Growing Repertoire of AraC/XylS Activators

The AraC/XylS family of transcription activator proteins is defined by a 100-amino-acid region of sequence similarity (15) that forms an independently folding domain containing two helix-turn-helix (HTH) DNA binding motifs. A recent database search (http://hits.isb-sib.ch) (35) by the author using the AraC/XylS family profile (PS01124) (15) identified 830 family members! A number of AraC/XylS family members involved in carbon metabolism and responses to environmental stress have been intensively studied, including Ada, AraC, MarA, MelR, RhaR, RhaS, Rob, SoxS, and XylS (described below). More recently, several family members that regulate virulence factor expression in animal and plant pathogens have been well characterized. These virulence regulators are often part of complex regulatory schemes and are revealing new twists to regulation by the AraC/XylS family. Here I will briefly discuss some of the features of this large family and then focus on novel findings regarding AraC/XylS regulation of virulence factor expression, including those by Hulbert and Taylor in this issue (19a).

STRUCTURE OF AraC/XylS FAMILY PROTEINS

Detailed structures have been published for the AraC/XylS family domains of only two proteins, MarA and Rob (25, 36). The backbone structures of these domains are virtually superimposable. The Rob structure exhibited an unusual interaction with DNA in which only one of the two HTH motifs made base-specific contacts with the DNA (25). However, the DNA interaction exhibited by MarA, in which each of the HTH motifs is inserted into an adjacent major groove of the DNA, appears to occur more widely (2, 16, 17, 33, 41). This interaction of each domain with two adjacent major grooves of DNA results in an unusually long binding site of 17 to 21 bp for each AraC/XylS family monomer. Most of the characterized family members function as transcription activators, although a few exclusively or additionally function as repressors (15). In addition to their DNA binding functions, at least some, and likely most, of the AraC/XylS family domains also contain transcription activation functions (3, 4, 21, 22, 24, 29).

Among the 830 AraC/XylS family members identified in the search described above, a large majority shared common protein architectures. These proteins have the family domain at the C-terminal end of an approximately 300-amino-acid-long protein that contains at least one additional domain. Many of these multidomain proteins use their nonfamily domain to form homodimers (15). In addition, many of the multidomain proteins directly respond to a ligand, and, again, this function is performed by the nonfamily domain (15). Given the large size of this family and the modular nature of domains, it is not surprising, however, that there are many AraC/XylS family members that do not fit this most common architecture. The two most common alternative architectures each comprise approximately 5% of the family and consist of either single domain proteins or those in which the AraC/XylS family domain is located at the N terminus of a protein with at least one additional domain.

BIOCHEMICAL FUNCTIONS OF AraC/XylS FAMILY PROTEINS

One of the most noteworthy features of AraC/XylS family members is the ability of some of them to mediate DNA looping. This mechanism was first demonstrated for AraC by the Schleif laboratory (39) and functions to repress araBAD expression in the absence of arabinose. The mechanism used by AraC to form a DNA loop involves the interaction of an amino acid “arm” near the very N terminus of AraC with either the dimerization domain and arabinose, when present, or with the DNA binding domain (38). The interdomain interaction in the absence of arabinose apparently constrains the domains such that DNA looping is preferred over binding to adjacent sites. More recently, a second family member, MelR, has also been shown to use DNA looping to repress transcription, although the mechanism has not yet been demonstrated (43). While many AraC/XylS family members may not utilize DNA looping, a mechanism similar to that used by AraC could constrain the relative orientation of the domains in the absence of ligand and thereby block binding to adjacent DNA sites required for activation.

