|Home | About | Journals | Submit | Contact Us | Français|
A transcriptional response to singlet oxygen in Rhodobacter sphaeroides is controlled by the group IV σ factor σE and its cognate anti-σ ChrR. Crystal structures of the σE/ChrR complex reveal a modular, two-domain architecture for ChrR. The ChrR N-terminal anti-σ domain (ASD) binds a Zn2+ ion, contacts σE, and is sufficient to inhibit σE-dependent transcription. The ChrR C-terminal domain adopts a cupin fold, can coordinate an additional Zn2+, and is required for the transcriptional response to singlet oxygen. Structure-based sequence analyses predict that the ASD defines a common structural fold among predicted group IV antiσs. These ASDs are fused to diverse C-terminal domains that are likely involved in responding to specific environmental signals that control the activity of their cognate σ factor.
Singlet oxygen (1O2), a potent oxidant that damages cellular biomolecules and can kill cells (Davies, 2005; Seis and Menck, 1992), is a byproduct of light-induced reactions, so photosynthetic cells harbor carotenoid pigments to quench this reactive oxygen species (Cogdell et al., 2000). Although cellular responses to 1O2 exist in prokaryotes (Anthony et al., 2005; Glaeser and Klug, 2005) and eukaryotes (Kochevar, 2004), molecular analysis has been limited. In the photosynthetic bacterium Rhodobacter sphaeroides (Rsp), the group IV σ factor σE is essential to mount a transcriptional response to 1O2 and for viability when carotenoids are limiting (Anthony et al., 2005).
Group IV σ factors are alternative σs that direct transcription of regulons in response to environmental conditions (Helmann, 2002). The group IV σs are related to the well-characterized σ70 from Escherichia coli (Campbell et al., 2003; Gruber and Bryant, 1997; Lonetto et al., 1992). Among the σ70-family members, the group IV subfamily is the largest and most diverse (Helmann, 2002), suggesting it plays a major role in the bacterial transcriptional program. The activity of many alternative σs is negatively regulated by a specific anti-σ factor. These anti-σs are strikingly diverse in sequence, structure, and mechanism (Campbell and Darst, 2005), reflecting the wide range of signals and regulatory mechanisms that control anti-σ function.
Rsp σE activity is inhibited by ChrR (Newman et al., 2001), a member of the ZAS (Zn2+ Anti-σ) subfamily of group IV anti-σs (Paget et al., 2001). Here, we investigate the structural and functional basis for inhibition of Rsp σE by ChrR and for the relief of this inhibition by 1O2. We show that ChrR comprises two structural and functional modules. The ChrR N-terminal domain is sufficient to inhibit σE-dependent transcription, whereas the C-terminal domain is needed for ChrR to release σE in response to 1O2. Surprisingly, we find that the N-terminal domains of ChrR and the E. coli (Ec) anti-σE factor RseA (Campbell et al., 2003) are similar in structure, leading us to term this the antisigma domain (ASD). Comparative bioinformatics indicates that group IV σs represent the largest group of proteins in the σ70 superfamily and that more than one-third of these σs are regulated by a protein containing a related ASD. These structurally conserved group IV ASDs are fused to a wide variety of C-terminal domains that likely detect signals to control the activity of their cognate σ factor.
Limited proteolysis of the σE/ChrR complex, combined with N-terminal sequencing and mass spectrometry, identified a ten-residue C-terminal truncation of ChrR (ChrR203) that still bound and inhibited σE. The σE/ChrR203 complex was coexpressed in Ec, and the purified heterodimeric complex crystallized in space group P21, with four 40.7 kDa complexes per asymmetric unit. These crystals diffracted to 2.6 Å resolution but had high mosaicity (Table 1). Crystals of the selenomethionyl σE/ChrR203 complex were analyzed (Hendrickson et al., 1990), but the Se substructure could not be solved, probably due to the high mosaicity.
Secondary structure predictions (Rost et al., 1994) led us to produce another C-terminal truncation of ChrR, ChrR195. The σE/ChrR195 complex crystallized in space group I222, with two 39.8 kDa complexes per asymmetric unit. The crystals diffracted to better than 2.4 Å resolution (Table 1). The structure was solved by multiwavelength anomalous dispersion (Hendrickson, 1991) using data from native crystals collected at the Zn K edge, as well as data from selenomethionyl-crystals collected at the Se K edge (Table 1). Excellent electron density was observed for one σE/ChrR195 heterodimer in the asymmetric unit (the AB heterodimer; Figure S1 in the Supplemental Data available with this article online), while the electron density for a significant portion of the CD heterodimer was “smeary” and appeared in some regions to reflect the superimposition of two slightly different conformations. Our investigations (Supplemental Data) lead us to believe that these crystals suffer an unusual twinning defect, where a part of the CD heterodimer has slightly different conformations throughout the crystal. The best refinement behavior (Table 1) was achieved by modeling the two different conformations as a slight tilt of one domain with respect to the other and treating each conformation separately with 0.5 occupancy using Refmac (Murshodov et al., 1997). Nevertheless, it was not possible to improve the Rfree beyond 0.32. In spite of this, the AB heterodimer is ordered and has excellent electron density, and its refined structure has excellent geometry (97% of residues in the most favored regions of the Ramachandran plot, no residues in disallowed regions).
