Abstract
In addition to multiple virulence factors, Bacillus cereus a pathogen that causes food poisoning and life-threatening wound infections, secretes the pore-forming toxin hemolysin II (HlyII). The HlyII toxin has a unique 94 amino acid C-terminal domain (HlyIIC). HlyIIC exhibits splitting of NMR resonances due to cis/trans isomerization of a single proline near the C-terminus. To overcome heterogeneity, we solved the structure of P405M-HlyIIC, a mutant that exclusively stabilizes the trans state. The NMR structure of HlyIIC reveals a novel fold, consisting of two subdomains αA-β1-β2 and β3-β4-αB-β5, that come together in a barrel-like structure. The barrel core is fastened by three layers of hydrophobic residues. The barrel end opposite the HlyIIC-core has a positively charged surface, that by binding negatively charged moieties on cellular membranes, may play a role in target-cell surface recognition or stabilization of the heptameric pore complex. In the WT domain, dynamic flexibility occurs at the N-terminus and the first α-helix that connects the HlyIIC domain to the HlyII-core structure. In the destabilizing P405M mutant, increased flexibility is evident throughout the first subdomain, suggesting that the HlyIIC structure may have arisen through gene fusion.
Similar content being viewed by others
Introduction
The soil-dwelling, spore-forming B. cereus bacterium1,2,3 produces a number of virulence factors4 including several secreted pore-forming toxins (PFTs) that form lytic channels in the membranes of target cells5. One of these toxins, hemolysin II (HlyII), is present in several closely related Bacillus species including B. cereus, B. thuringiensis (a bacterium that parasitizes insects and has insecticide applications), and B. anthracis (the cause of anthrax)6, 7. In B. cereus, expression of HlyII is under the control of the HlyIIR protein8, and the Fur system that regulates iron homeostasis9, 10. Expression of HlyII is greater under oxic conditions than under conditions mimicking the intestinal tract, suggesting the toxin may not play a major role in gastrointestinal disease11. Although the physiological target of HlyII is not known, the purified toxin lyses rabbit and human erythrocytes12 as well as other cultured mammalian cells13, 14. In addition, the toxin can attack species like algae15 and insects. Studies in mice and insects suggest that HlyII is involved in virulence, and that the toxin causes apoptosis of macrophages in vitro and in vivo 16. HlyII belongs to a larger family of secreted toxins with similar predicted core structures including the B. cereus cytotoxin K (CytK)17, Staphylococcal hemolysins/leukocidins18, 19, and toxins secreted by a variety of Vibrio species20. Similar to other family members, HlyII is secreted as a water-soluble monomer that assembles into a heptameric pore following binding to cell membranes12, 21. A unique feature of HlyII is the attachment of a C-terminal domain consisting of 94 amino acids that shows no sequence or structural homology to other known proteins18. The C-terminal domain, henceforth referred to as HlyIIC, is the subject of the present study.
Removal of the HlyIIC domain reduces the activity of HlyII against rabbit erythrocytes 8-fold12, but the mechanism by which this domain affects PFT activity is unknown. Homologs of HlyII from Staphylococcal PFTs do not contain C-terminal extensions, but homologous toxins from Vibrio species have C-terminal domains with lectin-folds that bind glycan receptors on target cell surfaces22, 23. Since HlyIIC occupies a similar topological position in HlyII, it could act as a targeting domain against cell-surface receptors to aid binding to target membranes, analogous to the C-terminal extensions of Vibrio PFTs. These C-terminal lectin domains, however, have no sequence homology to HlyIIC and unrelated β-trefoil and β-prism structures24.
To investigate the HlyIIC domain, we expressed, purified, and solved the structure of the domain using solution-state nuclear magnetic resonance (NMR). The wild-type HlyIIC domain (WT-HlyIIC) exists in two states due to cis/trans isomerization of the single proline in the sequence at position 405 (here the HlyIIC domain is numbered according to the full length HlyII toxin). To reduce the spectral complexity arising from separate sets of NMR resonances for the cis and trans conformations25, we solved the structure of a P405M mutant of HlyIIC which eliminates the cis conformational state. The NMR structure of the HlyIIC domain reveals a novel, previously undescribed fold. In addition to structural studies, we characterized the backbone dynamics of the domain using NMR relaxation experiments and hydrogen exchange. To gain functional insights we modeled the domain in the context of the full-length HlyII protein. The NMR studies define the structural and dynamic properties of the HlyIIC domain and serve as a starting point for understanding its possible functions.
Results
The structure of HlyIIC reveals a novel fold
The NMR structure ensemble of the P405M-HlyIIC mutant is shown in Fig. 1A. Statistics for the NMR structure are given in Table 1. HlyIIC folds into an α + β architecture, consisting of two α-helices and five strands of anti-parallel β-sheet (Fig. 1B). The overall topology and secondary structure limits are summarized in Fig. 1C. The structure can be thought of as consisting of two subdomains. The first subdomain is comprised of the N-terminal α-helix αA and a β-hairpin: αA-β1-β2 (blue in Fig. 1A). The second subdomain has the last three β-strands with a non-sequential anti-parallel pairing between strands β3 and β5, and an α-helix intervening between strands β4 and β5: β3-β4-αB-β5 (magenta in Fig. 1A).
To investigate structure similarity relationships for HlyIIC we performed a structural homology search using the Dali server26. The search failed to identify homologs, with the closest hit having a Z-score of 1.8 that is not considered significant26. The closest hit was a portion of formyltetrahydrofolate deformylase (PDB code 3NRB), which gave an RMSD of 3.0 Å to HlyIIC. Inspection of this structure, however, showed that the two proteins do not share the same topology. We thus conclude that HlyIIC represents a previously uncharacterized, novel protein fold.