While our understanding of the mechanisms used by AraC/XylS family members to activate transcription is by no means complete, some information regarding interactions with RNA polymerase (RNAP) is available. The RNAP α-subunit C-terminal domain (α-CTD) is involved in transcription activation by a number of family members (18, 19, 20, 21, 22, 27, 37). For SoxS, MarA, and Rob, α-CTD is required for full activation regardless of the position of the activator; however, only in cases in which the activator binding site is not adjacent to RNAP is all activation lost upon deletion of α-CTD (20-22). Residues in the N-terminal domain of the RNAP α-subunit (α-NTD) important for cyclic AMP receptor protein (CRP) activation were not required for full activation by a number of AraC/XylS family members (13). However, residues near the C-terminal end of σ⁷⁰ are required for activation by a number of family members (3, 26, 28). In the case of RhaS, there is genetic evidence that a residue within the second HTH motif makes a direct interaction with an identified residue of σ⁷⁰ (3).

ToxT REGULATION IN VIBRIO CHOLERAE

The AraC/XylS family member ToxT directly activates the expression of both of the major virulence factors in V. cholerae—the toxin-coregulated pilus and the cholera toxin (7). Hulbert and Taylor (19a) have analyzed the activation of the operon encoding the toxin-coregulated pilus (tcpABCDEF) by ToxT. In a similar study, Yu and DiRita (44) analyzed the regulation of both the tcpA promoter and the cholera toxin-encoding operon (ctxAB) by ToxT. These studies identified the minimal upstream endpoints for full ToxT activation of tcpA and ctxA and showed that purified His-tagged ToxT could bind to DNA fragments containing these minimal regions. Point mutations in the tcpA promoter region that resulted in decreased expression clustered into two A-rich regions of the promoter and reduced His-tagged-ToxT binding (19a). It was found that the nucleoid protein H-NS (described below) is a negative regulator of both promoters; however, the effect of H-NS on ctxA was much greater than its effect on tcpA (34, 44). In addition, H-NS was found to negatively regulate the toxT promoter (34). His-tagged-ToxT footprinted a region from −111 to −41 upstream of ctxA and from −84 to −41 upstream of tcpA (44). Finally, Hulbert and Taylor (19a) determined that the RNAP α-CTD was required for ToxT-dependent activation of tcpA.

Based on the combined results of these studies on ToxT activation in V. cholerae, a model emerges in which H-NS blocks expression from the tcpA and ctxA promoters in the absence of ToxT activation (Fig. 1A). ToxT is able to largely overcome this repression by binding to the DNA, most likely as a dimer. At least at tcpA, the bound ToxT protein also interacts with α-CTD to activate transcription. Therefore, ToxT has both antirepression and direct activation functions. Assuming a binding site of at least 17 bp for each monomer, the extent of the DNase I footprint at tcpA suggests that one ToxT dimer binds in the −84 to −41 region. At ctxA, the footprint extended upstream to −111; however, DNA upstream of −76 was not required for full ToxT activation (44). This suggests that two dimers of ToxT may bind in the ctxA −111 to −41 region, with only the downstream dimer being required for full activation. The requirement for the upstream ToxT dimer may be eliminated by the removal of the H-NS sites upstream of −76 in the deleted promoter, or, alternatively, the upstream ToxT dimer may have a subtle role that was not detected in the experiments performed.

FIG. 1.

Mechanisms used by AraC/XylS family regulators of virulence factor expression. (A) In the absence of ToxT activation, the nucleoid protein H-NS represses tcpA and ctxA expression in V. cholerae, most likely by binding to a long stretch of DNA in the vicinity of the promoters. ToxT binding to these promoter regions presumably displaces H-NS and allows a direct interaction with the RNAP α-CTD. Each ToxT monomer is illustrated with two domains. (B) Activation by Rns of its own promoter in enterotoxigenic E. coli requires binding to sites at +83.5 as well as at −224.5. The size of the sites suggests the binding of a single Rns monomer to each and the possibility that a dimer of Rns simultaneously contacts each of these sites to form a DNA loop. How Rns binding to these two sites activates transcription remains to be determined. Each Rns monomer is illustrated with two domains. (C) Transcription activation by InvF in S. enterica serovar Typhimurium requires the chaperone protein SicA, although InvF can bind to the promoter DNAs in vitro in the absence of SicA. The InvF monomer is illustrated with two domains.

