Introduction

Staphylococcus aureus (SA) is a robust Gram-positive pathogen associated with life-threatening systemic infections and a leading cause of hospital-associated and community-associated infections (1-4). Pore forming toxins that are secreted by SA are crucial to its immune evasion. These toxins may create a survival advantage for the bacteria by forming pores into the membrane of target cells, inducing cell death and weakening the host during the first stages of infection. Because of the limited treatment modalities for SA infection, the emergence of methicillin-resistant SA (MRSA) poses a tremendous public health threat. While the molecular basis of the disease remains unclear, community-associated MRSA infection is closely linked to the presence of a Panton-Valentine leukocidin (PVL), a bipartite toxin consisting of the ~34 kDa LukF-PV and the ~32 kDa LukS-PV proteins (5). Since PVL plays an important role in MRSA infection, an understanding of its structure and function is crucial to the development of novel therapeutics.

PVL’s mechanism of action has been intensely studied. As shown in Scheme 1, the function of the two PVL components is synergistic and requires a sequence of events at the membrane surface of the target cell (6). In the first step, the secreted, water-soluble LukF-PV and LukS-PV monomers aggregate on the membrane surface, and subsequently assemble into heterodimers. In a stepwise fashion, these heterodimers further oligomerize into heterotetramers that are characterized by alternating LukF-PV and LukS-PV subunits. These heterotetramers further assemble into an octameric, disc-like structure that is comprised of alternating LukS-PV and LukF-PV subunits in a 1:1 stoichiometry (7). At this stage, experimental data indicates that PVL exists as an octamer in pre-pore conformation that is not fully functional and not transversing the cell membrane. Subsequently, the pre-pore structure undergoes major conformational changes that result in the formation of a single transmembrane pore that allows the influx of calcium ions, leading to cell death (8).


Scheme 1: Mechanism of cell intoxication by PVL. A) Secreted, water-soluble LukF-PV and LukS-PV monomers bind to the membrane surface, B) assemble into heterodimers, C) further oligomerize into heterotetramers that are characterized by alternating LukF-PV and LukS-PV subunits, D) these heterotetramers assemble into a disc-like, octameric pre-pore structure that is comprised of alternating LukS-PV and LukF-PV subunits in a 1:1 stoichiometry (7), and E) pre-pore structure undergoes major conformational changes that result in the formation of a single transmembrane pore that allows the influx of calcium ions, leading to cell death.

The crystal structures of the monomeric forms of both LukF-PV (9) and LukS-PV (10) have each been determined to 2.0 Å, and these structures provide useful insights into their biological function. As shown in Figure 1, the LukF-PV and LukS-PV possess similar oblong elliptical shapes with molecular dimensions of ~70Å × ~35 Å × ~25 Å. Both structures are rich in β-sheet structure and are defined by two characteristic domains, a β-sandwich and a rim domain. Positioned at the top of each structure is the β-sandwich or cap domain, which is formed by two antiparallel β-sheets. Also at the top of the monomer structure is the pre-stem region, which contains a β-hairpin that is inserted across the membrane during the pore formation. In addition, the cap domain includes a triangle region that functions as a transition between the cap and stem domains in the pore form. At the bottom of the two toxin structures is the smaller rim domain, which is responsible for binding to the membrane surface, and is characterized by hydrophobic residues that anchor the toxins into hydrophobic interior of the lipid bilayer


Figure 1: Crystal structures of LukF-PV (PDB code 1PVL) and LukS-PV (PDB code 1T5R). The structures are rendered in ribbon with sheet, helical and loop structures colored cyan, orange/red, and purple, respectively. The two proteins are rich in β-structure and are comprised of a β-sandwich and rim domain. The top of the structures also contains the pre-stem region which unfolds to form the transmembrane β-barrel and a triangle region that functions as the transition from the cap domain to the stem domain in the pore structure.