Hydrophobic core and electrostatic surface properties of the structure
Figure 2 shows some of the key side-chain interactions and surface properties of the structure. The topology of HlyIIC is unusual in that it combines α-helices and β-strands in the same layer - an arrangement thought to be disfavored since α-helices cannot hydrogen bond to β-strands27. Consistently, there are no main-chain hydrogen bonds between the two subdomains in the NMR structure, and we did not detect through H-bond couplings between the two subdomains in the lrHNCO (long-range HNCO)28 experiment. Rather, the two β-sheets are flanked on either side by helices αA and αB, and come together through hydrophobic contacts to form a barrel-like structure with pseudo-two-fold symmetry. Similar to 5-stranded anti-parallel Greek-key β-barrels seen in examples such as the OB-fold29,30,31, the hydrophobic core of the structure is arranged into three layers of non-polar residues (Figs 2A and S1A). The non-polar residues that make up the hydrophobic core are arranged with a 2-residue sequence periodicity in the β-strands and with a 3 to 4 residue periodicity in the α-helices (Fig. S1A). Representative NMR data illustrating NOESY distance contacts for core hydrophobic residues are shown in Fig. S2. Excluding the disordered first 11 amino acids, only five of 32 non-polar residues in the HlyIIC structure do not participate in the three-layer hydrophobic core structure and are surface exposed. Three of these, Y363, I365, Y367 are in the extended loop between strands β2 and β3 that connects subdomains 1 and 2 in the polypeptide chain. Residues F375 from β3 and Y406, are surface-exposed and close in the structure to the site of the cis/trans peptide bond isomerization, G404-P405.
The HlyIIC domain is decorated by a number of charged residues, including 12 lysine, 6 glutamate, and 4 aspartate residues and has a calculated pI of 8.8. An electrostatic map computed using the APBS method32 indicates a predominantly basic surface, with a positive patch located on the side of the domain opposite the N-terminal connection to the rest of the HlyII toxin. This positive patch, on the side of the molecule we will refer to as the ‘bottom’, is formed by a sequence consisting of the residues 337-KLNKGKGKL-345, together with K370 and K387 (Fig. 2B,C).
Sensitivity of the WT-HlyIIC structure to cis/trans isomerization of P405
The site of cis/trans peptide bond isomerization, G404-P405, which leads to conformational heterogeneity in NMR spectra of WT-HlyIIC is located at the top of the structure (Fig. 2D) near the N-terminal linker that connects HlyIIC to the core of the HlyII protein. We previously noted that cis/trans proline isomerization leads to the splitting of about half of the backbone crosspeaks in the 1H-15N HSQC spectrum of WT-HlyIIC25. With the structure of the HlyIIC domain, we can now map the extent to which the magnetic environments of amide protons are affected by cis/trans isomerization of the G404-P405 peptide bond (Fig. 2D). The largest composite 1H-15N chemical shift differences (red in Fig. 2D) are near P405 but also extend into the secondary structure elements flanking the proline-harboring loop, αB and β5. Intermediate perturbations (orange in Fig. 2D) occur in the loops between β1-β2 and β3-β4, which are close in space but not in the sequence to P405. Smaller composite chemical shift differences of 0.01 to 0.1 ppm (blue and green in Fig. 2D) extend throughout the entire barrel-like fold of the domain. The three-layer hydrophobic core of HlyIIC (Fig. 2A) may provide a conduit to communicate the P405 isomerization state to the rest of the structure.
CD spectroscopy of the HlyIIC domain
To determine the extent to which the P405M mutation perturbs the HlyIIC domain, we compared CD data for the WT and mutant (Fig. 3). The folded-state CD spectra for WT-HlyIIC and P405M at 20 °C are very similar and typical of folded proteins with a mixture of α-helix and β-sheet structure (Fig. 3A). For comparison, the spectra at 90 °C (dashed lines in Fig. 3A) are typical of unfolded proteins, and very different from the folded state spectra at 20 °C. The CD spectra together with the conservation of chemical shifts between WT-HlyIIC and the P405M-HlyIIC mutant25 indicate that the overall folding topology is conserved. Structural differences may occur in the loops surrounding P405, the site of the P405M mutation. The extent of these structural differences will have to await the NMR structure determination of the cis and trans forms of WT-HlyIIC. Based on the observations above, however, the P405M-HlyIIC mutant is a suitable model for the overall WT-HlyIIC structure.
Figure 3B compares thermal unfolding data for WT-HlyIIC and the P405M-HlyIIC mutant monitored by ellipticity at 220 nm. The P405M mutation, destabilizes the protein to thermal unfolding by a modest 5 °C compared to the WT. As described below, NMR hydrogen exchange data indicate a subtler effect of the mutation, that involves the uncoupling of the dynamics of subdomain 1 from 2. The P405M-HlyIIC mutant was selected because a bulky methionine should most favor a trans peptide bond33. However, P405A-HlyIIC with a smaller alanine substitution, has a similar 1H-15N HSQC spectrum indicating a conserved structure, similar circular dichroism (CD) spectra, and a similar stability to thermal unfolding within 1 degree for that of P405M-HlyIIC (Fig. S3).
Backbone dynamics indicate higher flexibility for the N-terminal subdomain
As shown in Fig. 1A and Table 1, the NMR structure of the C-terminal subdomain 2 (magenta) is more precisely determined than that of the N-terminal subdomain 1 (blue). Recall that the two subdomains are not linked through backbone hydrogen-bonds but come together solely through side-chain interactions. The differences in structural precision of the two segments could be due to differences in dynamics, or an artifact of a lower number of NMR restraints for the N-terminal half of the molecule (the distribution of long-range NOEs and H-bonds is 171 within subdomain 2, compared to 100 within subdomain 1, and 29 between the two subdomains). All aromatic residues in HlyIIC occur in the C-terminal subdomain 2. Aromatic residues in proteins typically improve spectral dispersion as well as contributing NOE distance contacts due to their hydrophobic nature, factors that may lead to the higher precision of the structure in subdomain 2. Alternatively, subdomain 1 may have intrinsically greater flexibility. To distinguish between these possibilities, we collected 15N NMR relaxation data on the backbone dynamics of P405M-HlyIIC (Fig. S4). The 15N relaxation data for P405M were analyzed with the Model-Free formalism34 to obtain S 2 order parameters that describe the amplitudes of backbone 1H-15N bond librations on the ps-ns timescale (Fig. 4A) and R2ex contributions to transverse relaxation that are sensitive to conformational exchange processes on the µs-ms timescale (Fig. 4B).