The AB heterodimer from the I222 crystal form was used as a molecular replacement search model to solve the P21 crystal form. Phases from the molecular replacement solution were used to locate 32 of a possible 36 Se sites by using the anomalous signal from the single-wavelength data collected from a selenomethionyl crystal (Table 1). An excellent electron density map was obtained by combining the single-wavelength anomalous diffraction (SAD) phases with the molecular replacement phases. Cycles of iterative model building and crystallographic refinement converged to an R/Rfree of 0.250/0.286 at 2.7 Å resolution (Table 1).
The structures reveal the predicted 1:1 stoichiometry of the σE/ChrR complex (Figure 1A). ChrR makes extensive contacts with σE (σE domains σE2 and σE4, containing conserved regions 2 and 4, are green and yellow, respectively; Figure 1A) with a total buried surface area of 3682 Å2. The overall structures of the six crystallographically independent σE/ChrR heterodimers (two from the I222 crystal form, four from the P21 crystal form) were very similar (Figure S2).
The anti-σ ChrR comprises two distinct domains that are connected by an eight-residue, disordered linker (Figure 1A). The vast majority of the ChrR interactions with σE (95% of the buried surface area) occur through the N-terminal ChrR domain (residues 1–77, colored magenta in Figure 1A), so we term this the antisigma domain (ASD). The ChrR-ASD chelates one Zn2+ ion and forms a compact globular structure with σE.
The ChrR C-terminal domain (beyond residue 86; pink in Figure 1A) adopts the cupin fold, a β-barrel structural motif characteristic of a large protein superfamily (Dunwell et al., 2004). We thus call the ChrR C-terminal domain the cupin-like domain (CLD). The minimal cupin fold contains six β strands as well as two sequence motifs that contain three His residues and one Glu residue (conserved in the ChrR-CLD, see below), which can ligate cations or other ligands in some superfamily members (Dunwell et al., 2004). Anomalous difference maps calculated from data collected at the Zn2+ edge indicated that each of two ChrR-CLDs in the asymmetric unit of the I222 crystal form chelates a Zn2+ (the ChrR-ASD Zn2+ positions gave 19σ and 18σ anomalous difference peaks, whereas the ChrR-CLD Zn2+ positions gave 8.4σ and 6.7σ peaks; Figure S1), but in the σE/ChrR203 complex (P21 crystal form), this second Zn2+ was absent, indicating the CLD Zn2+ is not essential for complex formation (see below).
Like other group IV σ factors, Rsp σE comprises two structural domains, σE2 and σE4 (Campbell et al., 2003), separated by a flexible linker. Superimposition of Rsp σE2 with E. coli σE2 (Campbell et al., 2003) yields a root-mean-square deviation (rmsd) of 1.44 Å over 53 aligned α carbons, indicating that Rsp σE2 adopts a fold common to all structurally studied σ factors (Campbell et al., 2002b, 2003; Li et al., 2002; Murakami et al., 2002; Sorenson et al., 2004; Vassylyev et al., 2002). Structural studies of liganded or unliganded σ4 (Campbell et al., 2002b, 2003; Sorenson et al., 2004; Murakami et al., 2002; Vassylyev et al., 2002) reveal a conserved structural core containing a helix-turn-helix (HTH) DNA binding motif (Figure 1B). The interaction of σ4 with the anti-σ and appropriator AsiA, however, induces a dramatic structural rearrangement of σ4 (Figure 1B; Lambert et al., 2004). Interestingly, there is also a major structural rearrangement of Rsp σE4 in the σE/ChrR complex (Figure 1B). Whereas the first two helices (H1 and H2, region 4.1, orange in Figure 1B) of Rsp σE4 align with the same region of Ec σE4 (1.24 Å rmsd over 18 α carbons), the hydrophobic core that maintains interactions between regions 4.1 and 4.2 is disrupted, and the first helix of the region 4.2 HTH motif (H3) is unfolded in the σE/ChrR complex (Figure 1B). In addition, the interactions between the second HTH-motif helix (H4) and region 4.1 are lost, and H4 is reoriented with respect to region 4.1 (Figure 1B). Comparison of the σE/ChrR and σ4/AsiA structures shows that the ChrR-ASD and AsiA are unrelated in both structures and in the way they interact with σ4, and the distortions induced in σ4 are different as well (Figure 1B).