At present, the three-dimensional structure of oligomeric PVL has not been determined. In this absence, the well-known crystal structure of heptameric α-hemolysin in its pore form (11) has been extensively used as a paradigm for PVL. However, there are significant differences between α-hemolysin and PVL. First, the heptameric α-hemolysin crystal structure consists of a single molecular constituent, whereas PVL is a binary component toxin. Accordingly, the heptameric α-hemolysin structure is characterized by a single type of protein-protein interface, whereas PVL contains two distinct protein-protein surfaces. Secondly, α-hemolysin assembles into a heptameric ring structure, whereas the functional state of PVL is an octameric ring-like structure. Accordingly, channel characteristics of the two toxins are different. Based on these factors, the determination of the octameric PVL structure is warranted. Of particular interest is the pre-pore conformation of PVL, since this model would explain previously published mutational results, and additionally, open the way for the design of chemotherapeutics at an earlier stage of SA infection.

Given the large size of the PVL structure, molecular mechanical methods are the only tools available for modeling PVL. Even with the amazing advances in computing power in recent years, the complex PVL macromolecular structure poses a tremendous modeling challenge, but recent works have shown that biochemically realistic models can be generated for large protein assemblies and to address important biochemical questions (12-23).

In this paper, we report an all-atom model of octameric PVL in its pre-pore conformation. A systematic approach was used to build the oligomer model from the crystal structures of the monomer subunits.

Materials and Methods General Method

A Dell Precision 690 workstation running Red Hat Enterprise Linux 4 was used to model the PVL octamer structure. Insight II (Accelrys, San Diego, CA) was used to build and visualize the models. Simulations were performed using the cff91 force field with the nonbonded interaction limited to within 13 Å. The cff91 force field contains the appropriate atom typing and atomic radii for protein modeling. Additional refinement and protein quality check was performed using the Maestro 9.0 program (Schrodinger LLC, New York). Because of the high number of atoms in the oligomeric models, implicit solvent was used in all simulations.

The quality of the protein model was assessed using Procheck (24,25) and HINT (26-28) programs. HINT was used to identify unfavorable hydrophobic-polar and base-base interactions in the model. HINT was employed using an exp (-1r) distance dependence for hydropathic constants on atom pair wise interactions. The distance function was set to 50% hydropathic and 50% steric, and the steric term is a Lennard-Jones 6-9 function which is compatible with the cff91 force field used throughout this study. Lone pairs for basic atoms were vector focused by a factor of 10. A tethered minimization protocol was used to optimize structures without significantly changing their conformations. Minimization involved up to 5000 steps of Fletcher-Powell optimization until the norm of the gradient was < 1.0 kcal/mol Ų. In the first step, the backbone atoms were fixed and a tethering force of 2000 kcal/mol Ų was applied to the side chain heavy atoms. After each cycle of minimization, tethering force was stepped off by a factor of 0.30 until all external force was removed. In the second step, a tethering force of 2000 kcal/mol Ų was applied to all heavy atoms and again stepped off by a factor 0.30.

Molecular dynamics incorporating distance constraints were used to bring together and move away atoms or pseudoatoms. Molecular dynamics were performed for 500 fs in 500 steps with a time step of 1 fs at 300K. This was followed by minimization which involved up to 5000 steps of Fletcher-Powell optimization until the norm of the gradient was < 1.0 kcal/mol Ų. For each cycle, the upper or lower limits of the distance constraints were decreased or increased, typically by increments of 0.01 Å, since this provided uniform and smooth translations of the domains. For all simulations, the force constant for the distance constraints was 100 kcal/mol and the restraint function was a flat-bottomed well. For distance constraints used to maintain secondary structure, the lower and upper distance limits were to set ± 0.5 Å of the distances between the hydrogen bonded heavy atoms measured from the crystal structures or the homology models.

Model building

The 1PVL(9), 1T5R (10) and 7AHL (11) crystal structures were used as templates. The structures of LukF-PV and LukS-PV monomers were obtained by the 1PVL and 1T5R crystal structures, respectively, while a template for assembling the dimers was the 7AHL crystal structure of heptameric α-hemolysin (11). To begin, LukF-PV and LukS-PV monomers were extracted from the 1PVL and 1T5R crystal structures, respectively. Missing residues in each structure were modeled into the polypeptide. The resulting structures were energy refined using tethered minimization (14). Since each subunit possess two interacting faces in the octamer ring-like structure, it was necessary to construct two different LukF-PV/LukS-PV heterodimer models to fully elucidate the binding interactions on each subunit. These two molecular models were identified as F_R-S^L, and S_R-F_L. This nomenclature is based on the side-by-side relationship of the LukF-PV and LukS-PV constituents in the octameric ring-like structure. F and S represent the two classes, and subscript R and L denote right and left faces, respectively. If one views the two subunits from inside the channel lumen, in the F_R-S^L model, the right side of LukF-PV (F_R) is bound with the left side of LukS-PV (S_L) and alternatively, in the S_R-F_L model, the right side of LukS-PV (S_R) is bound with the left side of LukF-PV (F_L).