The S 2 order parameter ranges from a value of one for rigid sites to zero for sites that have unhindered motion34. As shown in Fig. 4A, most residues in the regular secondary structure HlyIIC have S 2 order parameters in the rigid limit. Low S 2 order parameters indicative of dynamic flexibility occur for the N-terminal segment that connects HlyIIC to the rest of the HlyII protein, the N-terminal α-helix (αA), and the loops between secondary structure units - in particular the loop between helix αB and strand β5 that includes the site of the P405M mutation. The dynamic nature of the N-terminal linker segment corresponding to the first 11 amino acids of HlyIIC appears to be conserved in the full-length HlyII toxin, in as much as HlyIIC can be proteolytically cleaved from HlyII12, 21. The mean S 2 order parameters for residues 330–360 in subdomain 1 and 370–412 in subdomain 2 are near 0.9, and statistically indistinguishable. Therefore, it does not appear that the lower precision of subdomain 1 in the NMR structure is due to increased dynamics of this segment on the ps-ns timescale. The exception is helix αA, which has residues with moderately lower S 2 order parameters in an element of regular secondary structure that may contribute to the poorer definition of subdomain 1 in the NMR structure.
Figure 4B summarizes the sequence distribution of residues with R2ex line-broadening contributions to the transverse relaxation rate, R2. The non-zero R2ex terms are due to conformational exchange on the µs-ms timescale. The majority of residues that experience R2ex contributions are in the two helices αA and αB, with some additional ones in the loops connecting elements of regular secondary structure. By contrast, R2ex terms are not seen for the β-strands (Fig. 4B). These observations suggest that the α-helices and loops are involved in some type of conformational exchange process on the µs-ps timescale, that does not affect the β-sheets. A greater fraction of residues in subdomain 1 (36%) than subdomain 2 (19%) show conformational-exchange line broadening contributions, and the R2ex terms are somewhat larger for subdomain 1 in the first half of the protein (Fig. 4B). The lower precision of subdomain 1 in the NMR structure is thus likely to be related to its greater flexibility on the µs-ms timescale rather than on the faster ps-ns timescale.
Hydrogen exchange points to enhanced dynamics for the N-terminal subdomain in the P405M mutant
Figure 5 compares hydrogen exchange (HX) data for WT-HlyIIC and P405M-HlyIIC. In Fig. 5A, a control 1H-15N HSQC of WT-HlyIIC in H2O (black) is superimposed with those obtained after 2 h of exchange in D2O for WT-HlyIIC (cyan) and P405M-HlyIIC (red). About 40 of the 92 backbone amide protons in WT-HlyIIC persist after 2 h in D2O (cyan peaks in Fig. 5A). These protected amide protons are distributed throughout the regular secondary structure of the HlyIIC domain. By contrast in the P405M-HlyIIC mutant, only amide protons from residues in subdomain 2 survive after 2 h in D2O. The result suggests that HX from subdomain 1 is selectively enhanced in the P405M-HlyIIC mutant.
Protection factors that describe the fold-decrease in HX rates due to structure compared to the chemically-determined intrinsic HX rates35, are given for WT-HlyIIC and P405M-HlyIIC in Fig. 5B and C, respectively. There are ~10 protected backbone amide protons in WT-HlyIIC that give resolved peaks for the conformers related by cis/trans isomerization of P405 (the cis form resonances are indicated with green labels in Fig. 5A). The cis resonances are not seen in the P405M mutant which only has the trans form. Because cis/trans isomerization occurs on a much faster timescale than HX, the HX rates for the cis and trans forms are very similar and only the protection factors for the trans form of WT-HlyIIC are given in Fig. 5B. Protection factors are mapped on the NMR structure for WT-HlyIIC (Fig. 5D) and P405M-HlyIIC (Fig. 5E).
Backbone amide proton protection factors in WT-HlyIIC are uniformly large for residues in regular secondary structure. The exception is helix αA, where only a single residue shows weak protection (Fig. 5B,D). In the P405M mutant (Fig. 5C), the largest protection factors are reduced about 2.5-fold compared to WT, consistent with the lower stability of the mutant to unfolding (Fig. 3B). A much more dramatic change is seen when the structure distribution of protection factors in the P405M mutant is considered. The amide protons from subdomain 2 in the second half of the molecule are protected but only two amide protons from subdomain 1 show measurable protection, at the beginning of strand β2 (Fig. 5C,E). The two weakly protected amide protons from subdomain 1 exchange within 2 h of dissolving the protein in D2O (Fig. 5A). Moreover, within subdomain 2, the protection factors for helix αB are about 10-fold weaker than those from the β3-β4-β5 component of the structure (Fig. 5C). Taken together, these observations suggest that the P405M mutant, which is necessary to eliminate the cis proline conformation, not only globally destabilizes the HlyIIC domain by 5 °C compared to the WT, but also causes an uncoupling of the HX dynamics of subdomain 1 from subdomain 2. To a lesser extent, within subdomain 2 the dynamics of helix αB are also uncoupled from those of the β5-β3-β4 sheet. In addition to the backbone amide protons, the side-chain indole nitrogen proton of W372 from subdomain 2 is protected in both WT and the P405M mutant (not shown). In the NMR structure, the indole nitrogen proton of W372 forms an H-bond with the side-chain carboxylic acid group of E386 from subdomain 2, which in turn forms a salt-bridge to K342 from subdomain 1 (Fig. S1B).
The uncoupling of the P405M mutant structure into two subdomains may be related to the observation that this protein forms a folded dimer with well-dispersed NMR signals at protein concentrations above ~0.5 mM and temperatures above ~20 °C25. The conditions for the present study (0.4 mM protein concentration and a temperature of 15 °C) were chosen so as to avoid the dimeric form. In contrast to the P405M-HlyIIC mutant, the more cooperatively folded WT-HlyIIC shows no signs of oligomerization at protein concentrations as high as 2 mM and temperatures as high as 40 °C. Additional support for a weaker structural cooperativity in the P405M mutant comes from the thermal denaturation data in Fig. 3B, where the slope of the thermal unfolding curve for P405M-HlyIIC is shallower than that for the WT, a hallmark of a less cooperative folding transition.