ROLE OF H-NS IN REGULATION BY AraC/XylS FAMILY MEMBERS

H-NS is an abundant, nucleoid-associated protein that binds preferentially to curved DNA (1). The finding that H-NS plays a role in ToxT-mediated regulation is similar to several other cases of AraC/XylS family regulation of virulence factors, particularly in enterobacteria (for example, see references 6, 7, 23, and 40). In general, the AraC/XylS family members antagonize repression by H-NS. In at least some cases, H-NS repression is due to the formation of a hairpin-like structure, with H-NS bridging two double-stranded regions of the DNA (8-10). H-NS is used to inform the regulons of the status of a variety of environmental conditions, such as temperature, pH, and osmolarity (1), which may allow the organisms to distinguish host versus nonhost environments.

Rns ACTIVATION REQUIRES DOWNSTREAM BINDING SITES

Another interesting AraC/XylS family regulator of virulence genes is the Rns activator of various pilin-encoding operons in enterotoxigenic Escherichia coli (5). In addition to the pilin-encoding operons, Rns activates expression of its own gene (14). Rns regulation of at least the cooB operon also involves repression by H-NS (32). Perhaps the most interesting aspect of regulation by Rns, however, is the finding that Rns activation of its own operon requires two DNA binding sites: one at −224.5 and one at +83.5 relative to the rns transcription start site (Fig. 1B) (30). This requirement by an activator for a binding site downstream of the transcription start site is quite unusual. However, it was subsequently shown that several other closely related AraC/XylS family activators (CfaR, AggR, and VirF) could also activate rns and in each case also required the downstream binding site (31). Given the high degree of amino acid identity in the HTH motifs of CfaR, AggR, VirF, and Rns, Munson and Scott used known Rns binding sites to search for potential binding sites for CfaR, AggR, and VirF, as well as CsvR. Their results suggest that CfaR and CsvR may also regulate their own expression by using downstream as well as upstream binding sites (31). It will be interesting to see whether activation from such distant DNA sites might involve DNA looping (Fig. 1B), especially given that each of the downstream and upstream sites is only long enough to accommodate a single Rns monomer. This would differ from DNA looping by AraC and MelR, since the Rns loop would mediate transcription activation rather than repression.

InvF ACTIVATION REQUIRES THE SicA CHAPERONE

Finally, regulation by InvF from Salmonella enterica serovar Typhimurium adds yet another twist to regulation by AraC/XylS family proteins. InvF regulates expression of sicA, sigD, and sopE, but, interestingly, InvF alone is not sufficient for activation of these genes (11). In addition to InvF, the SicA protein is also required for expression of these operons (Fig. 1C). SicA functions as a chaperone to several proteins secreted by the type III secretion system (42). Darwin and Miller have shown that the SicA protein copurifies with His-tagged InvF, suggesting a direct protein-protein interaction between InvF and SicA (12). Purified His-tagged InvF bound to sicA promoter DNA; however, purified His-tagged SicA did not (12). Together, these findings suggest that InvF binds to DNA as a monomer, regardless of the presence of SicA, but may only activate transcription in the presence of SicA. Whether SicA causes a conformational change in VirF that allows transcription activation, or perhaps itself interacts with RNAP, remains to be determined (12).

CONCLUSIONS

It is clear from the understanding of only a small fraction of the AraC/XylS family, that there is considerable variety in the mechanisms they use to regulate gene expression. Clearly, as illustrated by just the few examples described here, analysis of regulation of virulence-related genes by AraC/XylS family proteins has increased our understanding of the repertoire of this family of proteins. Undoubtedly, further study of these and other members of this ubiquitous family of prokaryotic proteins will uncover additional novel mechanisms for the regulation of gene expression.