Journal list menu
Crystal structure of Campylobacter jejuni peroxide regulator
Abstract
In Campylobacter jejuni (Cj), the metal-cofactored peroxide response regulator (PerR) transcription factor allows C. jejuni to respond to oxidative stresses. The crystal structure of the metalated form of CjPerR shows that the protein folds as an asymmetric dimer displaying structural differences in the orientation of its DNA-binding domain. Comparative analysis shows that such asymmetry is a conserved feature among crystallized PerR proteins, and mutational analysis reveals that residues found in the first α-helix of CjPerR contribute to DNA binding. These studies present the structure of CjPerR protein and highlight structural heterogeneity in the orientation of the metalated PerR DNA-binding domain which may underlie the ability of PerR to recognize DNA, control gene expression, and contribute to bacterial pathogenesis.
Abbreviations
BMe, beta-mercaptoethanol
MCO, metal-catalyzed oxidation
ROS, reactive oxygen species
Campylobacteriosis is an infectious disease caused by bacteria of the genus Campylobacter [1]. Campylobacters are Gram-negative curved bacteria that commensally colonize the gastrointestinal tract of an extensive number of animals [2]. They are microaerophilic and as such require an oxygen concentration below atmospheric levels (3–15%) [2]. Similar to other prokaryotes, Campylobacter species are constantly exposed to reactive oxygen species (ROS) that are either endogenously produced by the enzymes forming the respiratory chain or produced by host defense systems. ROS take multiple forms including hydrogen peroxide (H2O2), superoxide anion radicals (
), and hydroxyl radicals (·OH), the latter often formed by the reaction of H2O2 with ferrous ions [3]. Campylobacter species use several enzymes to cope with ROS. Among those, superoxide dismutase (SodB) converts superoxide into hydrogen peroxide while catalase (KatA) and the alkyl hydroperoxide reductase (AhpC) convert hydrogen peroxide and alkyl hydroperoxides into oxygen and nontoxic alcohol intermediates, respectively [4]. To trigger the expression of these enzymes upon ROS exposure, bacteria primarily use either OxyR or peroxide response regulator (PerR) to sense the presence of H2O2. In Campylobacter jejuni, mutation of perR results in a 103-fold reduction of chick gut colonization [5] and the PerR-deleted C. jejuni strain displays induced hyper-resistance to ROS [5, 6]. Transcriptome profiling of PerR regulated genes in C. jejuni shows that in addition to katA and ahpC, CjPerR represses the expression of other genes involved in oxidative stress defense including sodB [5, 7], ferritin, trxB, and dps [5].
CjPerR is a member of the Fur metalloregulator superfamily of transcription factors and initially identified as a close homolog of the C. jejuni Fur [8]. Similar to other members of the Fur family, PerR activity is controlled by regulatory metals such as Mn2+ and Fe2+ yet depending on the metal bound form, DNA target selectivity differs greatly [9]. For example, while the Mn2+ bound form of PerR leads to strong repression of all PerR regulated genes, the metalloregulator iron complex selectively represses genes helping the Bacillus subtilis to cope with peroxide stress [10]. In addition, depending on the metal bound form of PerR, its sensitivity to H2O2 will differ. For example, the iron-bound form of PerR is > 104 fold more sensitive to H2O2 inactivation when compared to PerR metallated by Mn2+ [11]. Mechanistically, the metal bound form of PerR binds to PerR boxes typically located upstream of specific transcriptional start sites to repress the transcription of genes involved in peroxide defense [12]. Upon exposure to ROS, PerR undergoes metal-catalyzed oxidation (MCO) [13]. MCO is triggered by the formation of hydroxyl radical directly reacting with two histidine residues important for the coordination of the regulatory metal ion [14]. The oxidation of these histidine residues induces the formation of 2-oxo-histidine which leads to a structural reorganization of the metalloregulator in a conformation not favorable for DNA binding; resulting in PerR-DNA disengagement, target gene derepression [14], and degradation by the LonA protease [15].
