transduction cascade that modifies and activates the response regulator. In the SoxRS system, SoxR is synthesized constitutively in an inactive form. Upon encountering an oxidative stress, SoxR becomes activated when its dual 2Fe-2S centers become oxidized. Activated SoxR only has one function: to activate transcription of SoxS. Newly synthesized SoxS then activates transcription of the ~50 genes that carry out the defense response.
The initial, circumstantial evidence that SoxS does not activate transcription by recruitment. Most transcription activators in bacteria function by “recruitment”. In recruitment, the transcription activator first binds to a sequence in the promoter of the target gene; the DNA-bound activator then makes protein-protein interactions with RNAP that “recruits” it to the promoter and thereby stimulates transcription initiation at the target gene. The first in vivo evidence that SoxS does not activate transcription of the member genes of the SoxRS regulon by recruitment was largely circumstantial. Our earlier work had demonstrated that the binding site for SoxS is highly degenerate 1,2, so much so that the E. coli genome was estimated to contain ~13,000 “soxboxes” 3. Moreover, since rapidly growing cells have a DNA content equivalent to 4-6 genomes, such cells will contain ~65,000 SoxS binding sites. The conundrum posed is that the SoxRS regulon is comprised of only about 50 genes; i.e., rapidly growing cells will contain about 250 SoxS-dependent promoters. If recruitment were the mechanism of transcription activation by SoxS, then induction of de novo SoxS would have to rapidly produce enough SoxS molecules to bind to a large fraction of the total soxboxes in the cell, in order to ensure that SoxS will be able to activate transcription of the defense response genes despite the large excess of sequence-equivalent but non-functional soxboxes scattered around the chromosome.
To address the question of whether SoxS functions by recruitment, we purified SoxS, prepared antibody against it, and used it to determine the number of SoxS molecules per cell after induction of its synthesis 3. A maximum of 2,500 SoxS molecules was produced 3. Thus, although the number of SoxS molecules per cell exceeds by tenfold the number of SoxS-dependent promoters per cell, the number of soxboxes per cell exceeds the number of SoxS molecules per cell by 26-fold. Accordingly, we decided that this disparity makes it very unlikely that SoxS functions by recruitment. Instead, we proposed that SoxS activates transcription by a new mechanism, which we call “pre-recruitment” 3.
Transcription activation by pre-recruitment. In pre-recruitment (see Image 1), we propose that the activator, SoxS, first makes protein-protein contacts with RNAP, forming a SoxS-RNAP binary complex. This complex then scans the chromosome for SoxS-dependent promoters using two “reading heads”, the sigma factor of RNAP, which binds to the -10 and -35 promoter elements, and SoxS, which binds to properly positioned and oriented soxboxes. In so doing, the sequence-equivalent but non-functional soxboxes are ignored and the complex only binds to the small subset of soxboxes that are embedded in promoters. Moreover, since SoxS is a monomer 4, transcription activation of the regulon’s genes does not depend on SoxS achieving an intracellular concentration sufficient for oligomerization. Instead, the first molecules synthesized upon induction have the potential to interact with RNAP in solution, scan the chromosome for target promoters, and activate transcription from them. And, with SoxS being only 107 amino acids long and protein synthesis occurring at a rate of 15 amino acids per second, it only takes seven seconds to synthesize a functional SoxS polypeptide.
Genetic and biochemical evidence for pre-recruitment. We have conducted an in vivo test of the pre-recruitment hypothesis. First, we carried out a complete alanine scanning mutagenesis of SoxS 5, replacing each of the 102 non-alanine residues with alanine. We then determined the effects of the alanine substitutions on DNA binding and transcription activation in vivo 5. We also purified ~70 mutant proteins and determined whether they could bind DNA normally or not 5. Two classes of mutants relevant to the test of pre-recruitment were identified: mutants defective in DNA binding and “positive control” mutants, which bind DNA normally but which are defective in transcription activation 5.