Modeling of the HlyIIC domain in the context of full-length HlyII
To better understand the role of the HlyIIC domain in the context of the full-length HlyII toxin we calculated homology models of the latter (Fig. 6). The HlyII-core was modeled using the monomer (PDB 4IDJ) and protomer (7AHL) X-ray structures of the homologous toxin α-hemolysin from S. aureus with the SWISS-MODEL server36. The non-conserved HlyIIC domain, which is unique to HlyII, was linked and docked to the respective core structures with the Ab Initio Domain Assembly (AIDA) server37. The AIDA program is used to predict the orientations of domains in multi-domain proteins37. The structure of each individual domain is treated as invariant, and only the structure of the linker between the domains is varied to energy-minimize an ab initio folding potential that seeks to optimize docking of the domains37. A model of the membrane-spanning pore structure was constructed by superposing the HlyII-core structure of the protomer calculated with the AIDA program onto each of the seven protomers in the 7AHL heptamer structure (Fig. 6B,C). An initial model without distance restraints resulted in steric overlap between the HlyIIC domain and the adjacent protomer that was not included in the docking simulation (Fig. S5). Noticing there were exposed aromatic residues in the HlyII-core structure (Y54, F302, F303) positioned to interact with exposed aromatic residues in the HlyIIC domain (F375,Y406), we set six distance restraints of 10 Å between the aromatic residues to obtain a revised final model (Fig. 6B–D). The orientations of the HlyII-core and HlyIIC domain were similar to the original model (Fig. S5), but in the revised model there were no longer steric clashes between the protomers. Of the aromatic residues, Y54 in the HlyII core, and F375 and Y406 in the HlyIIC domain appear to be conserved. The two residues F302 and F303 are conserved in sequence homologs that have the HlyIIC domain, but not in those where it is absent, suggesting a possible role of these aromatic residues in docking HlyIIC against the HlyII core.
Discussion
The present work demonstrates that HlyII contains a structurally unique C-terminal domain, not present on PFTs that otherwise share structural homology and similar function. Indeed, the structure of HlyIIC, which consists of two β-sheets flanked by two α-helices in a barrel-like structure, represents a novel fold with no identifiable structural homologs. Because the HlyIIC structure represents a novel fold we anticipate it will shed light on structure-function relationships of yet to be discovered sequence and structure homologs.
A BLAST search of the HlyIIC sequence revealed that besides other proteins in the HlyII family, the closest sequence homologs are two gene fragments of unknown function found in B. anthracis. The first fragment corresponding to residues 319–371 in the N-terminal subdomain 1 of HlyIIC, occurs on the chromosome of the Sterne strain of B. anthracis (sequence ID: AAT54665.1, 87% sequence identity). The second fragment corresponding to residues 364–412 in the C-terminal subdomain 2 of HlyIIC, is found in the XO1 plasmid (sequence ID: WP_010890031.1, 38% sequence identity)12, 38. The XO1 plasmid, together with XO2, is needed for anthrax toxicity.
Because the HlyIIC domain exists in a dynamic equilibrium due to cis/trans isomerization of P405, we solved the NMR structure of the proline to methionine single point mutant P405M-HlyIIC. CD data (Fig. 3A), as well as the similarity of NMR chemical shifts between the WT and the mutant25, indicate that the structure of WT-HlyIIC is very similar to the P405M-HlyIIC mutant. A notable difference is that strand β5 is extended by three to four residues in the cis compared to the trans form of WT-HlyIIC, based on a dihedral angle analysis of backbone NMR chemical shifts25.
The P405M mutation destabilizes the HlyIIC domain by 5 °C to thermal denaturation (Fig. 3B) and leads to an uncoupling of the dynamics of subdomain 1 from subdomain 2 (Fig. 5B–E). That the two subdomains have properties of independent folding units and the observation that the subdomains have homology to separate gene fragments in B. anthracis, suggests that the HlyIIC domain may have arisen through a gene fusion event. If the HlyIIC domain originated from the fusion of subdomain 1 (αA-β1-β2) with subdomain 2 (β3-β4-αB-β5), evolutionary pressure would have favored the accumulation of mutations that maximized the folding cooperativity of the structure. Because it is destabilizing, the P405M mutation could relieve the requirement for maximum folding cooperativity, and reveal the original bipartite organization of the structure into two subdomains with partially autonomous folding properties. A similar mechanism has been proposed for the organization of subdomains into cooperatively folded structures in the OB-fold family of proteins29. Conversely, the two subdomains encoded on separate genes in B. anthracis, could come together through fragment complexation to form a protein complex with a similar structure to the HlyIIC domain.
The function of the HlyIIC domain is currently unknown. An earlier structural model for HlyIIC was proposed21 based on a sequence to structure ab initio prediction using the program ROSETTA, although the convergence of the predicted structures was poor21. Based on this model it was suggested that the HlyIIC domain had a structure similar to the iron-binding protein frataxin, and a possible iron-binding function39. While the two proteins have similar secondary structure elements, the β-sheet topology of HlyIIC is different from frataxin, where the β-strand connections are all sequential and the last α-helix follows the last β-strand. Frataxin forms a flat continuous β-sheet with the two α-helices on one side of the sheet, while HlyIIC forms a barrel-like structure. The coordinate RMSD between the NMR structure of HlyIIC and frataxin (PDB code 1EKG) is ~14 Å, indicating quite different structures.
Because many PFTs contain accessory domains that are carbohydrate-binding lectins40, we submitted the HlyIIC domain to a glycan array screen run by the Consortium for Functional Glycomics (CFG) (www.functionalglycomics.org). Screening a library of over 600 mammalian glycans failed to identify any hits for HlyIIC (CFG accession code: primscreen_5868). Although this negative result suggests that HlyIIC is not a carbohydrate-binding protein, it does not rule out this possibility. First, the HlyIIC domain on its own, may have a weaker affinity for carbohydrates than when part of the PFT formed from full-length HlyII. Second, the B. cereus HlyII toxin and HlyIIC domain are also found in B. thuringiensis, a bacterium that infects insects. If HlyIIC binds carbohydrates specific to insects, these may have been missed in the mammalian glycan library.