Crystal structures of PerR from B. subtilis (BsPerR) [16], Streptococcus pyogenes (SpPerR) [17, 18] and Leptospira interrogans (LiPerR) [19] revealed that all these metalloregulators adopt a caliper-like structure composed of two protomers. Despite such conserved structural organization, the presence of structural appendages on the N terminus of SpPerR [17, 18] and the lack of the structural metal-binding site in LiPerR highlights the structural diversity of PerR proteins [19]. In this study, we report the crystal structure of the Mn2+-metallated dimeric form of CjPerR. The structure shows that CjPerR DNA binding domain shows atypical structural divergence in the orientation of its first α-helix between the CjPerR dimer. We also show that CjPerR, akin to other PerR proteins, folds as an asymmetric dimer. These findings highlight a novel mechanism employed by these metal sensing transcription factors to enable their DNA binding domain to move in space; a mechanism that could help the recognition of divergent DNA-binding sequences.
Materials and methods
CjPerR purification and crystallization
CjPerR was overexpressed in Escherichia coli Rosetta cells in fusion with a Strep epitope placed at the C terminus of the transcription regulator. Following induction with 0.1 mm isopropyl β-d-1-thiogalactopyranoside for 16 h at 18 °C, Rosetta cells were centrifuged at 2560 g for 30 min at 4 °C and resuspended in Buffer A (50 mm Sodium phosphate (NaPi) pH 7.0, 500 mm NaCl, 5 mm beta-mercaptoethanol (BMe)). Cells were lysed by sonication and centrifuged at 31 000 g for 30 min at 4 °C and the supernatant was filtered using 0.45 μm filters. The filtered solution was applied to Strep-TACTIN (Novagen) beads and the column was washed with 10-column volumes of Buffer A. The fusion protein was eluted according to manufacturer's instructions and the Strep tag was removed using Tev protease during 16 h incubation at 4 °C in Buffer B (50 mm NaPi pH 7.0, 150 mm NaCl, 5 mm BMe). The protein was further purified by size-exclusion chromatography using Superdex 75 column (GE Healthcare) equilibrated in Buffer C (20 mm Tris pH 8.0, 250 mm NaCl, 5 mm BMe). CjPerR eluted as a single peak at an apparent molecular weight of ~ 32 kDa, which corresponds to a molecular weight of a CjPerR dimer. CjPerR fractions were then pooled and concentrated to 20 mg·mL−1 and excess MnCl2 (2 : 1 metal to protein molar ratio) was gradually added to the protein solution. CjPerR was then incubated with the metal ion during 16 h at 4 °C. In order to remove excess free metal ions, a second gel filtration was performed. Fractions corresponding to a protein dimer were then pooled and concentrated to 15 mg·mL−1 and several crystallization conditions were tested. CjPerR crystals were grown in 0.1 m HEPES pH 7.2–7.4, 35–60% hexylene glycol (MPD) at 5 mg·mL−1 protein concentration. Crystals were harvested in the mother liquor and flash-frozen in liquid nitrogen.
Crystal structure determination of CjPerR
A single dataset was collected at the 21ID-D beamline of the Life-Sciences Collaborative Access Team at the Advance Photon Source, Argonne National Laboratories. The CjPerR crystal structure was determined at 2.7Å resolution using single-wavelength anomalous dispersion at the zinc peak wavelength. The reflections were recorded on a single crystal at 1.2822 Å using a MarMosaic300 CCD detector (Rayonix). The dataset was subsequently processed and scaled using autoproc [20]. Using AutoSol from phenix [21] two zinc atoms were located in the asymmetric unit and a partial model (~ 68%) was automatically built. The CjPerR model was further refined using repetitive cycles of model building using coot [22] and structure refinement using phenix [21]. The final refined model has an Rwork of 22.1% and Rfree of 27.4%. molprobity [23] assigned a score of 1.61 with no Ramachandran outliers (Table 1). Assignment and placement of the zinc and manganese ions in the S1 and S2 sites respectively were based on previous structural studies of B. subtilis PerR [16].
| Space group | P212121 |
| Cell dimensions | |
| a, b, c (Å) | 48.1, 70.2, 85.2 |
| Resolution | 54–2.71 (2.72–2.71)a |
| R sym | 9.3 (58.8) |
| I/σI | 24.0 (4.9) |
| Completeness | 100.0 (100.0) |
| Redundancy | 14.2 (14.7) |
| Redundancyanom | 7.6 (7.6) |
| Atoms found | 2 |
| Site determination FOM | 29.1 |
| Density modification FOM | 68.0 |
| Resolution (Å) | 54.16/2.71 |
| Reflections | 8254 |
| Rwork/Rfree | 21.5/27.0 |
| No. atoms | |
| Protein (A/B) | 1105/924 |
| Ligands (Mn2+/Zn2+) | 2/2 |
| Water | 17 |
| B-factors (Å2) | |
| Protein (A/B) | 52.6/72.7 |
| Ligands (Mn2+/Zn2+) | 52.3/167.9 |
| Water | 44.9 |
| R.m.s. deviations | |
| Bond lengths (Å) | 0.01 |
| Bond angles (°) | 1.21 |
| Molprobity score | 1.84 |
| Ramachandran favored (%) | 100 |
| RCSB code | 6DK4 |
- aValues in parentheses refer to the highest resolution shell.