The F_R-S_L model was assembled by superimposing the modeled LukF-PV and LukS-PV structures with the corresponding β-sandwiches of subunits A and B in the heptameric α-hemolysin 7AHL crystal structure, respectively. Subsequently, the atom-atom contacts at the LukF-PV/LukS-PV interface were optimized, using state-of-the-art modeling methodologies previously developed by our group (14). The bad contacts across the protein-protein interface were determined using the HINT program as well as Maestro 9.0 structure analysis tool. To relieve the unfavorable interactions, iterative cycles of distance-constrained energy minimization and molecular dynamics in a CFF91 force field were used, followed by hydropathic analysis with the HINT program to determine the interaction quality (14). A model of S_R-F_L heterodimer was constructed in the same fashion with the initial alignment obtained from an alternative alignment of the two monomer crystal structure with the B and C subunits of the 7AHL crystal structure instead.

The next step was to arrange the F_R-S_L and S_R-F_L models into a tetrameric structure. Because of the alternating stoichiometry of LukF-PV and LukS-PV in the octameric ring structure, the tetramer was assembled by first, superimposing FR-SL over subunits A and B in the 7AHL crystal structure and S_R-F_L over subunits B and C and then removing the SR structure. A second F_R-S_L structure was then superimposed onto the terminal F_L unit of the existing trimer and the overlapping FR unit was removed, to give a tetramer of alternating LukF-PV and LukS-PV subunits. The tetramer model was energy refined to remove unfavorable protein-protein contacts. To assemble the octamer structure, the energy-refined tetramer structure was copied in the same Cartesian space, and the second tetramer copy was rotated 180° about the z-axis, which is the channel axis in the octamer ring-like structure and subsequently translated along the xy-plane until the protein interfaces for the A and H subunits and for the D and E subunits were not overlapped. At this stage, the A/H and D/E protein interfaces are not optimal. Additionally, the pre-stem region (residues 109-131) of different LukS-PV subunits overlapped one another in the channel lumen, essentially occupying the same Cartesian space. Based on the unfavorable packing of the pre-stem region in the channel lumen, it is apparent that the conformation of the pre-stem region in monomeric LukS-PV is not necessarily its conformation in the pre-pore octameric ring structure.

Accordingly, the protein conformations at the two A/H and D/E interfaces as well as in the pre-stem structures in the channel lumen were refined. First, with subunits B-G fixed in Cartesian space, the A/H protein interface was optimized using our standard protocol. This was followed by optimization of the D/E protein interface using the same protocol. Using the loop refinement protocol in the Modeler module in InsightII, the pre-stem structures in the D and H subunits of the octamer model were modeled. The best loop conformations were selected. Then the all-atom octamer model was energy minimized using Schrodinger’s Macromodel program in an OPLS 2005 force-field and a distance-dependent dielectric. The model was structurally evaluated using Schrodinger’s Protein Report program, which identifies unfavorable stereoelectronic interactions. The protein quality was further evaluated using Procheck (25). The final structure consists of 36556 atoms characterized by eight alternating LukF-PV (residues 1-301) and LukS-PV (residues 7-284) subunits.

Results

Modeling the PVL Octamer

The starting building blocks for the PVL octamer model were the monomer LukF-PV and LukS-PV structures, which were obtained from the 1PVL and 1T5R crystal structures, respectively. To fashion the heterodimer and then the heterotetramer as shown in Scheme 1, the monomeric LukF-PV and LukS-PV subunits were arrayed in an alternating fashion using the subunits in the 7AHL crystal structure of heptameric α-hemolysin (11) as a template. The protein-protein interfaces within the heterodimer and between heterodimers were refined using a previously described modeling protocol (13,14). In brief, bad contacts across the protein-protein interface were determined using the HINT program as well as Maestro 9.0 structure analysis tool. To relieve the unfavorable interactions, iterative cycles of manual adjustment of torsion bond angles, distance-constrained energy minimization and molecular dynamics in a CFF91 force field were used. This was followed by hydropathic analysis with the HINT program to determine the interaction quality (14).