To further explore how the HlyIIC domain interacts with the rest of the toxin, we modeled its NMR structure in the context of full-length HlyII (Fig. 6). Without functional data on the full-length HlyII toxin, which is difficult to express, such modeling is speculative but could provide clues about possible roles for HlyIIC based on its position in the overall structure. We wondered if the HlyIIC domain could displace the stem loop (gray in Fig. 6A) in the monomer structure, so as to favor the membrane-bound protomer conformation of HlyII, in which the stem loop participates in the heptameric β-barrel pore. In the monomer model (Fig. 6A), the HlyIIC domain extends from the toxin core at an edge of the β-sandwich structure opposite that of the stem loop, so that an interaction between the two segments is highly unlikely. Fig. 6B and C show the homology model of the heptameric membrane-spanning pore complex. Even with a fully extended linker, the domain appears to be too far away to interact with the pore component of the structure (Fig. 6B and C), in agreement with experimental results that indicate HlyIIC does not obstruct the toxin pore in voltage gating experiments12. Rather, the HlyIIC domain in the heptamer model is positioned towards the membrane with the positively charged patch formed by the lysine residues at the bottom of the structure poised to interact with negatively charged molecules such as lipids carbohydrates, or proteins embedded in the membrane (Fig. 6C). Biochemical studies showed that HlyII with HlyIIC deleted, can still form oligomeric pores that have only an 8-fold reduction in activity compared to the full-length wild type12. Oligomers of the toxin without the HlyIIC domain were less stable in SDS than those with the domain, the former dissociating at 78 °C compared to 82 °C for the latter12. The HlyII toxin is relatively non-specific being able to lyse a variety of eukaryotic cell types13, 14, as well as being able to form active pores in model membrane systems such as liposomes without the C-terminal HlyIIC domain21. Thus, putative membrane interactions of the positively-charged HlyIIC surface with negatively-charged lipid, carbohydrate, or protein moieties on cell membranes could afford a general mechanism for cell targeting or stabilization of the PFT complex in the membrane.
An interesting question regards the role played by cis/trans isomerization of P405 in the HlyII toxin. Splitting of backbone 1H-15N signals from sites distributed throughout the structure is evident in the 1H-15N HSQC spectrum of the HlyIIC domain (Fig. 2D), so that binding events at distal sites could be communicated to P405. For example, the proposed binding of the positively charged patch of HlyIIC to negatively charged moieties in the membrane could affect the ratio of cis to trans isomers for P405, located in a loop between strands β4 and β5 that is directly adjacent to the N-terminal linker that connects the HlyIIC to the HlyII-core (Fig. 6D). The linker segment between the N-terminus of HlyIIC domain and the C-terminus of the HlyII core is highly flexible and can be proteolytically cleaved12, 21, making it unlikely to affect communication between the two domains. However, P405 immediately precedes the conserved exposed aromatic residue Y406 in HlyIIC, that may form contacts with aromatic residues from the HlyII-core (Fig. 6D). A considerably different alternative role for the proline, is that it could constitute a binding site for cyclophilin. The site of proline isomerization in HlyIIC, P405, is part of the sequence 404-GPYI-407, which is close to the consensus sequence GPxL recognized by cyclophilins41, 42. Moreover, cyclophilin-binding should favor prolines undergoing cis/trans isomerization43, such as P405 in HlyIIC. It has recently been reported that leukotoxin binds to cyclophilin D in neutrophil mitochondria thereby triggering the disruption of the mitochondrial membrane potential, release of cytochrome c, and apoptosis44, 45. The induction of apoptosis in neutrophils may help bacteria evade immune defenses. It remains to be seen whether HlyII can interact with mitochondrial cyclophilin D, and serve a similar role.
In conclusion, the structure of the HlyIIC domain represents a new fold in which two subdomains each comprised of an α-helix and a β-sheet (αA-β1-β2 and β3-β4-αB-β5) come together to form a barrel-like structure, stabilized by a core formed from three layers of hydrophobic residues. The different dynamic properties of the two subdomains when the HlyIIC domain is destabilized by the P405M mutation, together with the existence of two homologous gene fragments in B. anthracis, suggests that the HlyIIC domain may have arisen from a genetic fusion of the two subdomains. The most likely function for the HlyIIC domain is to stabilize the heptameric PFT, either through interactions with adjacent protomers in the heptamer or through membrane binding. The unique proline P405 in the HlyIIC domain, which exists in two conformational states due to cis/trans isomerization, could act as a conduit to communicate binding events to the core of the HlyII toxin or could constitute a binding site for cyclophilin D. The structure and dynamics of HlyIIC characterized in this work pave the way for experiments that can better establish the function of this novel accessory domain in PFTs.
Methods
Expression and purification of HlyIIC
The HlyIIC domain P405M mutant (P405M-HlyIIC) was expressed and purified as described previously25 and isotopically enriched with 15N and 13C using established protocols46. Triple-labeled samples (15N13,C, and 2H) were produced by slowly acclimatizing E. coli to grow in deuterium oxide (D2O) by incremental amounts47. Briefly, T7 Express cells (New England Biolabs Inc., Ipswich, MA) containing the HlyIIC gene fragment (residues D319-I412) cloned into the pET28b plasmid, were grown overnight at 37 °C in M9 minimal media supplemented with 30 μg/ml kanamycin using glucose (3 g/L) as a carbon source. A 10 ml culture was inoculated with 500 μL of the overnight culture, and grown to mid-log phase (OD600 = 0.6–0.8) with vigorous aeration. This culture was used to inoculate new M9 media containing 30% D2O and grown to the same OD as the first. In a similar fashion, each culture was used to inoculate fresh media with successively increasing amounts of D2O (0%, 30%, 60%, 90%, 100%). The final culture (500 ml) contained 1.5 g/l 15N-labeled NH4Cl and 3 g/l 2H, 13C-enriched glucose in D2O. Protein was expressed for 18 hours at 37 °C and purified by Ni-NTA affinity and size-exclusion chromatography as described previously25. The His6-affinity tag was removed by cleavage with a 1:100 wt/wt ratio of thrombin for 1 h at room temperature (Haemotologic Technologies Inc., Essex Junction, VT).