The ChrR-ASD sterically occludes core RNAP binding determinants in both σE2 and σE4, explaining how ChrR prevents σE from binding RNAP (Anthony et al., 2004). Specifically, H4 of the ChrR-ASD occludes the RNAP β ′-coiled-coil binding surface of σE2, whereas H3 of the ChrR-ASD occupies a groove between the two region 4.1 helices of σE4 where the RNAP β-flap-tip-helix would bind (Murakami et al., 2002; Vassylyev et al., 2002). The structure of the σE/ChrR complex is also consistent with the findings from Rsp σE region 2.1 mutant proteins that alter the sensitivity to ChrR inhibition (Anthony et al., 2004). Similarly, the Streptomyces coelicolor (Scoel) ZAS RsrA also interacts with σ2 of its cognate σ, σR (Li et al., 2002).
Zn2+ is required for ChrR to inhibit σE-dependent transcription, and a potential metal-binding motif in the ASD (His31X3Cys35X2Cys38) was identified (Newman et al., 2001). The σE/ChrR structure reveals that the ChrR-ASD coordinates Zn2+ via side chains of His31, Cys35, Cys38, and His6, the latter being in an N-terminal loop preceding the first α helix (H1) of the ChrR-ASD (Figure 2A). Zn2+ binding mediates interactions between the N-terminal loop and the first three α helices of the ChrR-ASD and appears to help maintain the overall fold of this domain.
To confirm the structural identification of the ChrR-ASD Zn2+ ligands, and to test their role in anti-σ function, we generated single Ala substitutions at each of these residues (H6A, H31A, C35A, and C38A) as well as a mutant protein in which all four residues were converted to alanine (ChrR4A). When the ability of these ChrR mutant proteins to inhibit σE activity was tested in an Ec tester strain (Newman et al., 1999), we found that collectively all four Zn2+ ligands, as well as individual side chains at His6, Cys35, or Cys38, were required for ChrR anti-σ function (reporter gene activity was high; Figure 2A). In contrast, the ChrR-H5A protein inhibited σE activity, suggesting that this side chain does not contribute to Zn2+ ligation, as expected from the structure. When Ser substitutions were generated at Cys35 or Cys38, they either reduced or abolished Zn2+ binding (Newman et al., 2001). Thus, it would appear that the inability of these Ala-substituted ChrR proteins to inhibit σE activity in vivo reflects a defect in binding the N-terminal Zn2+, which is required for anti-σ function.
Surprisingly, the side chain of His31 was not required for ChrR to inhibit σE activity (Figure 2A). Most Zn2+ metallo-proteins bind Zn2+ with four protein side-chain ligands (His, Cys, Asp, or Glu), but some can bind the metal with fewer protein ligands if water or some other molecule provides additional coordination (Alberts et al., 1998; Harding, 2001). Thus, it appears that Zn2+ can bind to the ChrR-ASD when only three of the four normal ligands (His6, Cys35, and Cys38) are present. Similarly, Scoel RsrA, another ZAS anti-σ, can bind Zn2+ when the corresponding residue of its Zn2+-binding motif (His37) is changed to Ala (Zdanowski et al., 2006).
In the σE/ChrR complex, the ChrR-ASD makes extensive contacts with σE (Figure 1A). To test if the ChrR-ASD is sufficient to inhibit σE activity, we assayed the ability of a truncated ChrR protein (comprising residues 1–85, or ChrR85), which lacks the CLD to inhibit σE activity. We found that cells containing ChrR85 had low σE activity in the Ec tester strain (Figure 2A). We also observed a decrease in the amount of transcript from the σE-dependent rpoE P1 promoter as concentrations of ChrR85 were increased, while transcript levels from a control σ70-dependent promoter (RNA1) were unaffected (Figure 2B). Inductively coupled plasma-mass spectrometry (ICP-MS) indicated that ChrR85 contained 0.72 (~1) molar equivalents of Zn2+ to protein and insignificant (<0.012 equivalents) amounts of other metals when compared to storage buffer. Thus, we conclude that the ChrR-ASD is both necessary and sufficient to bind Zn2+ and inhibit σE activity in vivo and in vitro.