The next step was to assemble a preliminary octameric PVL structure. The refined PVL heterotetramer was copied, and the second copy was rotated and translated to form a ring-like structure of octameric PVL. Again the protein-protein interfaces were energetically and hydropathically refined as previously described (13,14). The remaining crucial challenge was modeling the fold of the pre-stem regions inside the channel lumen of this disc-like structure. Residues in the pre-stem region of both the monomer 1PVL and 1T5R starting structures were missing. When the missing residues were modeled into the subunits of the initial octamer PVL structure, it was evident that the eight pre-stem regions were overlapped in Cartesian space. Therefore, the conformations of these projected pre-stem regions from the monomer crystal structures were not remotely similar to those present in the octameric form.

To optimize the molecular packing of the pre-stem regions in the channel lumen, multiple conformations of the pre-stem regions were modeled and examined. The best pre-stem conformations were selected and mapped into the octamer PVL structure. The 36,556-atom macromolecular structure was then energy minimized to give the reported octamer PVL model.

Overall Structure

As shown in Figure 2, the disc-like structure of the ~264 kDa octameric PVL has a diameter of ~115 Å and a height of ~70 Å. Within the ring structure, the LukF-PV and LukS-PV subunits are arrayed in an alternating fashion, and each subunit is characterized by a ~45° tilt angle relative to the channel axis. The rim domains of the eight subunits are positioned at the bottom of the macromolecular structure and are oriented for membrane binding.

The LukF-PV and LukS-PV constituents in the pre-pore PVL model retained their overall secondary structure from their monomer crystal structures. In the octamer structure, LukF-PV and LukS-PV subunits have a backbone r.m.s. deviation of ~1 Å relative to their conformations in the 1PVL and 1T5R monomer crystal structures. This low r.m.s. deviation value indicates that the conformations of the starting monomer crystal structures are within the natural dynamic motion of the monomer structures in solution.

Octamerization of the LukF-PV and LukS-PV constituents buries a surface area of ~1,600 Ų for each subunit and results in a PVL octamer with a surface area of 66,800 Ų. As shown in Figure 3A, the subunits in the octameric structure are tightly packed. Mapping the electrostatic potential onto the surface of the PVL octamer model shows that there are only small patches of acidic and basic surfaces. A cut-away view of the protein structure reveals that the channel interior has a bottom-heavy, hour-glass shape (Figure 3B). The top of the channel has a diameter of ~35 Šbut narrows to ~5 Šin diameter near the top of the structure due to the presences of the pre-stem regions from the eight subunits lining the channel interior. Moving downwards pass the pre-stem regions, the channel regains its ~35 Šdiameter and progressively increases in diameter until a ~55 Šdiameter is reached at the base of the structure.

As shown in Figure 4A, a top view of the PVL octamer structure illustrates how the folding of the pre-stem regions from the eight subunits occludes the channel lumen to a narrow ~5 Å diameter opening. Since the PVL octamer structure is a static pre-pore model, it is possible that the natural motion of the pre-stem regions may result in complete closure of the channel lumen. This occlusion contrasts with the base of the PVL octamer structure, which has an opening of ~55 Å in diameter that likely remains constant (Figure 4B). A side view of the PVL octamer model (Figure 3A) shows that the electrostatic surface of the membrane-anchoring regions is largely hydrophobic.

Protein-protein interface

While the protein interfaces in the PVL octamer model are nonidentical, they display similar interchain interactions. Listed in Table I are the hydrogen bonds at one such F_R-S_L and one such S_R-F_L interface. With 22 hydrogen bonds, the F_R-S_L interface is stabilized by significantly more hydrogen bonds than the S_R-F_L interface with its 5 hydrogen bonds. Conversely, the S_R-F_L interface is stabilized by more hydrophobic interactions compared to that of the F_R-S_L interaction. Figure 2: Disc-like structure of octameric PVL model. The structure is rendered in ribbon with LukF-PV colored orange, and LukS-PV colored green. (A) A side view shows that the constituents have a ~45° tilt angle relative to the channel axis. (B) A top view shows that the pre-stem regions of each subunit fill the interior of the channel lumen and occlude the top of the channel.

