Key features of σS required for specific recognition by Crl, a transcription factor promoting assembly of RNA polymerase holoenzyme Amy B. Bantaa, Robert S. Chumanovb, Andy H. Yuanc, Hueylie Lina, Elizabeth A. Campbelld, Richard R. Burgessb, and Richard L. Goursea,1 aDepartment of Bacteriology and bMcArdle Laboratory for Cancer Research, University of Wisconsin–Madison, Madison, WI 53706; cDepartment of Microbiology and Immunobiology, Harvard Medical School, Boston, MA 02115; and dLaboratory of Molecular Biophysics, The Rockefeller University, New York, NY 10065 Edited by Jeffrey W. Roberts, Cornell University, Ithaca, NY, and approved August 15, 2013 (received for review June 18, 2013) Bacteria use multiple sigma factors to coordinate gene expression connected by flexible linkers (3). The most highly conserved do- in response to environmental perturbations. In Escherichia coli and main, σ2 (composed of homology regions 1.2–2.4), interacts with other γ-proteobacteria, the transcription factor Crl stimulates σS- the β′ clamp helices (β′CH) in core RNAP (residues 262–309, also dependent transcription during times of cellular stress by promot- known as the β′ coiled coil) (15), and σ4 interacts with the flap ing the association of σS with core RNA polymerase. The molecular domain of β (16). Free sigma factors undergo conformational basis for specific recognition of σS by Crl, rather than the homol- changes to associate with core RNAP and then undergo further ogous and more abundant primary sigma factor σ70, is unknown. conformational changes in the context of holoenzyme to bind to the Here we use bacterial two-hybrid analysis in vivo and p-benzoyl- promoter and to melt the DNA in the vicinity of the −10 element phenylalanine cross-linking in vitro to define the features in σS and transcription start site (15, 17–20). S responsible for specific recognition by Crl. We identify residues Direct interaction of Crl and σ in E. coli and Salmonella S S enterica fi in σ conserved domain 2 (σ 2) that are necessary and sufficient was reported previously in vitro by gel ltration (14), to allow recognition of σ70 conserved domain 2 by Crl, one near surface plasmon resonance (SPR) (12), and bacterial two-hybrid the promoter-melting region and the other at the position where (BTH) analysis (12, 21). BTH analysis further suggested that a large nonconserved region interrupts the sequence of σ70.We S. enterica Crl interacts with fragments containing conserved σS σS then use luminescence resonance energy transfer to demonstrate domain 2 of ( 2) (21). Although most activators interact with directly that Crl promotes holoenzyme assembly using these spec- sigma by recognizing domain 4 (reviewed in refs. 2, 22), two ificity determinants on σS. Our results explain how Crl distin- activators exhibiting no sequence similarity to Crl [Chlamydia guishes between sigma factors that are largely homologous and trachomatis GrgA (23) and Streptomyces coelicolor RbpA (24)] activates discrete sets of promoters even though it does not bind were recently shown to interact with domain 2 of the primary to promoter DNA. sigma factor, suggesting that such interactions may be more widespread than appreciated previously. E. coli σS RNAP formation | transcription initiation | bacterial stress response | Here we investigate the interaction between and Crl fi RpoS | curli fiber to gain insight into its mechanism(s) of action. We de ne two S surface-exposed patches on σ 2 needed for specific recognition 70 S by Crl; create a σ 2–σ 2 chimera to show that these residues are ranscription initiation in bacteria requires the assembly of fi σS σ both necessary and suf cient for Crl to distinguish from other Ta sigma factor ( ) with the RNA polymerase (RNAP) cata- σ70 α β β′ sigma factors, including ; and demonstrate directly that Crl lytic core (E, composed of 2 -subunits and one each of , , and functions as a holoenzyme assembly factor by using these spec- ω σ BIOCHEMISTRY ) to form RNAP holoenzyme (E ), which in turn recognizes ificity determinants on σS. We thereby determine how this promoter sequences (1) (reviewed in ref. 2). Multiple sigma factors compete for binding to core RNAP (reviewed in refs. 3, Significance 4), and each sigma factor controls a specific set of promoters. In Escherichia coli, which has seven sigma factors, σ70 is the primary sigma, and σS is important for certain stress responses Bacteria can change the relative amounts of different sigma and during the stationary phase of growth (5). EσS-dependent factors that associate with the RNA polymerase (RNAP) cata- transcription initiation is regulated by σS, whose concentration is lytic core enzyme, thereby altering the relative amounts of the itself regulated at the levels of transcription, translation, and