Electrophoretic mobility shift assays
Forward (5′-ATTAAAAATAATAAAAAATATTATTTAATATACTTC-3′) and reverse (5′-GAAGTATATTAAATAATATTTTTTATTATTTTTAAT-3′) Cy5-labeled primers (Integrated DNA Technologies) corresponding to a 36-bp DNA fragment of the perR promoter region and containing the 15-bp PerR box (highlighted in italics) were annealed in NEB2 buffer at 95 °C during 10 min and slowly cooled to room temperature. For the gel shift assays, 2 nm of Cy5 double-stranded DNA fragment was incubated with increasing concentrations (5, 10, 50 μm) of purified recombinant CjPerR-WT or mutants during 1 h on ice in binding buffer (20 mm Tris/HCl pH 8.0, 5% glycerol, 50 μg·mL−1 BSA, 50 mm KCl, 50 μm MnCl2, 30 μm MgCl2, 5 ng·μL−1 poly-dIdC nonspecific DNA competitor). Samples were separated on a 6% nondenaturing polyacrylamide gel (19 : 1 acrylamide/bisacrylamide) during 45 min at 100 V and 4 °C. Gels were freshly prepared with 0.5× Tris-Glycine buffer and 100 μm MnCl2 and prerun during 20 min at 150 V at 4 °C. All steps were carried out in the dark to limit the exposure of the Cy5 fluorophore to light, which results in decreased fluorescence signals. Gels were visualized using a Typhoon Trio Variable Mode phosphoimager (GE Healthcare).
Results and Discussion
Campylobacter jejuni encodes a PerR showing 35% sequence identity with the well characterized B. subtilis PerR protein. Sequence alignment of CjPerR with other PerR proteins shows that the residues forming the regulatory metal and zinc-binding site are conserved (Fig. 1 and Fig. S1). However, the CjPerR polypeptide is shorter than the majority of PerRs and its N terminus differs in length and sequence when compared to other crystallized PerRs. To understand the conformation of this divergent region of PerR and the overall structure of CjPerR, we solved its structure at a 2.7 Å resolution using single-wavelength anomalous dispersion at the zinc peak wavelength. The asymmetric unit consists of two polypeptide chains of CjPerR. As defined by the electron density, the first chain (referred to as protomer A) of CjPerR contains residues 1–92 and 94–135, while the other chain (referred to as protomer B) includes residues 18–44, 48–71, and 73–135. Each polypeptide chain is formed by two distinct domains. The N-terminal domain folds as a bundle of three α-helices (α1, α2, and α3) followed by a small antiparallel β-sheet formed by two β-strands (β1 and β2). The C-terminal domain adopts a three-stranded antiparallel β-sheet (β3, β4, and β5) in which β4 and β5 are disrupted by a 310 and an α-helix (α4) (Fig. 2A,B). Each subunit of CjPerR harbors two metal-binding sites. In the first metal-binding site (referred to as S1 or the structural site [9]), a Zn2+ ion is tetra-coordinated by four Cys residues (C89, C92, C132, and C135), while the second metal-binding site (referred to as S2 or the regulatory site [9]) contains a Mn2+ ion. Located at the interdomain region positioned between the DBD and the DD, the Mn2+ ion is pentacoordinated by three histidine residues (H30, H78, and H86) and two aspartate residues (D78 and D97) (Fig. 2C,D). Comparative analysis of the CjPerR and BsPerR structures (Fig. 3A) shows that both metalloregulators employ the same type of residues and an identical number of coordinations to stabilize the Mn2+ in their regulatory sites (Fig. 3B). In a recent study, incubation of C. jejuni with excess manganese in vivo failed to repress the expression of a reporter element under the control of a PerR box [7], while similar conditions triggered the repression of PerR regulated genes in B. subtilis [10]. However, it should be noted that BsPerR is believed to be involved in sensing the ratio between Fe/Mn in B. subtilis and directly regulates fur transcription [9, 10]. In contrast, CjPerR does not regulate fur in C. jejuni and thus may not be involved in manganese regulation in this organism [5, 24]. Nevertheless, our crystal structure suggests that manganese enables the formation of a V-shaped structure; a conformation conducive for binding DNA and repressing gene expression.