Figure 3: Views of the protein surface of PVL octamer model orthogonal to the channel axis. (A) The protein-protein surfaces are tightly packed. The electrostatic surface potential shows a dispersion of charge with the absence of large patches of electronegative (red) or electropositive (blue) surface. (B) Cut-away of interior of the disc-like structure of PVL octamer shows the shape and contour of the channel. The protein is rendered as an orange mesh surface with the subunits also shown as colored ribbons. The channel resembles a bottom-heavy, hour glass. Starting from the top, the channel diameters moves from ~35 Å to ~5 Å where the pre-stem motifs are folded and then to ~55 Å at the base. Figure 4: Electrostatic surface rendering of PVL octamer model colored red for electronegative and blue for electropositive potentials. (A) A top view shows a cluster of positive electrostatic potential at the channel opening. The channel narrows to a ~5 Å diameter where the pre-stem regions from the eight subunits are located. (B) A bottom view shows that the channel interior is not highly charged and that the membrane anchor regions in the rim domain are largely hydrophobic.

Among the 27 hydrogen bonds, there are two salt bridge interactions at the SR-FL interface: 1) Asp90 of LukS-PV to His3 of LukF-PV, and 2) His173 of LukS-PV to Glu209 of LukF-PV, and a single salt bridge interaction at the FR-SL interface: Asp47 of LukF-PV to Arg16 of LukS-PV. Salt bridge interactions play an important role in molecular recognition and interface stability (29). However, these three identified salt bridge interactions are located at the top of channel in the PVL structure as in the case of the Asp90-His3 and Asp47-Arg16, and at the bottom as in the case of the H173-E209 salt bridge. This topology modulates the roles of these salt bridge interactions in oligomerization. A 50 ps molecular dynamics simulation showed that these three salt bridge interactions persisted during the time course. At the interior of the S_R-F_L interface, a hydrophobic cluster contributes to oligomer stability. As shown in Figure 5A, a hydrophobic linear patch on the S_R side is formed by the hydrophobic side chains of Leu91, Tyr94, and Val146. This is matched on the F_L side of the S_R-F_L interface by a linear patch composed of the hydrophobic residues Ile35, Ala56, Phe221, and Pro223. At the center, Tyr94 of LukS-PV is wedged itself into a hydrophobic pocket in LukF-PV.

The F_R-SL interface is also stabilized by hydrophobic packing. At the center of the F_R-S_L interface, the aromatic ring of the LukS-PV Phe53 is packed against the side chains of LukF-PV Tyr99 and Leu153 (Figure 5B).

Discussion

Structural Basis for Mutational Results

Because the PVL octamer model is consistent with experimental data, it provides dimensionality to biophysical data.

Residue Thr28 in LukS-PV is homologous to His35 in α-hemolysin, which was previously shown to be crucial to the assembly and biological function of α-hemolysin (30-32). In work by Guillet et al., (10), Thr28 in LukS-PV was found to be similarly important to the activity of wild-type PVL toxin. A conservative Thr28Ser mutation gave no apparent decrease in toxin activity relative to the wild-type; less conservative mutations as in the Thr28Cys and Thr28His mutants resulted in slightly less activity than wild-type. Finally, Guillet et al., found that more radical mutations in Thr28Leu, Thr28Phe, Thr28Asn, and Thr28Asp resulted in inactive mutants.