RNAP holoenzymes and modifying gene expression globally protein stability (reviewed in ref. 6). EσS-dependent transcription under changing nutritional or environmental conditions. We is also activated by Crl (7), a small protein that increases expres- show here that the Escherichia coli transcription factor Crl sion of many stress response genes and those required for forma- increases assembly of the holoenzyme containing the stress tion of amyloid curli fibers (which accounts for its name) involved response sigma factor, sigma S. We identify the key determi- in adhesion and biofilm formation (reviewed in refs. 2, 6, 8). nants necessary for Crl to distinguish between sigma S and S The effect of Crl on σ -dependent transcription in vivo is most other sigma factors, thereby specifically promoting the sigma S pronounced during the transition into stationary phase (9). It has transcriptional program. been proposed that Crl functions by increasing the concentration S S of Eσ holoenzyme by facilitating assembly of σ with core Author contributions: A.B.B., R.S.C., A.H.Y., R.R.B., and R.L.G. designed research; A.B.B., RNAP (10–12) because Crl’s effects on transcription are greatest R.S.C., A.H.Y., H.L., and E.A.C. performed research; A.B.B. contributed new reagents/ in vitro when the concentration of σS is lowest, and over- analytic tools; A.B.B., R.S.C., A.H.Y., H.L., E.A.C., R.R.B., and R.L.G. analyzed data; and expression of σS complements a crl deletion in vivo (13). Effects A.B.B. and R.L.G. wrote the paper. of Crl have also been reported on postholoenzyme assembly The authors declare no conflict of interest. steps including promoter binding (14) and open complex for- This article is a PNAS Direct Submission. mation (12). 1To whom correspondence should be addressed. E-mail: [email protected]. Sigma factors contain several protease-resistant domains, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. each composed of subregions defined by sequence conservation, 1073/pnas.1311642110/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1311642110 PNAS | October 1, 2013 | vol. 110 | no. 40 | 15955–15960 Downloaded by guest on September 24, 2021 S nonconventional transcriptional activator distinguishes between these variants (representing 30 different positions in σ 2) were S S sigma factors to stimulate σ -dependent gene expression. unique (Table S1). Second, we tested the 39 σ 2 variants for interaction with the β′CH fragment to identify those that were Results defective in binding to Crl but that still bound to core RNAP. E. coli σS S Crl Interacts with Conserved Domain 2 in . It was reported The interaction between σ 2 and Crl was reduced (<65% of that S. enterica σS fi S recently that Crl interacts with 2 (21). To con rm with wild-type σ 2) in 15 of the variants that still interacted with E. coli E. coli S that this interaction is conserved in ( Crl has 83% core RNAP (>51% of that with wild-type σ 2)(Fig. S2A and identity/92% similarity to S. enterica Crl, and E. coli σS has 99% Table S1). Substitutions at four positions that were not con- identity/99% similarity to S. enterica σS), we constructed a library served between σS and σ70, D87, D135, P136, and E137 (teal in S S of σ fragments based on the predicted endpoints of its stable Fig. 2A and Fig. S2A, shaded rows in Table S1), disrupted the σ 2 domains and tested their interactions with Crl using a BTH assay interaction with Crl (<40% of the wild type) without dramati- S σS β′ (25). Crl interacted with the σ 2 fragment and with larger frag- cally perturbing the 2 interaction with the CH (gray in Fig. S A fi ments encompassing σ 2, and it did not interact with fragments 2 ), suggesting that these residues represent speci c determi- S lacking σ 2 (Fig. 1). As the lack of detectable protein–protein nants for Crl binding. We also note that many other substitutions interactions in the BTH assay could be attributable to misfolding near D87 were obtained in the screen for amino acid residues S A or instability of σ 2 fusion proteins, we verified that all fragments altering Crl binding (Fig. S2 and Table S1), but these sub- of σS used in our study interacted with fragments of core RNAP stitutions also diminished the interaction with the β′CH, com- containing either the β′CH or the β-flap (Fig. S1A). plicating the interpretation of the role of these residues in We also investigated the interaction of Crl with other sigma Crl binding. S factors. Previous in vitro transcription data suggested that Crl may We constructed a structural homology model of σ 2 based on have small effects on transcription by Eσ70 and EσH (10) but an the X-ray crystal structure of σ70 (1SIG) and the alignment in 70 B C–F σS interaction between Crl and σ was not detected by SPR (12), Fig.