These mutants were used in genetic dominance tests 6. The pre-recruitment hypothesis predicts that overexpression of a mutant defective in DNA binding will be dominant to the wild type allele, because it will form a binary complex with RNAP that is unable to bind to and activate transcription of SoxS-dependent promoters. In contrast, DNA binding mutants of an activator that functions by recruitment should be recessive, because overexpression of the mutant protein cannot interfere with the ability of the wild type protein to recruit RNAP to the promoter.
Moreover, the pre-recruitment model predicts that positive control mutants of SoxS will be recessive to the wild type allele, because the mutant protein cannot form binary complexes with RNAP and therefore cannot interfere with the function of the wild type protein. In contrast, positive control mutants of an activator that functions by recruitment are dominant, because the mutant protein binds to its target sequence in the promoter and although it cannot recruit RNAP by occupying the activator binding site it excludes binding by the wild type protein and thereby prevents it from recruiting RNAP.
We conducted the genetic dominance tests with our DNA binding and positive control mutations of SoxS. The results were exactly those predicted of the pre-recruitment hypothesis 6. Thus, this experiment provides convincing genetic evidence, consistent with the above-described circumstantial evidence, that SoxS activates transcription by pre-recruitment, not recruitment.
In addition to this genetic test, Robert Martin and Lee Rosner of the NIH demonstrated that SoxS (and MarA) can form binary complexes with RNAP in solution and in the absence of specific DNA binding. Thus, pre-recruitment is supported by circumstantial evidence, by in vivo genetic tests, and by in vitro biochemical experiments.
What’s next? Our next task is to identify all the member genes of the SoxRS regulon. This will be accomplished by conducting Chromatin Immunoprecipitation (ChIP) experiments with anti-SoxS antibodies and then hybridizing the immunoprecipitated DNA to high-density, tiled microarrays. These ChIP-chip experiments will be done kinetically so that we can determine if the order of appearance of the genes of the regulon following imposition of the oxidative stress makes sense physiologically.
SoxS-RNAP binary complexes arise by interaction between one surface of SoxS but two surfaces of RNAP. We have determined the components of RNAP that interact with SoxS and the regions of SoxS that interact with RNAP 7. This was initially accomplished with a yeast two-hybrid system. With SoxS as “bait”, we found that it makes protein-protein contacts with both the C-terminal domain (CTD) of the alpha subunit of RNAP 7 and also with Region 4 of the sigma factor (Shah and Wolf, unpublished results). Alanine substitutions of the “class I/II surface” of SoxS, which prevent transcription activation of both class I promoters (where the soxbox resides upstream of the -35 promoter hexamer and in the “backwards” orientation) and class II promoters (where the soxbox lies in the “forward” orientation and overlaps the promoter hexamer) prevent interaction with both domains of RNAP 7. These results were confirmed by in vitro “pull-down” experiments with SoxS and the full-length alpha and sigma subunits 7. Thus the same class I/II surface of SoxS is used to form two different binary complexes with RNAP. Perhaps, the interaction with alpha is used to activate transcription from class I promoters and the interaction with sigma is used to activate transcription from class II promoters.
The role of the class II surface in transcription activation is unknown. Interestingly, alanine substitutions of the “class II surface” of SoxS, which prevent transcription activation of class II promoters, have no effect on the interactions of SoxS with the alpha-CTD or sigma Region 4 in either the yeast two-hybrid system or in the pull-down experiments (Shah and Wolf, unpublished results). Moreover, although genetic epistasis experiments have revealed specific amino acid contacts between residues of sigma Region 4 and residues of the class I/II surface, no interactions were identified that involve the amino acids of the class II surface (Shah and Wolf, unpublished results). The failure to find interactions between the class II surface of SoxS and sigma Region 4 was unexpected because the class II surfaces of other proteins closely related to SoxS, e.g., Rob, RhaS, and MelR, are known to make protein-protein contacts with sigma Region 4. Hence, determining the role of the class II surface of SoxS is an important objective of our current research.