Because Thr28 is a surface residue in the 1T5R crystal structure of LukS-PV, the available crystal structure can not explain these mutational results. In the monomer 1T5R crystal structure, the microenvironment around Thr28 is formed by the side chains of Ser21 and Phe53 from neighboring β-strands and the side chain ofAsn30 from its own β-strand. Given this, simple in silico amino acid replacement shows that the side chains of inactive mutants can fit into the stereoelectronic space provided by the 1T5R monomer crystal structure. For example, the steric bulk of the inactive Thr28Phe mutant and the negative charge of the inactive Thr28Asp mutant do not clash with the surrounding protein structure, and therefore, the inactivity of these mutants can not be structurally explained.
,br> In contrast, the PVL octamer model provides a structural basis for these results. In the octamer model, the LukS-PV Thr28 side chain is tightly packed against residues Val96, Asn158 and Phe159 from the neighboring LukF-PV subunit. As shown in Figure 6, the side chain hydroxyl of Thr28 in LukS-PV is potentially hydrogen bonded to the Phe159 backbone NH of LukF-PV. Additionally, the side chain methyl of Thr28 in LukS-PV is packed against the hydrophobic side chains from Val96 and Phe159. At one LukS-PV-LukF-PV interface, the carbon-to-carbon distances of the Thr28 –CH₃ to Val96 γCH₃ and Phe159 ΒCH₂ are ~5.3 Å and ~3.6 Å, respectively. These distances indicate the close packing of the LukS-PV Thr28 side chain to residues in the neighboring LukF-PV subunit. In the LukS-PV Thr28Ser mutant, the interactions involving the Thr28 –CH3 to Val96 γCH₃ and Phe159 βCH₂ are lost. While it has been estimated that the intrinsic binding energy of a methyl group is 0.8 kcal/mol (33), this loss in binding energy is insufficient to disrupt the biological function of LukS-PV. This is consistent with the similar activity of the Thr28Ser mutant to wild-type.



Figure 5: Hydrophobic interactions at the two distinct protein-protein interfaces in the PVL model. The identified residues in LukF-PV and LukS-PV are rendered in CPK surface with hydrogen atoms not shown for clarity. LukF-PV and LukS-PV are colored orange and green, respectively. (A) SR-FL interface. (B) FR-SL interface.



Figure 6: Interactions of LukS-PV Thr28 residues across the protein-protein interface in the octamer model used to explain mutational data at that position (10). Residues are rendered in stick with hydrogen, nitrogen and oxygen atoms colored white, blue and red, respectively. Carbon atoms of LukS-PV and LukF-PV are colored green and purple, respectively. Hydrogen bonds are represented by yellow dashed lines. Nonpolar hydrogen atoms were removed for clarity.

In contrast, in the LukS-PV Thr28Cys mutant, this loss in methyl binding energy is coupled with the replacement of a strong O-H•••N hydrogen bond to the Phe159 backbone NH with a weaker S-H•••N hydrogen bond. These two different losses in binding energy would explain the decreased activity of the Thr28Cys mutant relative to wild-type. Similarly, for the LukS-PV Thr28His mutant, our model indicates that the imidazole side chain from histidine is potentially hydrogen bonded to the backbone oxygen atom of Val96 or the side chain oxygen of Asn30; this hydrogen bonding interaction may allow the LukS-PV Thr28His to retain some activity, although histidine is bulkier and more polarized than threonine, which creates unfavorable interactions with the residues of the neighboring subunit.

Furthermore, modeling of the Thr28Leu and Thr28Phe mutants indicates that larger side chains such as those of leucine and phenylalanine can not be sterically accommodated at the interface. Finally, substitutions with negatively charged and polarized residues such as aspartate and asparagine resulted in unfavorable acid-acid clashes with the Val96 backbone oxygen from the neighboring LukF-PV subunit as well as acid-hydrophobic clashes with the Val96 hydrophobic side chain.

Pre-stem Region Fold

The PVL octamer model presents an important snapshot of PVL’s mechanism of action. Pore formation involves major conformational changes in the pre-pore structure. A central step is the unfolding of the eight pre-stem regions in the channel lumen before insertion into the membrane. The model shows how these flexible motifs are oriented in the channel lumen. This structural detail is absent in both the LukF-PV and LukS-PV monomeric structures, since the pre-stem segments were not located in the electron density map of these two reported crystal structures.