We were surprised to find two molecules of Zn2+ in each σE/ChrR195 complex (I222 crystal form; Figure S1C), because previous analyses indicated one mole of Zn2+/mole ChrR (Newman et al., 2001). The CLD residues coordinating the Zn2+ correspond to conserved residues of the cupin fold: His141, His143, Glu147, and His177 (Figure 3A). The presence of Zn2+ in the ChrR-CLD of both complexes within the asymmetric unit of the σE/ChrR195 crystals was indicated by anomalous difference maps (Figure S1C).
We found that σE/ChrR complex prepared in buffers containing DTT had only one Zn2+ in the ChrR-ASD (σE/ChrR230, P21 crystal form), whereas complexes not exposed to DTT had two Zn2+ (one each in the ChrR-ASD and ChrR-CLD [Supplemental Data]). DTT can compete with protein for Zn2+ chelation (Gracy and Noltmann, 1968).
We were able to dialyze a second Zn2+ ion into σE/ChrR complexes containing ~1 metal equivalent by using buffer supplemented with β-mercaptoethanol. Dialysis of σE/ChrR complexes containing ~1 Zn2+ equivalent against buffer supplemented with β-mercaptoethanol and either 100 μM FeSO4, CuSO4, or MnSO4 led to significant protein precipitation with only ~1 Zn2+ equivalent in the remaining soluble complexes, even when using anoxic buffers to prevent formation of metal or protein oxidation products (data not shown). We conclude that the ChrR-ASD Zn2+ does not exchange easily with solution, whereas Zn2+ coordinated in the ChrR-CLD can be exchanged with solution in the presence of DTT. The ChrR-ASD-Zn2+ is not solvent exposed, but the ChrR-CLD-Zn2+ is solvent accessible via a small channel (Figure 3B), perhaps explaining why it is exchangeable in the presence of DTT. Thus, past (Newman et al., 2001) and current results indicate that Zn2+ binding to the ChrR-ASD is required for anti-σ function, but the physiological function of metal binding to the ChrR-CLD remains to be determined.
The ability of ChrR85 to inhibit σE activity led us to test the role of the ChrR-CLD in the transcriptional response to 1O2 (Anthony et al., 2005) by monitoring β-galactosidase (β-gal) synthesis from the σE-dependent rpoE P1::lacZ reporter gene before and after exposure to 1O2. Prior to generating 1O2, β-gal synthesis was low in wild-type Rsp cells, in cells lacking the rpoEchrR operon (TF18), or in cells containing ChrR85 as the sole source of ChrR (TD284; Table 2), showing that ChrR85 inhibits σE activity in Rsp.
When photosynthetic cells containing ChrR85 as their only source of ChrR are exposed to O2 in the light, there is no increase in σE activity (Table 2). The cells continue to grow under these conditions because the carotenoids within the photosynthetic apparatus quench 1O2 (Anthony et al., 2005). This indicates that the ChrR-CLD is required for the transcriptional response to 1O2 seen in cells containing a wild-type ChrR (Table 2).
When aerobically grown cells (limited for carotenoids) containing ChrR85 are exposed to 1O2, there is also no increase in σE activity (Table 2). In addition, exposure of aerobic cells containing ChrR85 to 1O2 is bacteriostatic (Table 2), providing another indication that the ChrR-CLD is required for the transcriptional response of cells to this reactive oxygen species.
In the σE/ChrR complex, interactions between the ChrR-ASD and the ChrR-CLD bury a surface area of 1354 Å2. The ChrR-ASD contacts the ChrR-CLD through ASD residues in the N-terminal loop and helices H1 and H2 (Figure 4A). This interface involves mostly van der Waals interactions (Figure 4A); polar interactions are restricted to hydrogen bonds between the side chain of ChrR-ASD-Glu22 and ChrR-CLD-Tyr97(OH) and the main-chain nitrogen of ChrR-CLD-Gly122, as well as a network of hydrogen bonds with two of the ChrR-ASD Zn2+ ligands, His6 and His31, plus adjacent residues (Figure 4B). Interactions between the ChrR-ASD and the ChrR-CLD involving the ChrR-ASD Zn2+ ligands include van der Waals contacts between His6 and Ile168, between His31 and Leu150, and a hydrogen bond between His31(Nδ1) and the main-chain carboxyl of Gly166 (Figure 4B).
Given the amino acid sequence diversity among group IV anti-σs, we were surprised that a search for proteins structurally related to the ChrR-ASD (http://www.ebi.ac.uk/dali/; Holm and Sander, 1996) yielded the N-terminal domain of the Ec anti-σE factor RseA (Campbell et al., 2003). Optimization of the alignment revealed that ChrR-ASD helices 1–3 align with RseA helices 1–3 with an rmsd of 1.4 Å over 33 atoms (Figure 5A). This degree of similarity suggests that these two group IV anti-σs contain a common structural motif (the ASD). H4 of the Rse-ASD and ChrR-ASD each occlude the RNAP β′-coiled-coil binding determinant on σE2, and its position is similar with respect to each cognate σE (Figure 5B). This suggests that the ASD-H4 interactions have been conserved across group IV anti-σs to block core RNAP binding (see below). However, the disposition of H4 of each ASD is different relative to H1–H3 (Figure 5A). In addition, the RseA-ASD does not contain a metal binding motif and does not require Zn2+ for maintaining its tertiary structure.