SoxS is a “co-sigma factor”, since in one SoxS-RNAP binary complex, the class I/II surface of SoxS makes protein-protein contacts with the DNA binding determinant of the alpha-CTD. Yet another interesting and unexpected result that came from the yeast two-hybrid experiments is that alanine substitutions of the DNA binding determinant of the alpha-CTD prevent the interaction with SoxS 7. This interaction was also confirmed by pull-down experiments. Normally, the DNA binding determinant of the alpha-CTD binds to the “UP element”, a third, A/T-rich promoter element found upstream of strong promoters, e.g., the promoters for the rRNA genes. If SoxS binds to this region of the alpha-CTD during oxidative stress, then transcription from strong promoters would be severely reduced. We tested this hypothesis in E. coli after inducing SoxS synthesis with paraquat, a redox-cycling compound that endogenously generates superoxide. Indeed, rRNA synthesis was reduced several-fold, but only if the promoters contained the UP element and only if the class I/II surface of SoxS was intact 7.
Thus, these experiments revealed yet another remarkable and surprising property of SoxS: it functions as a “co-sigma factor” that appropriates RNAP, diverting it from transcription of strong promoters to SoxS-dependent promoters. This redeployment enhances the ability of the cell to survive the oxidative stress, albeit at the expense of the ability to grow rapidly. Currently, we are using the ChIP-chip procedure to identify on a genome-wide scale the genes whose transcription is reduced when synthesis of SoxS is induced. We are also using this approach to determine whether SoxS “rides” with RNAP as it transcribes the genes of the SoxRS regulon. If so, as soon as SoxS finishes transcription of one regulon gene, it would be able to activate transcription of another without having to form a new binary complex.
If SoxS captures RNAP and diverts it from “housekeeping” promoters to stress-inducible promoters, how does the cell compete once the stress has been overcome? This redeployment of RNAP raised the question of how the SoxRS system resets once the oxidative stress has been relieved. We reasoned that the most logical mechanism to bring the SoxRS system back to the ground state would be if SoxS were unstable. Indeed, we measured the half-life of SoxS after removing an inducer of oxidative stress and found that it is only about two minutes, whereas most bacterial proteins are infinitely stable 8. Thus, when the stress has been relieved, de novo synthesis of SoxS ceases, the SoxS-RNAP binary complexes dissociate, SoxS is degraded, and RNAP is free to transcribe its normal “housekeeping” genes.
Proteolytic degradation of SoxS is regulated. Using a set of mutants defective in the proteases of E. coli, we determined that Lon is the main enzyme responsible for degradation of SoxS 8. Then, in a series of genetic experiments, we demonstrated that the 17 amino acids at the N-terminus of SoxS are the substrate for Lon protease 9. We purified SoxS and demonstrated that purified Lon protease is sufficient for degradation of SoxS, provided that the N-terminus is accessible 9. We also discovered that the binding of SoxS to soxbox DNA or to RNAP blocks proteolysis by Lon 10, the first example of the ability of DNA and RNAP to serve as inhibitors of the degradation of a regulatory protein by a specific protease. This protection from degradation also makes sense physiologically: when SoxS is being synthesized de novo in response to oxidative stress, binary complex formation and transcription activation protect it from degradation so that it can continue to mount the defense response. Then, when the threat has been defeated, SoxR becomes reduced and SoxS is no longer synthesized. Then, SoxS dissociates from RNAP and being no longer protected, it is degraded by Lon protease and RNAP returns to its normal duties.
Summary of why we study Rob and the significant discoveries we have made with it. Rob is about 50% identical in amino acid sequence to SoxS, MarA, and TetD over the 107 amino acid length of SoxS, the shortest member of the subset of AraC/XylS family members 11. However, Rob differs from SoxS, MarA and TetD in a very basic way: it has a second, C-terminal domain (CTD) that is responsible for fundamental differences between it and its paralogs. In particular, Rob’s ability to function as a transcription activator is regulated by a new mechanism of induction, “sequestration-dispersal” 12, that is dependent on the CTD. By regulating the activity of the N-terminal domain, (NTD), which carries out DNA binding and transcription activation by the same pre-recruitment mechanism as SoxS, the CTD is functioning as an anti-sigma factor. In addition, whereas the mono-domain proteins SoxS, MarA and TetD are highly unstable, Rob is completely stable and this stability is dependent on the CTD 12.