Using molecular modeling, we were able to construct and visualize a biochemically feasible topology for these pre-stem regions. In the octamer model, the missing residues in the pre-stem regions were modeled, folded, and refined inside the channel lumen. Facilitating their compact fold in the channel is the fact that the pre-stem regions are glycine-rich and consist of loop structures. For LukS-PV, this loop structure encompasses residues 106 to 137 and for LukF-PV, it is a residue 124-135 segment. In constructing the PVL octamer model, it was soon realized that the conformations of pre-stem regions in the monomer are dramatically different than those in the octamer structure. By modeling the pre-stem regions in all eight subunits of the PVL pre-pore structure, we present a paradigm of how a complex oligomer macromolecular can efficiently pack its membrane-inserting motifs inside a small diameter channel structure. Figure 7 shows the molecular packing of the pre-stem loops at the F-S-F interfaces. The octamer model also depicts another key component in PVL’s mechanism of action in the relative topology of the N-terminal segments of LukF-PV and LukS-PV. In α-hemolysin, the amino-terminal segment, which is also termed the amino latch, is responsible for initiating oligomerization (34). In the PVL octamer model, the N-terminal segments of LukS-PV and LukF-PV are positioned atop the cluster of pre-stem regions at the channel lumen. The interactions of the N-terminal segments with the pre-stem regions provide additional stability across the different subunits and may also trigger oligomerization. The N-terminal segments may “push” down on the pre-stem cluster in order to initiate pore formation.

Figure 7: Stereo view of the pre-stem regions from LukS-PV (subunit S in the model, colored purple) and LukF-PV (subunit G in the model, colored cyan). The protein is rendered in blue ribbon with sheet structure colored blue and loop structure grey. The N-terminal residue Ile7 has its backbone drawn in purple CPK to depict the relative orientation of the N-terminal segment above the pre-stem motif. Since the N-segment sits atop of the pre-stem regions of LukF-PV and LukS-PV, it potentially promotes pore formation by “pushing” down on the cluster of pre-stem structures at the channel lumen.

Conclusion

We have modeled the pre-pore conformation of octameric PVL in order to provide a structural basis for understanding experimental results and for extending a therapeutic paradigm to an earlier stage of toxin assembly. As described herein, the model rationalizes the activity of the reported mutants at position 28 of LukS-PV, whereas the current crystal structure does not. Additionally, the model delineates a potential mechanism of oligomerization and pore-formation involving the amino latch and pre-stem region of the toxin constituents.

Acknowledgment

This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract N01-CO-12400. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. This research was supported in part by the Developmental Therapeutics Program in the Division of Cancer Treatment and Diagnosis of the National Cancer Institute.