Based on binding affinities, it is believed that under most growth conditions, PerR is primarily bound to iron [10] and the iron form predominates when both iron and manganese are present at equivalent concentrations [13]. The reactivity of the iron-bound form of PerR to H2O2 varies between species and one element that influences such sensitivity may be the solvent accessibility of the regulatory site. In contrast to biochemical studies performed on BsPerR [25], our attempts to purify the nonoxidized form of CjPerR failed despite the incorporation of chelating agents in the purification buffers and extensive treatment of glassware with dithionite (data not shown). Based on our observation that CjPerR α1 protrudes out of the V-shaped dimer (Fig. 3C,D and Fig. S2), we surmise that the regulatory site might be more solvent exposed compared to other metalloregulators. This protrusion by α1 leads to the formation of new pocket in close proximity to the regulatory site. Strikingly, the regulatory metal can be observed in the surface representation of CjPerR while in BsPerR, the metal ion is completely shielded from the solvent (Fig. 3E,F). Our findings suggest that the extra space created by the movement of α1 leads to increased propensity of the protein to undergo MCO. These observations are consistent with previous findings showing that decreasing the accessibility of the solvent to the S2 metal-binding site by replacing an aspartate (D104 in BsPerR; D97 in CjPerR) for a glutamate in BsPerR protects the metalloregulator from oxidization [25]. Considering that C. jejuni is a microaerophilic bacterium, it is tempting to speculate that the increased sensitivity of CjPerR, and the reorganization of its first α-helix, renders the protein more susceptible to oxidation and thus more responsive to low ROS levels. This idea is supported by extensive alignment of PerR homologs which shows that several strict anaerobe and microaerophilic bacteria display high sequence homology with the N terminus of CjPerR, while PerR in facultative aerobe and aerobic bacteria harbors a longer N-terminal domain and resembles BsPerR.
Collectively, the structure shows that CjPerR folds as a V-shaped dimer which is characteristic of the Fur family metalloregulators. However, CjPerR folds differently from other PerR proteins. The DBDs of CjPerR contains only three α-helices (Fig. 4A) while the same domain of BsPerR [16] (Fig. 4B), SpPerR [17, 18] (Fig. 4C), and LiPerR [19] (Fig. 4D) folds as a four α-helix bundle. These differences stem from the first α-helix of CjPerR that is elongated and protrudes out of the V-shaped structure. Similar to LiPerR [19] and SpPerR [17, 18], the residues forming CjPerR protomer's DD fold as an antiparallel β-sheet created by three β-strands intersected by one α-helix. Upon dimerization, the β5 of each protomer interacts to ultimately form a six-stranded β-sheet. Such organization is not conserved across PerR proteins as BsPerR forms a four-stranded β-sheet. However, these structural differences do not prevent protein dimerization.
Alignment of the protomers forming CjPerR dimer revealed that both subunits align relatively well with a r.m.s.d. of 0.5 Å. However, while the DD aligns very well, the DBD shows some structural differences. This is not only highlighted by the absence of the first 18 residues in the protomer B but also in the slight displacement of α3 toward the inside portion of the V-shaped dimer and a concomitant movement of α2 toward α3 (Fig. 5A). To measure whether such asymmetry is also observed in other PerR proteins, we aligned each protomer forming the BsPerR (PDB code 3FN8 [16]) and SpPerR (PDB code 4I7H [17]) dimers. As shown in Fig. 5 and similar to CjPerR, the DD of each transcription factor's protomer share high structural similarity. However, the DBD of each subunit does not align equivalently well. This is reflected by a r.m.s.d. of 0.3 and 0.5 Å for BsPerR and SpPerR, respectively. In BsPerR, these differences are predominantly due to a shift of α2 and α4 N-termini toward the interior of the dimer (Fig. 5B). BsPerR α3 slightly moves toward its neighboring α-helices α2 and α4. Akin to BsPerR and CjPerR, SpPerR α2, α3, and α4 shift toward the inside portion of the V-shaped dimer and the hairpin between β1 and β2 is shifted upward in one of the protomer (Fig. 5C). Collectively, these observations suggest that while the DD is symmetric, the DBD of these transcription factors are structurally asymmetric. Such asymmetry has been previously suggested by previous NMR studies revealing that peak assignment for members of Fur proteins was difficult [26, 27] highlighting the possible existence of structural heterogeneity between each protomer. Such heterogeneity may facilitate the binding of AT rich elements which are more prone to structure flexibility compared to GC-rich sequences [28]. Moreover, the independent mobility of each DBD likely enables the transcription factor to adapt to divergent sequences forming the 5′ and 3′ ends of the inverted repeat. This divergence is illustrated in Fig. 5D showing that while the central core of the PerR boxes is conserved between C. jejuni, B. subtilis, Staphylococcus aureus, and Str. pyogenes, the type of nucleotides at position −4 and −6 are not matched with the corresponding base pairing nucleotide in position 4 and 6. Moreover, close inspection of CjPerR box elements shows that in contrast to the X-1-X inverted repeats (where X represents either seven or eight nucleotides) observed in BsPerR [10] and SaPerR [29], an additional nucleotide is observed at one end of the PerR box in C. jejuni [7] highlighting the possible role of CjPerR's extended α-helix in the readout of this extra nucleotide.