Similarities and differences between Rob and its paralogs. Because of the extensive amino acid sequence homology between Rob and SoxS, MarA and TetD, especially within the dual helix-turn-helix DNA binding motifs 11, monomeric Rob binds to the same degenerate sequence and activates the same set of genes as its paralogs 4. Moreover, genetic dominance tests like those used with SoxS have shown that Rob activates transcription by pre-recruitment (E.F. Keen III and Wolf, unpublished results). And, like SoxS, activated Rob binds to the DNA-binding determinant of the alpha-CTD, redeploying it from strong, UP element-containing promoters to stress-inducible promoters activated by Rob (Keen and Wolf, unpublished results). Accordingly, Rob also functions as a co-sigma factor.
Despite these significant similarities, Rob differs in a fundamental way from its paralogs. Whereas SoxS, MarA and TetD effect their specific defense responses as part of two-gene, two-stage systems in which the activators are synthesized de novo in active but unstable forms, K. Skarstad in the lab of A. Kornberg showed that Rob is synthesized constitutively in an inactive, stable form. Then, the labs of J.L. Rosner and H. Nikaido showed that Rob activity is induced by dipyridyl (DIP) and bile salts and medium chain fatty acids like decanoate (DEC), respectively, and these inducers interact directly with Rob’s CTD. Lastly, T.A. Azam, in the lab of A. Ishihama, discovered that the 10,000 molecules of Rob per cell reside in 3-4 punctate, immuno-stainable foci. Based on these findings, we hypothesized that the inducers of Rob activity function by releasing it from the clusters.
Sequestration-dispersal as a new mechanism for regulating the activity of a transcription activator. Most transcription activators are synthesized constitutively in an inactive form. Inducing signals activate activators by a variety of mechanisms that convert them to an active conformation, e.g., by direct interaction between a small molecule ligand or indirectly through a signal transduction pathway that covalently modifies the activator. In addition, some activators are unstable and the inducing signal leads to their stabilization.
We used indirect immunofluorescence microscopy (IFM) to test the hypothesis that inducers activate sequestered Rob by dispersing it from the clusters such that it can gain access to RNAP, form binary complexes with it and thereby activate transcription by pre-recruitment. Being unable to secure stable, high-titer antibodies against Rob, we prepared a SoxS-Rob chimera and used it as a Rob surrogate, because we had affinity-purified anti-SoxS antibodies. The SoxS-Rob chimera, in which Rob’s CTD is fused to the C-terminus of SoxS, activates transcription like SoxS, but is regulated like Rob.
We conducted IFM with the SoxS-Rob chimera in the absence and presence of inducers DIP and DEC. As in Azam’s IFM of native Rob, we found that under normal, non-inducing conditions the SoxS-Rob chimera resides in 3-4 punctate foci 12. In contrast, SoxS without Rob’s CTD is fully dispersed during normal growth. More importantly, we found that within five minutes after inducing treatment with DIP or DEC, the clusters of the SoxS-Rob chimera begin to disappear and within ten minutes the SoxS-Rob chimera is almost completely dispersed 12. Coincident with dispersal, Rob becomes active in that it is able to activate transcription of Rob-dependent promoters like those of inaA and fumC 12.
Although these initial experiments were done with SoxS-Rob expressed from a medium copy plasmid, we replaced the N-terminal 107 amino acids of chromosomally encoded Rob with those of SoxS by “Recombinant Enhancement by Selection for Survival”, a scar less method of recombineering of our design. We then carried out IFM in the absence and presence of inducer and found that chromosomally encoded SoxS-Rob is sequestered in the absence of inducer and rapidly dispersed upon induction 12. Importantly, the activity of the chromosomally encoded chimera is enhanced by treatment with inducer 12. Thus, following an inducing treatment, the ability of SoxS-Rob produced under native conditions to activate target gene transcription is increased because the chimera becomes dispersed. Moreover, since induction leads dispersed Rob to function as a co-sigma factor, removal of inducer and subsequent sequestration of Rob indicates that Rob’s CTD regulates Rob activity by functioning as an anti-sigma factor.