References

1. V. Nizet. J Allergy Clin Immunol 120, 13-22 (2007).
2. B. L. Kotzin, D. Y. Leung, J. Kappler, and P. Marrack. Adv Immunol 54, 99-166 (1993).
3. R. D. Meyer, S. R. Monday, G. A. Bohach, and S. M. Schlievert. Int J Infect Dis 5, 163-166 (2001).
4. H. J. Schuberth, C. Krueger, H. Zerbe, E. Bleckmann, and W. Leibold. Vet Microbiol 82, 187-199 (2001).
5. H. F. Chambers. The New England Journal of Medicine 352, 1485-1487 (2005).
6. J. Kaneko and Y. Kamio. Bioscience, Biotechnology, and Biochemistry 68, 981-1003 (2004).
7. L. Jayasinghe and H. Bayley. Protein Sci 14, 2550-2561 (2005).
8. V. T. Nguyen, Y. Kamio, and H. Higuchi. The EMBO Journal 22, 4968-4979 (2003).
9. J. D. Pedelacq, L. Maveyraud, G. Prevost, L. Baba-Moussa, A. Gonzalez, E. Courcelle, W. Shepard, H. Monteil, J. P. Samama, and L. Mourey. Structure 7, 277-287 (1999).
10. V. Guillet, P. Roblin, S. Werner, M. Coraiola, G. Menestrina, H. Monteil, G. Prevost, and L. Mourey. The Journal of Biological Chemistry 279, 41028-41037 (2004).
11. L. Song, M. R. Hobaugh, C. Shustak, S. Cheley, H. Bayley, and J. E. Gouaux. Science New York, NY 274, 1859-1866 (1996).
12. D. Josa, E. F. F. Da Cunha, T. C. Ramalho, T. C. S. Souza, and M. S. Caetano. Journal of Biomolecular Structure and Dynamics 25, 373-376 (2008).
13. T. L. Nguyen. Journal of Biomolecular Structure & Dynamics 22, 253-265 (2004).
14. T. L. Nguyen, G. Schoehn, W. Weissenhorn, A. R. Hermone, J. C. Burnett, R. G. Panchal, C. McGrath, D. W. Zaharevitz, M. J. Aman, R. Gussio, and S. Bavari. J Struct Biol 151, 30-40 (2005).
15. B. P. Mukhopadhyay, B. Ghosh, H. R. Bairagya, T. K. Nandi, B. Chakrabarti, and A. K. Bera. Journal of Biomolecular Structure & Dynamics 25, 543-551 (2008).
16. V. I. Poltev, V. M. Anisimov, V. I. Danilov, A. Deriabina, E. Gonzalez, A. Jurkiewicz, A. Les, and N. Polteva. Journal of Biomolecular Structure and Dynamics 25, 563-571 (2008).
17. G. Barone, C. F. Guerra, N. Gambino, A. Silvestri, A. Lauria, A. M. Almerico, and F. M. Bickelhaupt. Journal of Biomolecular Structure and Dynamics 26, 115-129 (2008).
18. G. H. Braun, D. M. M. Jorge, H. P. Ramos, R. M. Alves, V. B. Da Silva, S. Giuliatti, S. V. Sampaio, C. A. Taft, and C. H. T. P. Silva. Journal of Biomolecular Structure and Dynamics 25, 347-355 (2008).
19. J. Yoon, J. Park, S. Jang, K. Lee, and S. Shin. Journal of Biomolecular Structure and Dynamics 25, 505-515 (2008).
20. V. Rajakrishnan, V. R. Manoj, and G. Subba Rao. Journal of Biomolecular Structure and Dynamics 25, 535-542 (2008).
21. A. N. Nekrasov and A. A. Zinchenko. Journal of Biomolecular Structure and Dynamics 25, 553-561 (2008).
22. E. D. Akten, S. Cansu, and P. Doruker. Journal of Biomolecular Structure and Dynamics 27, 13-25 (2009).
23. H. M. Martinez, J. V. Maizel Jr, and B. A. Shapiro. Journal of Biomolecular Structure and Dynamics 25, 669-683 (2008).
24. R. A. Laskowski, M. W. MacArthur, and J. M. Thornton. Curr Opin Struct Biol 8, 631-639 (1998).
25. R. A. Laskowski, J. A. Rullmannn, M. W. MacArthur, R. Kaptein, and J. M. Thornton. J Biomol NMR 8, 477-486 (1996).
26. G. E. Kellogg, S. F. Semus, and D. J. Abraham. Journal of Computer-aided Molecular Design 5, 545-552 (1991).
27. A. Marabotti, L. Balestreri, P. Cozzini, A. Mozzarelli, G. E. Kellogg, and D. J. Abraham. Bioorganic & Medicinal Chemistry Letters 10, 2129-2132 (2000).
28. E. C. Meng, I. D. Kuntz, D. J. Abraham, and G. E. Kellogg. Journal of Computer-Aided Molecular Design 8, 299-306 (1994).
29. M. Vijayakumar, K. Y. Wong, G. Schreiber, A. R. Fersht, A. Szabo, and H. X. Zhou. Journal of Molecular Biology 278, 1015-1024 (1998).
30. R. Jursch, A. Hildebrand, G. Hobom, J. Tranum-Jensen, R. Ward, M. Kehoe, and S. Bhakdi. Infection and Immunity 62, 2249-2256 (1994).
31. S. Bhakdi, R. Jursch, M. Broker, H. Ronneberger, and K. D. Hungerer. Behring Institute Mitteilungen, 80-84 (1994).
32. B. E. Menzies and D. S. Kernodle. Infection and Immunity 62, 1843-1847 (1994).
33. P. R. Andrews, D. J. Craik, and J. L. Martin. Journal of Medicinal Chemistry 27, 1648-1657 (1984).
34. L. Jayasinghe, G. Miles, and H. Bayley. The Journal of Biological Chemistry 281, 2195-2204 (2006).