The intramolecular contacts between ChrR-ASD helices 1–3 contain only seven contacts, excluding interactions of the Zn2+ ligands. This is in contrast to the RseA-ASD, where there are 45 contacts between helices 1 and 3, with 40 of them being nonpolar. The number of intramolecular contacts between ChrR helices 1 and 3 increases to 24 when the contacts conferred by the ligands to Zn2+ and to other residues are included. Thus, the fold of helices 1–3 in the N-terminal domain of RseA is maintained by a well-packed hydrophobic core, whereas ChrR uses Zn2+ coordination to overcome the lack of a hydrophobic core. Loss of Zn2+ from ChrR causes aggregation and precipitation, accompanied by loss of anti-σ function (Newman et al., 2001), presumably because the ChrR-ASD cannot maintain its tertiary structure without this metal. Thus, it appears that the ChrR-ASD-Zn2+ ion plays a structural role, although we cannot rule out the possibility that the Zn2+ may also serve a functional role in detecting the 1O2 signal (see Discussion). The interactions of helices 1–3 in the ChrR- and RseA-ASDs with their cognate σs are not conserved. For instance, the structure of Ec σE4 is not distorted when bound to RseA, whereas the fold of Rsp σE4 is distorted when bound to ChrR (Figure 1B). Thus, although the secondary and partial tertiary structure of helices 1–3 of the ChrR- and RseA-ASDs are conserved, the amino acid sequences and the nature of the contacts with their cognate σs diverge.
The finding that the ASDs of Rsp ChrR and Ec RseA are conserved in structure despite only 21% sequence identity (over 63 aligned residues; Figure 5C) suggests that this fold may exist in other group IV anti-σs. We applied bioinformatics and visualization analysis (Sofia et al., 2001) to numerate the occurrence of group IV σs and to ask if this conserved ASD structure is present in other predicted members of this anti-σ family.
Using PSI-BLAST searches (Altschul et al., 1997), we detected ~6000 σ70-family members in the NCBI nonredundant protein database, 60% of which (~3600) cluster with group IV σs. The entire set of σ70-family proteins was used to extract 11,484 neighbor proteins from Gen-Bank, providing a database for sequence similarity searching against Rsp ChrR-ASD (residues 1–81) and Ec RseA-ASD (residues 1–90). These searches predicted 1265 structurally related sequences, including known anti-σs not previously recognized as containing an N-terminal ASD, such as FecR (Braun et al., 2003) and PrtR (Burger et al., 2000). The searches also uncovered an unusual group of serine-threonine kinases and many ZAS anti-σs with the conserved HisX3CysX2Cys Zn2+-binding motif in an N-terminal ASD (Figure 6 and Table S1). The genes encoding each of the 1265 predicted ASDs are directly adjacent to a group IV σ except for one, which is encoded on the opposite strand from a group III σ. Most σ/anti-σ pairs (1208 cases, 95%) are in a likely operon with the ASD-encoding gene downstream of the group IV σ.
Alignment of representative sequences (Figure 6) reveals conserved residues that correlate both with predicted secondary structure of the ASD and with the three-dimensional structural elements from ChrR and RseA. Thus, our results predict that ~33% of the group IV σs (~1265/3600) are regulated by anti-σs that contain an ASD (Table S1). The ASD is typically predicted to be N-terminal, cytoplasmic, and to precede a transmembrane-spanning region (912 cases, 72%), most often as a single transmembrane segment (857 cases). However, 248 ASD-containing proteins are not predicted to contain a membrane-spanning domain, including 73 ChrR homologs, 115 proteins < 150 residues in length (like Scoel RsrA), 9 of 11 predicted serine-threonine kinases, and various other family members. The ZAS motif (HisX3CysX2Cys) is found in the ASD of 38% of the potential group IV anti-σs (482 out of 1265), and it is overrepresented in predicted cytoplasmic proteins, with 92% (227 out of 248 cases) compared to 25% (255 out of 1017 cases) of those containing membrane-spanning C-terminal domains.