Rob’s NTD, like native SoxS, is intrinsically unstable 12. Thus, since full-length Rob is fully stable, Rob’s CTD confers stability on the NTD. Moreover, the SoxS-Rob chimera is also fully stable and thus Rob’s CTD can not only confer regulation by sequestration-dispersal on another protein when fused to it but it can also stabilize an intrinsically unstable protein 12. Accordingly, it will be interesting to determine whether Rob can confer these properties on a totally unrelated protein.
1. Wood, T. I., Griffith, K. L., Fawcett, W. P., Jair, K.-W., Schneider, T. D. & Wolf, R. E., Jr. (1999). 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. Mol. Microbiol. 34, 414-430.
2. Griffith, K. L. & Wolf, R. E., Jr. (2001). 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. Mol. Microbiol. 40, 1141-1154. Erratum Mol. Microbiol. 42:571.
3. Griffith, K. L., Shah, I. M., Myers, T. E., O'Neill, M. C. & Wolf, R. E., Jr. (2002). Evidence for "pre-recruitment" as a new mechanism for transcription activation in Escherichia coli: the large excess of SoxS binding sites per cell relative to the number of SoxS molecules per cell. Biochem. Biophys. Res. Commun. 291, 979-986 Erratum. Biochem. Biophys. Res. Commun. 294:1191.
4. Jair, K.-W., Yu, X., Skarstad, K., Thöny, B., Fujita, N., Ishihama, A. & Wolf, R. E., Jr. (1996). Transcriptional activation of promoters of the superoxide and multiple antibiotic resistance regulons by Rob, a binding protein of the Escherichia coli origin of chromosomal replication. J. Bacteriol. 178, 2507-2513.
5. Griffith, K. L. & Wolf, R. E., Jr. (2002). A comprehensive alanine scanning mutagenesis of the Escherichia coli transcriptional activator SoxS: identifying amino acids important for DNA binding and transcription activation. J. Mol. Biol. 322, 237-257.
6. Griffith, K. L. & Wolf, R. E., Jr. (2004). Genetic evidence for pre-recruitment as the mechanism of transcription activation by SoxS of Escherichia coli: the dominance of DNA binding mutations of SoxS. J. Mol. Biol. 344, 1-10.
7. Shah, I. M. & Wolf, R. E., Jr. (2004). Novel protein-protein interaction between Escherichia coli SoxS and the DNA binding determinant of the RNA polymerase a subunit: SoxS functions as a co-sigma factor and redeploys RNA polymerase from UP-element-containing promoters to SoxS-dependent promoters during oxidative stress. J. Mol. Biol. 343, 513-532.
8. Griffith, K. L., Shah, I. M. & Wolf, R. E., Jr. (2004). 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. Mol. Microbiol. 51, 1801-1816.
9. Shah, I. M. & Wolf, R. E., Jr. (2006). Sequence requirements for Lon-dependent degradation of the Escherichia coli transcription activator SoxS: identification of the SoxS residues critical to proteolysis and specific inhibition of in vitro degradation by a peptide comprised of the N-terminal 21 amino acid residues. J. Mol. Biol. 357, 718-731.
10. Shah, I. M. & Wolf, R. E., Jr. (2006). Inhibition of Lon-dependent degradation of the Escherichia coli transcription activator SoxS by interaction with "soxbox" DNA or RNA polymerase. Mol. Microbiol. 60, 199-208.
11. Griffith, K. L., Becker, S. M. & Wolf, R. E., Jr. (2005). Characterization of TetD as a transcriptional activator of a subset of genes of the Escherichia coli SoxS/MarA/Rob regulon. Mol. Microbiol. 56, 1103-1117.
12. Griffith, K. L., Fitzpatrick, M. M. & Wolf, R. E., Jr. (2009). Two functions of the C-terminal domain of Escherichia coli Rob: mediating “sequestration-dispersal” as a novel off-on switch for regulating Rob’s activity as a transcription activator and preventing degradation of Rob by Lon protease. J. Mol. Biol., http://dx.doi.org/doi:10.1016/j.jmb.2009.03.023