NMR structure determination
P405M-HlyIIC samples for NMR were at a concentration of 0.4 mM in 20 mM NaH2PO4 buffer, pH 6.1. All NMR work on the mutant was done at a temperature of 15 °C to suppress signals from a dimeric form unique to the P405M-HlyIIC mutant, which is stabilized at large protein concentrations and high temperature25. NMR spectra were recorded using an 800 MHz Varian Inova Spectrometer equipped with a cryogenic probe. Chemical shift assignments for P405M-HlyIIC and the cis and trans forms of WT-HlyIIC were published previously25. NOE-based distance restraints were obtained from 3D 15N- and 13C-NOESY experiments with mixing times of 125 ms. These were supplemented with hydrogen bond restraints from a long-range HNCO (lrHNCO) experiment that identifies hydrogen bond acceptors and donors via through-H-bond couplings28. The lrHNCO experiment was recorded in TROSY mode on a 2H, 13C, 15N triple-labeled sample of HlyIIC. Backbone (ϕ, ψ) and sidechain (χ1) torsion angles were calculated from HN, Hα, Cα, Cβ, CO, and N chemical shifts using the program TALOS-N48. Stereospecific assignments for methylene protons were determined from a 3D HNHB spectrum, 50 ms mixing time 2D NOESY spectra, and 1H-15N NOESY-HSQC experiments49. NMR structures of P405M-HlyIIC were calculated from 1736 experimental restraints (Table 1) using the programs X-PLOR NIH 2.3650 and ARIA 2.351. In the final optimization step, structures from X-PLOR NIH were refined in water with ARIA. The 25 lowest energy NMR structures have been deposited in the Protein Data Bank under the PDB accession code 2N67.
Backbone dynamics from NMR relaxation experiments
Backbone dynamics of P405M-HlyIIC were investigated using 15N longitudinal (R1), transverse (R2), and cross-relaxation (1H-15N NOE) experiments. All relaxation data were acquired at a field strength of 800 MHz. R1 rates were obtained using interleaved relaxation delays of 0.05, 0.13, 0.21, 0.49, 0.57, 0.71 and 0.99 s. R2 rates were determined using interleaved relaxation delays of 0.01, 0.03, 0.05, 0.07, 0.09, 0.11, 0.15, 0.25 s. A 2 s pre-acquisition delay was used for recovery to thermal equilibrium. 1H-15N NOE values were determined from the ratio of crosspeak intensities in a spectrum for which the proton signals were saturated for 4 s and a control spectrum in which the saturation period was replaced by a pre-acquisition delay of equivalent length. The processing and analysis of relaxation parameters was done according to published methods52, 53. Model-free analyses34 of the 15N relaxation data to obtain S 2 order parameters and R2ex conformational-exchange terms were performed with the program Tensor254, and yielded an optimal global rotational correlation time of 9.7 ns.
Circular dichroism (CD)
HlyIIC-WT, HlyIIC-P405M, and HlyIIC-P405A were dialyzed into buffer containing 10 mM NaH2PO4, pH 7.5 and 50 mM NaCl, and diluted to a concentration of 4 μM. CD data were collected on a Jasco J-810 spectrometer (Jasco Inc., Easton, MD). Thermal denaturation was monitored using the change in molar ellipticity at 220 nm while changing the temperature in 5 degree increments from 20 °C to 90 °C. Tm values were calculated from the mid-point of sigmoidal fits to the temperature data.
Hydrogen exchange experiments
Hydrogen exchange (HX) experiments on 0.4 mM samples of WT-HlyIIC and the P405M-HlyIIC mutant were performed on a 600 MHz spectrometer at pH 6.2 and a temperature of 10 °C. To initiate HX, lyophilized 15N-labeled samples were dissolved in 99.96% D2O and 1H-15N HSQC spectra were recorded as a function of time. Decays of 1H-15N crosspeak intensities with D2O incubation times were fitted to single exponentials, to obtain HX rate constants and their associated uncertainties55. Protection factors were calculated as the intrinsic HX rates divided by experimental HX rates. Intrinsic HX rates for the HlyIIC sequence were calculated using the SPHERE server (http://landing.foxchase.org/research/labs/roder/sphere/)56.
Modeling of full-length HlyII
The core structure of HlyII was modeled using the homologous S. aureus α-hemolysin toxin with the Swiss-Model server (https://swissmodel.expasy.org)36. The NMR structure of the non-conserved HlyIIC domain was linked and docked to the homology model of the HlyII core structure with the Ab Initio Domain Assembly (AIDA) server (http://ffas.sanfordburnham.org/AIDA/), a program used to predict the relative orientations of domains in multi-domain proteins37. A model of the heptameric membrane-spanning pore, was constructed by superposing the HlyII-core structure calculated with AIDA onto each of the seven protomers in the α-hemolysin structure.
Data availability
The atomic coordinates and restraint files for HlyIIC-P405M are deposited in the PDB with accession code 2N67. Chemical shifts are deposited under BMRB entries 19463 (HlyIIC-P405M), 19462 (trans WT-HlyIIC), and 19461 (cis WT-HlyIIC). Additional datasets if not included in this published article (and its Supplementary Information Files) are available from the corresponding author on reasonable request.
References
Bottone, E. J. Bacillus cereus, a volatile human pathogen. Clin. Microbiol. Rev. 23, 382–398 (2010).
Drobniewski, F. A. Bacillus cereus and related species. Clin. Microbiol. Rev. 6, 324–338 (1993).
Logan, N. A. Bacillus and relatives in foodborne illness. J. Appl. Microbiol. 112, 417–429 (2012).
Stenfors Arnesen, L. P., Fagerlund, A. & Granum, P. E. From soil to gut: Bacillus cereus and its food poisoning toxins. FEMS Microbiol. Rev. 32, 579–606 (2008).
Ramarao, N. & Sanchis, V. The pore-forming haemolysins of bacillus cereus: a review. Toxins (Basel) 5, 1119–1139 (2013).
Budarina, Z. I. et al. Hemolysin II is more characteristic of Bacillus thuringiensis than Bacillus cereus. Arch. Microbiol. 161, 252–257 (1994).
Sinev, M. A., Budarina, Z. I., Gavrilenko, I. V., Tomashevskiĭ, A. I. & Kuzmin, N. P. Evidence of the existence of hemolysin II from Bacillus cereus: cloning the genetic determinant of hemolysin II. Mol. Biol. (Mosk.) 27, 1218–1229 (1993).