The Rsp group IV σ factor σE controls a transcriptional response to 1O2, a damaging byproduct of photosynthesis (Anthony et al., 2005). The activity of Rsp σE, in turn, is regulated by an anti-σ factor, ChrR, which binds σE and prevents it from binding the core RNAP (Anthony et al., 2004). We have solved crystal structures of the Rsp σE/ChrR complex and we have performed in vivo and in vitro studies to probe the basis for both ChrR anti-σ function and the response to 1O2. In addition to the structural and functional details of the σE/ChrR complex, two broad conclusions emerge from this work. First, the anti-σ ChrR is modular in both its structure and function, with an N-terminal ASD responsible for binding and sequestering σE and a C-terminal domain required for the transcriptional response. Second, the N-terminal ASD and the overall structure/function modularity are properties common to a large family of group IV anti-σs.
The modularity of ChrR structure and function is illustrated by the fact that a truncated ChrR protein comprising the N-terminal ASD (ChrR85) is sufficient for inhibiting σE activity in vivo and in vitro. Recurring themes of σ/anti-σ interactions are that anti-σs block core RNAP binding determinants of the σ, that the interactions tend to occur through multiple σ domains, and that the promoter binding determinants within σ2 and σ4 are held in a conformation not amenable for RNAP holoenzyme formation (Campbell et al., 2003; Campbell and Darst, 2005; Campbell et al., 2002a; Sorenson et al., 2004). These themes bear true in the structure of the Rsp σE/ChrR complex.
The behavior of ChrR85 indicates that the C-terminal CLD of the anti-σ is necessary for activation of σE by 1O2. We can envision several reasons why cells containing ChrR85 fail to mount a transcriptional response to 1O2. One possibility is that amino acid side chains or a ligand (Zn2+ or some other compound) in the ChrR-CLD is a target for an unknown chemical modification by 1O2 that leads to dissociation of the σE/ChrR complex. In this model, the signal for dissociation of the σE/ChrR complex could be communicated to the ChrR-ASD through contacts with the ChrR-CLD. In this scenario, cells containing ChrR85 are unable to mount a transcriptional response to 1O2 because the truncated protein lacks a critical target for 1O2 modification.
Another possibility is that the σE/ChrR85 complex is not properly localized. In photosynthetic cells, 1O2 is generated by integral membrane proteins of the photosynthetic apparatus (Cogdell et al., 2000). The high reactivity of 1O2 makes it likely that it will not diffuse from the membrane to the bacterial cytoplasm (Kochevar, 2004). If the ChrR-CLD plays a role in promoting association of the σE/ChrR complex with the photosynthetic membrane, then the properties of cells containing ChrR85 could reflect a defect in subcellular localization. Experiments to test the predictions of these models in vivo and in vitro, and to determine how 1O2 promotes dissociation of the σE-ChrR complex, are in progress.
These models do not preclude a role for Zn2+ release from the ChrR-ASD in promoting dissociation of the σE/ChrR complex. Precedent for such a role exists because oxidants remove Zn2+ from the Scoel anti-σ factor RsrA, causing dissociation of the σR/RsrA complex (Paget and Buttner, 2003). In addition, activity of the bacterial chaperone Hsp33 increases when Zn2+ is removed by reactive oxygen species (Ilbert et al., 2006). Thus, Zn2+ release from the ChrR-ASD could promote unfolding of this domain, thereby releasing σE.
Unlike Scoel σR, which is activated by superoxide, hydrogen peroxide, or diamide in vivo or in vitro (Kang et al., 1999), sustained increases in Rsp σE activity are not caused by these compounds (Anthony et al., 2004). Thus, there must be some selectivity of the cognate ZASs for different reactive oxygen species. It is possible that the surface accessibility or chemical nature of the Zn2+ ligands contributes to the reactivity with different reactive oxygen species. For example, Scoel RsrA oxidation induces a disulfide bond between two of the Zn2+ ligands (Cys11 and Cys44; Zdanowski et al., 2006), but ChrR has a histidine side chain (His6) as a metal ligand at the equivalent position to RsrA-Cys11. Thus, the inability of the ChrR-ASD to form a similar disulfide bond could explain why ChrR is insensitive to reactive oxygen species other than 1O2 (Newman et al., 2001; Anthony et al., 2004).
The presence of an enzyme-like pocket that harbors the ChrR-CLD Zn2+ (Figure 3B) also leads one to consider that this domain has some activity, possibly related to the stress response. Catalytic Zn2+ ions tend to be coordinated by His, Asp, and Glu (the ChrR-CLD Zn2+ has 3His and 1Glu ligands), whereas structural Zn2+ ions are preferentially coordinated by Cys (Alberts et al., 1998). It is also worth noting that the free thiol (Cys189) of the ChrR-CLD along with the four ChrR-CLD Zn2+ ligands (His141, His143, Glu147, and His177) are each solvent exposed (Figure 3B), making them prime targets for 1O2 modification (Davies, 2005). Finally, the structure of the CLD is unchanged in the absence of the Zn2+, indicating that metal binding is not needed for maintaining its fold. Studies are in progress to determine the biological activity and significance of the ChrR-CLD and the role of Zn2+ or other ligands.