To identify the residues important for binding DNA, we first generated a representation of the CjPerR biological dimer (Fig. 6A). Protomer A was overlaid onto protomer B and the resulting dimer was used to generate the electrostatic potential surface of a CjPerR dimer. As shown in Fig. 6B, several positively charged residues that may contribute to binding to DNA line the top face of the V-shaped dimer (Fig. 6C). To test this hypothesis, we first substituted K8, K9, and K13, three residues located in the first α-helix of CjPerR, to glutamate. Increasing concentrations of each mutant was incubated with Cy5-labeled DNA corresponding to the perR promoter region [7] and separated on native PAGE. As depicted in Fig. 6D, in contrast to wild-type CjPerR, the three K → E mutants failed to bind DNA. The negative impact of the K8E and K9E mutants is consistent with previous studies showing that the residues located in the N-terminal helix are important for PaFur binding to DNA [30]. The deleterious effect of the K13E mutant on DNA binding was expected as the crystal structure of MgFur in complex DNA showed that the corresponding residue, K15, binds in the DNA minor groove to make an hydrogen bond with the base of a nucleotide. Similarly, the analogous residue in EcZur, R23, makes several polar contacts with DNA phosphate backbone. We also mutated residues Y53 and K54 (both located in α3) to determine their role in CjPerR DNA binding. Similar to the K41E mutant, substitution of Y53 and K54 into an alanine and a glutamate, respectively, incapacitate CjPerR DNA-binding activity. These findings are consistent with the DNA-bound forms of MgFur and EcZur which showed both of these residues makes several contacts with DNA. Y64 in EcZur contacts the DNA phosphate backbone and the deoxyribose moiety [31]. The EcZur R65 also makes a salt bridge interaction with a phosphate group and polar contacts with the nucleotide base. Analogously, the corresponding residue in MgFur, R57, makes several contacts with DNA [32]. Similar to our findings, mutation of these residues in MgFur and EcZur negatively impact DNA binding and gene regulatory activity. Collectively, our findings suggest that CjPerR, despite its structural differences when compared to other PerRs, likely binds DNA in a similar fashion to other metalloregulators.
Conclusion
Here, we have presented the high-resolution structure of CjPerR. The transcription factor adopts a caliper-like conformation formed by two protomers. Each protomer is composed of a DNA-binding domain and a dimerization domain. Based on the crystal structure, each protomer of CjPerR binds one zinc and one manganese ion. The peculiar orientation of the CjPerR N terminus suggests that this region of the protein is flexible and that such movement may help in finding the optimal orientation for binding DNA. This idea is further supported by DNA-binding assays showing that this region is important for interaction with a DNA element corresponding to the CjPerR box.
Acknowledgements
This work was supported by a grant from the Canadian Institutes of Health Research awarded to Drs. Jean-Francois Couture (J.-F.C) and Alain Stintzi (A.S). S.S acknowledges a scholarship from the Fonds de la Recherche en Santé du Québec.
Author contributions
J.-F.C and A.S conceptually designed the experiments and J.S.B collected the CjPerR dataset. S.S solved the crystal structure and performed the in vitro binding studies. F.C contributed to the in vitro binding assays and purification of CjPerR. J.B helped with the writing of the manuscript.