Budarina, Z. I. et al. A new Bacillus cereus DNA-binding protein, HlyIIR, negatively regulates expression of B. cereus haemolysin II. Microbiology (Reading, Engl.) 150, 3691–3701 (2004).
Harvie, D. R., Vilchez, S., Steggles, J. R. & Ellar, D. J. Bacillus cereus Fur regulates iron metabolism and is required for full virulence. Microbiology (Reading, Engl.) 151, 569–577 (2005).
Sineva, E. et al. Iron regulates expression of Bacillus cereus hemolysin II via global regulator Fur. J Bacteriol 194, 3327–3335 (2012).
Clair, G., Roussi, S., Armengaud, J. & Duport, C. Expanding the known repertoire of virulence factors produced by Bacillus cereus through early secretome profiling in three redox conditions. Mol. Cell Proteomics 9, 1486–1498 (2010).
Miles, G., Bayley, H. & Cheley, S. Properties of Bacillus cereus hemolysin II: a heptameric transmembrane pore. Protein Sci 11, 1813–1824 (2002).
Andreeva, Z. I. et al. Purification and cytotoxic properties of Bacillus cereus hemolysin II. Protein Expression and Purification 47, 186–193 (2006).
Kholodkov, O. A., Budarina, Z., Kovalevskaya, J. I., Siunov, A. V. & Solonin, A. Effect of Bacillus cereus hemolysin II on hepatocyte cells. Prikl. Biokhim. Mikrobiol. 51, 258–267 (2015).
Kataev, A. A., Andreeva-Kovalevskaya, Z. I., Solonin, A. S. & Ternovsky, V. I. Bacillus cereus can attack the cell membranes of the alga Chara corallina by means of HlyII. Biochim. Biophys. Acta 1818, 1235–1241 (2012).
Tran, S.-L. et al. Haemolysin II is a Bacillus cereus virulence factor that induces apoptosis of macrophages. Cell. Microbiol. 13, 92–108 (2011).
Lund, T., De Buyser, M. L. & Granum, P. E. A new cytotoxin from Bacillus cereus that may cause necrotic enteritis. Mol Microbiol 38, 254–261 (2000).
Baida, G., Budarina, Z. I., Kuzmin, N. P. & Solonin, A. S. Complete nucleotide sequence and molecular characterization of hemolysin II gene from Bacillus cereus. FEMS Microbiol. Lett. 180, 7–14 (1999).
Menestrina, G., Serra, M. D. & Prévost, G. Mode of action of beta-barrel pore-forming toxins of the staphylococcal alpha-hemolysin family. Toxicon 39, 1661–1672 (2001).
Olson, R. & Gouaux, E. Vibrio cholerae cytolysin is composed of an alpha-hemolysin-like core. Protein Sci. 12, 379–383 (2003).
Andreeva, Z. I. et al. The properties of Bacillus cereus hemolysin II pores depend on environmental conditions. Biochimica et biophysica acta 1768, 253–263 (2007).
Kaus, K., Lary, J. W., Cole, J. L. & Olson, R. Glycan specificity of the Vibrio vulnificus hemolysin lectin outlines evolutionary history of membrane targeting by a toxin family. J. Mol. Biol. 426, 2800–2812 (2014).
Levan, S., De, S. & Olson, R. Vibrio cholerae cytolysin recognizes the heptasaccharide core of complex N-glycans with nanomolar affinity. J. Mol. Biol. 425, 944–957 (2013).
Olson, R. & Gouaux, E. Crystal structure of the Vibrio cholarae cytolysin (VCC) pro-toxin and its assembly into a heptameric transmembrane pore. Journal of molecular biology 350, 997–1016, doi:10.1016/j.jmb.2005.05.045 (2005).
Kaplan, A. R., Maciejewski, M. W., Olson, R. & Alexandrescu, A. T. NMR assignments for the cis and trans forms of the hemolysin II C-terminal domain. Biomolecular NMR assignments 8, 419–423 (2014).
Holm, L. & Rosenström, P. Dali server: conservation mapping in 3D. Nucleic Acids Research 38, W545–549 (2010).
Finkelstein, A. V. & Ptitsyn, O. B. Protein Physics: A Course of Lectures. (Academic Press 2002).
Cordier, F. & Grzesiek, S. Direct observation of hydrogen bonds in proteins by interresidue 3HJNC’ scalar couplings. Journal of the American Chemical Society 121, 1601–1602 (1999).
Alexandrescu, A. T., Gittis, A. G., Abeygunawardana, C. & Shortle, D. NMR structure of a stable ‘OB-fold’ sub-domain isolated from staphylococcal nuclease. J. Mol. Biol. 250, 134–143 (1995).
Murzin, A. G. OB(oligonucleotide/oligosaccharide binding)-fold: common structural and functional solution for non-homologous sequences. EMBO J 12, 861–867 (1993).
Murzin, A. G., Lesk, A. M. & Chothia, C. Principles determining the structure of beta-sheet barrels in proteins. I. A theoretical analysis. Journal of molecular biology 236, 1369–1381 (1994).
Baker, N. A., Sept, D., Joseph, S., Holst, M. J. & McCammon, J. A. Electrostatics of nanosystems: application to microtubules and the ribosome. Proc Natl Acad Sci USA 98, 10037–10041 (2001).
Johansson, C., Ullner, M. & Drakenberg, T. The solution structures of mutant calbindin D9k’s, as determined by nMR, show that the calcium-binding site can adopt different folds. Biochemistry 32, 8429–8438 (1993).
Lipari, G. & Szabo, A. Model-Free Approach to the Interpretation of Nuclear Magnetic Resonance Relaxation in Macromolecules. 1. Theory and Validity. J. Am. Chem. Soc. 104, 4546–4559 (1982).
Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W. Protein stability parameters measured by hydrogen exchange. Proteins 20, 4–14 (1994).
Arnold, K., Bordoli, L., Kopp, J. & Schwede, T. The SWISS-MODEL Workspace: A web-based environment for protein structure homology modelling. Bioinformatics 22, 195–201 (2006).