Structural homology exists between the ChrR-ASD and another Group IV anti-σ, Ec RseA, despite sequence similarity so weak as to be undetectable (Figure 5). Extension of this analysis allows us to predict that the N-terminal ASD, and the structure/function modularity found in ChrR, is a conserved property of a diverse group of anti-σs that may regulate ~35% of all group IV σs (~20% of all σ70 family members). Nearly 40% of these group IV anti-σs contain the ZAS motif and are thus predicted to bind Zn2+. The presence or absence of a ZAS motif in the ASD, combined with its fusion to different classes of soluble or membrane-spanning C-terminal signaling domains, appears to have produced a versatile suite of transcriptional regulators that have permeated broadly among bacteria. We propose that the various C-terminal domains that have been fused to these ASDs allow bacteria to respond to the many environmental stresses encountered in nature.
The Rsp rpoEchrR operon was cloned from pJDN48 (Newman et al., 2001) into a pET28A (Novagen) derivative that contains a precision protease-cleavable His6 tag. The resulting plasmid contains a His6-tagged rpoE in an operon fusion with chrR under the control of an inducible T7 promoter. Plasmids encoding σE/ChrR203 or σE/ChrR195 complexes were created similarly.
The plasmid containing the σE/ChrR203 operon was transformed into Ec BL21 (DE3) (Novagen). Expression and purification of the complex were as described in detail in the Supplemental Data. Briefly, the σE/ChrR203 complex was purified to homogeneity in the following steps: 1) metal-chelating chromatography, (2) gel filtration chromatography, and (3) ion-exchange chromatography. The purified complex was then concentrated by centrifugal filtration to 12 mg/ml. Selenomethiony-substituted (SeMet) protein was prepared by suppression of methionine biosynthesis (Doublie, 1997). Expression and purification of SeMet protein was as above.
The σE/ChrR195 complex was expressed and purified as described for σE/ChrR203 but with modifications (see Supplemental Data). Most importantly, this sample was never treated with DTT, but 10 mM β-mercaptoethanol was used throughout the purification.
Crystals of σE/ChrR203 (Form II, Table 1) were grown by vapor diffusion at room temperature with crystallization solution (0.1 M MES [pH 6.5], 0.2 M ammonium sulfate, and 14%–18% polyethylene glycol mono-methyl ether 5000 [PEG5KMME]). Crystals were prepared for cryo-crystallography by serial transfers in six steps to 0.1 M MES (pH 6.5), 25% PEG5KMME, 10 μM ZnCl2, 5 mM DTT, and 10% ethylene glycol, then frozen into liquid ethane.
A SeMet data set (Table 1) was collected at the peak wavelength of the fluorescence emission spectrum (0.9798 Å). The data were processed with Denzo and Scalepack (Otwinowski and Minor, 1997). We were unable to solve the Se substructure from these data, possibly due to the high mosaicity. The structure was solved by molecular replacement using a model of the σE/ChrR195 heterodimer (see below) as a search model in CNS (Adams et al., 1997). The Se and Zn2+ sites were then located by using difference Fourier methods. The resulting SAD phases were combined with the molecular replacement phases for the initial electron density map. The molecular replacement model was optimized through iterative rounds of refinement and model building against the SeMet amplitudes and SIGMAA-weighted phase combination with CNS and O (Jones et al., 1991). Analysis of the final model using PROCHECK (Laskowski et al., 1993) showed no residues in disallowed regions of the Ramachandran plot (87.2% in most favored regions, 11.7% in additional allowed regions, and 1% in generously allowed regions) and an overall G factor of 0.07.
Crystals of σE/ChrR195 (Form I, Table 1) were grown by vapor diffusion at room temperature with crystallization solution (0.1 M sodium acetate, [pH 5.0] and 1.8–2 M sodium formate). Crystals of the SeMet-labeled complex grew in a crystallization solution of sodium acetate (pH 5.0) and 1.6–1.9 M sodium formate. Crystals were prepared for cryocrystallography by a quick soak in 0.1 M sodium acetate (pH 5.0), 6 M sodium formate, and 1 mM TCEP, then frozen in liquid ethane.
Native II datasets (Table 1) were collected at the peak and inflection of the Zn2+ fluorescence spectrum. The SeMet dataset was collected at the peak, inflection, and remote wavelengths. A higher resolution native dataset (NativeI) was also collected at the Zn2+ fluorescence peak. Using the Semet and NativeII datasets, 15 Se and 4 Zn2+ sites were located within the asymmetric unit (which contained two σE/ChrR195 heterodimers) using SnB (Weeks and Miller, 1999). Heavy atom refinement, phasing, and density modification calculations using SHARP (de La Fortelle et al., 1997) yielded an interpretable electron density map. Refinement against the NativeI amplitudes was initially done with CNS, and later with REFMAC (Murshodov et al., 1997). The electron density maps unambiguously defined the A/B dimer; however, the maps were not clear for parts of the C/D dimer. Our investigations (see Supplemental Data) suggest an unusual type of twinning, in which the C/D dimer has two similar but distinct conformations. Refinement of the model containing one A/B and the two C/D dimers, each with 0.5 occupancy, partly improved the electron density maps and lowered the Rcryst and Rfree to 0.28 and 0.32, respectively.
Plasmid pJDN48 (Newman et al., 2001) was used to generate and confirm the sequence of mutant ChrR proteins (Table 2). Point mutations were generated via QuikChange site-directed mutagenesis (Stratagene, La Jolla, CA). ChrR85 was generated by placing a stop codon at the 86th codon of chrR. To assay function of ChrR85 in Rsp, the chrR-85 allele was placed downstream of a wild-type rpoE gene (pRSG35), cloned into pJDN18 (Anthony et al., 2004; Newman et al., 1999), transformed into S17-1, and mated into Rsp TF18 containing a low-copy rpoE P1::lacZ reporter plasmid, pJDN30 (Anthony et al., 2004; Newman et al., 1999). ChrR85 was purified as a C-terminal His-tag protein after cloning chrR-85 into pET37b (Novagen) by using primers creating an NdeI site upstream of the chrR start codon and a HindIII site at the chrR-85 stop codon. The PCR product was digested, ligated into pET37b that was digested with the same enzymes (pRSG41), and sequenced to confirm construction of the desired gene. After pRSG41 was transformed into BL21(DE3), expression of ChrR85-His6 was induced, and the protein was purified (Anthony et al., 2004).
Metal content of ChrR proteins was determined by ICP-MS, and ChrR protein concentrations were determined either by previously determined extinction coefficients (Newman et al., 2001) or by Bradford assays. Experiments to test the effects of 1O2 on ChrR function (Anthony et al., 2005) are described in the Supplemental Data.
Recombinant σE, wild-type ChrR, and Rsp core RNAP were purified as previously described (Anthony et al., 2003; Newman et al., 2001; Newman et al., 1999). ChrR was incubated with 70 nM RpoE-His (Anthony et al., 2004) in transcription buffer (40 mM Tris-HCl [pH 7.9], 200 mM KCl, 10 mM Mg acetate, 1 mM DTT, and 62.5 mg/ml acetylated BSA) and pJDN34 (Newman et al., 2001) for 30 min at 30°C. Next, ~70 nM Rsp core RNAP was added and incubated for 30 min before transcription reactions were initiated and product analyzed (Anthony et al., 2004; Anthony et al., 2005).
Details of the bioinformatic analyses are described in the Supplemental Data.
Supplemental Data include Supplemental Results, Supplemental Experimental Procedures, three figures, and one table and can be found with this article online at http://www.molecule.org/cgi/content/full/27/5/793/DC1/.
Results were derived in part from work performed at Argonne National Laboratory, Structural Biology Center beamline 19ID at the Advanced Photon Source, and beamlines X25 and X9A at the National Synchrotron Light Source at Brookhaven National Laboratory. We thank the beamline staff at these facilities for their support. Argonne is operated by UChicago Argonne, LLC, for the U.S. DOE, Office of Biological and Environmental Research under contract DE-AC02-06CH11357. Financial support for the National Synchrotron Light Source beamline X25 comes principally from the U.S. DOE Offices of Biological and Environmental Research and of Basic Energy Sciences, and from the NIH/NCRR. During this work J.R.A was supported by a Louis and Elsa Thomsen Distinguished Graduate Fellowship from the College of Agricultural and Life Sciences and the University of Wisconsin Foundation. This work was supported in part by the DOE ASCR project in Data-Intensive Computing for Complex Biological Systems (H.J.S.), NIH GM075273, DOE DE-FG02-05ER15653, UW-Madison Hatch WIS04951 (T.J.D), and NIH GM053759 (S.A.D.).
Structure coordinates and structure factors from the form I and form II crystals have been deposited in the Protein Data Bank under ID codes 2Q1Z and 2Z2S, respectively.