Xu, D., Jaroszewski, L., Li, Z. & Godzik, A. AIDA: ab initio domain assembly server. Nucleic Acids Res 42, W308–W313 (2014).
Perlinska, A. & Grynberg, M. Bacillus anthracis pXO1 plasmid encodes a putative membrane-bound bacteriocin. PeerJ 2, e679, doi:10.7717/peerj.679 (2014).
Bakulina, A. Y., Sineva, E. V., Solonin, A. S. & Maksyutov, A. Z. In Proceedings of the Fifth International Conference on Bioinformatics of Genome Regulation, Structure, July 16–22, 231–234 (2006).
Dai Peraro, M. & van der Goot, F. G. Pore-forming toxins: ancient, but never really out of fashion. Nat Rev Microbiol. 14, 77–92 (2016).
Coaker, G., Zhu, G., Ding, G., Van Doren, S. R. & Staskawicz, B. Eukaryotic cyclophilin as a molecular switch for effector activation. Molecular Microbiology 61, 1485–1496 (2006).
Piotukh, K. et al. Cyclophilin A Binds to Linear Peptide Motifs Containing a Consensus That Is Present in Many Human Proteins. Journal of Biological Chemistry 280, 23668–23674 (2005).
Sarkar, P., Reichman, C., Saleh, T., Birge, R. B. & Kalodimos, C. G. Proline cis-trans Isomerization Controls Autoinhibition of a Signaling Protein. Molecular Cell 25, 413–426 (2007).
Atapattu, D. N., Albrecht, R. M., McClenahan, D. J. & Czuprynski, C. J. Dynamin-2-dependent targeting of Mannheimia haemolytica leukotoxin to mitochondrial Cyclophilin D in bovine lymphoblastoid cells. Infection and Immunity 75, 5357–5365 (2008).
Aulik, N. A., Hellenbrand, K. M., Kisela, D. & Czuprynski, C. J. Mannheimia haemolytica leukotoxin binds cyclophilin D on bovine neutrophil mitochondria. Microbial Pathogenesis 50, 168–178 (2011).
Marley, J., Lu, M. & Bracken, C. A method for efficient isotopic labeling of recombinant proteins. J. Biomol. NMR 20, 71–75 (2001).
Sosa-Peinado, A., Mustafi, D. & Makinen, M. W. Overexpression and biosynthetic deuterium enrichment of TEM-1 beta-lactamase for structural characterization by magnetic resonance methods. Protein Expression and Purification 19, 235–245 (2000).
Shen, Y. & Bax, A. Prediction of Xaa-Pro peptide bond conformation from sequence and chemical shifts. J Biomol NMR 46, 199–204, doi:10.1007/s10858-009-9395-y (2009).
Case, D. A., Dyson, H. J. & Wright, P. E. Use of chemical shifts and coupling constants in nuclear magnetic resonance structural studies on peptides and proteins. Methods Enzymol 239, 392–416 (1994).
Schwieters, C. D., Kuszewski, J. J., Tjandra, N. & Clore, G. M. The Xplor-NIH NMR molecular structure determination package. Journal of magnetic resonance 160, 65–73 (2003).
Bardiaux, B., Malliavin, T. & Nilges, M. ARIA for solution and solid-state NMR. Methods in molecular biology 831, 453–483, doi:10.1007/978-1-61779-480-3_23 (2012).
Alexandrescu, A. T. & Shortle, D. Backbone dynamics of a highly disordered 131 residue fragment of staphylococcal nuclease. Journal of molecular biology 242, 527–546 (1994).
Mandel, A. M., Akke, M. & Palmer, A. G. 3rd Backbone dynamics of Escherichia coli ribonuclease HI: correlations with structure and function in an active enzyme. Journal of molecular biology 246, 144–163 (1995).
Dosset, P., Hus, J. C., Blackledge, M. & Marion, D. Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J Biomol NMR 16, 23–28 (2000).
Newcomer, R. L., Fraser, L. C., Teschke, C. M. & Alexandrescu, A. T. Mechanism of Protein Denaturation: Partial Unfolding of the P22 Coat Protein I-Domain by Urea Binding. Biophys J 109, 2666–2677, doi:10.1016/j.bpj.2015.11.010 (2015).
Zhang, Y.-Z. Protein and peptide structure and interactions studied by hydrogen exchange and NMR., (Ph.D. Thesis, University of Pennsylvania 1995).
Waelder, S. F. & Redfield, A. G. Nuclear Magnetic Resonance studies of exchangeable protons. II. The solvent exchange rate of the indole nitrogen proton of tryptophan derivatives. Biopolymers 16, 623–629 (1977).
Nabuurs, S. B. et al. Quantitative evaluation of experimental NMR restraints. Journal of the American Chemical Society 125, 12026–12034, doi:10.1021/ja035440f (2003).
Bhattacharya, A., Tejero, R. & Montelione, G. T. Evaluating protein structures determined by structural genomics consortia. Proteins 66, 778–795, doi:10.1002/prot.21165 (2007).
Acknowledgements
We thank Dr. Mark W. Maciejewski for assistance with NMR data collection and Prof. Irina Russu for useful discussion. This work was supported by startup funds and project grants from Wesleyan University to R.O., and by a University of Connecticut Research Foundation (UCRF) grant to A.T.A. K.K. is supported by an NIH training grant in molecular biophysics (Grant T32 GM008271) and A.R.K. by NSF-GRFP.
Author information
Authors and Affiliations
Contributions
R.O. and A.T.A. conceived and designed the project. K.K. and S.D. prepared protein samples and carried out the CD experiments. A.R.K. and A.T.A. performed the NMR experiments and determined the NMR structure of HlyIIC. A.T.A did the homology modeling of full-length HlyII. A.T.A. and R.O. wrote the manuscript with contributions from A.R.K. and K.K. All authors discussed the results and commented on the manuscript.
Corresponding authors
Ethics declarations
Competing Interests
The authors declare that they have no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kaplan, A.R., Kaus, K., De, S. et al. NMR structure of the Bacillus cereus hemolysin II C-terminal domain reveals a novel fold. Sci Rep 7, 3277 (2017). https://doi.org/10.1038/s41598-017-02917-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-017-02917-4
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.