Differential Effect of Solution Conditions on the Conformation of the Actinoporins Sticholysin Ii and Equinatoxin Ii

Actinoporins are a family of pore-forming proteins with hemolytic activity. The structural basis for such activity appears to depend on their correct folding. Such folding encompasses a phosphocholine binding site, a tryptophan-rich region and the activity-related N-terminus segment. Additionally, different solution conditions are known to be able to influence the pore formation by actinoporins, as for Sticholysin II (StnII) and Equinatoxin II (EqtxII). In this context, the current work intends to characterize the influence of distinct solution conditions in the conformational behavior of these proteins through molecular dynamics (MD) simulations. The obtained data offer structural insights into actinoporins dynamics in solution, characterizing its conformational behavior at the atomic level, in accordance with previous experimental data on StnII and EqtxII hemolytic activities.

It is proposed that the phosphocholine binding site is responsible for the recognition of its molecular target and interaction between the Trp-rich region and the lipid bilayers (Mancheño et al. 2003, Hong et al. 2002).Such event is followed by an N-terminal displacement into the lipid bilayer, thought to be mandatory for the functional pore formation (Casallanovo et al. 2006, Kristan et al. 2004).
The exposure of StnII and EqtxII to different solution conditions, including urea solutions and different pH values, affects the hemolytic activity of these proteins, with variable degrees of conformational changes (Ulrih et al. 2004, Álvarez et al. 2001, Mancheño et al. 2001, Martínez et al. 2001, Malavasic et al. 1996).The main conformational differences between NMR and crystal structures of these toxins lie on the C-terminus α-helix (Gln128-Tyr136 in StnII and Gln130-Tyr138 in EqtxII), at the phosphocholine binding site (Álvarez et al. 2009, Miles et al. 2008, Hinds et al. 2002).Since actinoporins are considered a potential target for biotechnological applications (Tejuca et al. 2009), the current work intends to characterize the dynamics of StnII and EqtxII under distinct physical-chemical conditions as a strategy to gain insights into the structural basis for the actinoporins hemolytic activity.

NOMENCLATURE AND SOFTWARE
The nomenclature guidelines proposed by IUPAC (1970) were used.The molecular dynamics (MD) calculations and analyses were performed using GROMACS simulation suite (van der Spoel et al. 2005) and GROMOS96 43a1 force field (van Gunsteren et al. 1996).The manipulation of structures was carried out with VMD (Humphrey et al. 1996), Pymol (Delano 2002) and Swiss PDB viewer (Guex and Peitsch 1997).The monitoring of secondary structure content was performed by PROCHECK (Laskowski et al. 1993) and ellipticity at 222 nm was calculated with g_helix tool from GROMACS package, according to Hirst and Brooks (Hirst and Brooks 1994), and plotted using SigmaPlot v.11.

MOLECULAR DYNAMICS SIMULATIONS
The crystal structures of actinoporins StnII (PDB ID 1GWY) and EqtxII (PDB ID 1IAZ) were retrieved from Protein Data Bank and the respective monomeric forms were employed in the simulations.Each monomer was solvated in a triclinic box using periodic boundary conditions and SPC water model (Berendsen et al. 1981).Counter ions were added to neutralize the systems whenever necessary.Amino acids ionization was automatically adjusted to pH = 7.0 or pH = 3.0 within GROMACS according to the standard pKa values of the amino acids side chains.Accordingly, the side chains of residues Asp18, Asp36, Asp56, Asp76, Asp107, Asp116, Asp127, Asp133, Asp143, Glu22, Glu23, Glu38, Glu61, Glu132, Glu148, Glu166, and His147 in StnII,and Asp10,Asp17,Asp38,Asp58,Asp78,Asp96,Asp109,Asp129,Asp146,Glu24,Glu40,Glu134,Glu135,Glu173,His63,His67,His150,His169,and His175 in EqtxII were protonated when in acidic medium with no change to the water model.In neutral pH, these residues are in its non-protonated forms.A solution of 10.0 M urea was also used for StnII simulation (Rocco et al. 2008).The employed MD protocol was based on previous studies (de Groot and Grubmüller 2001).The Lincs method (Hess et al. 1997) was applied to constrain covalent bond lengths, allowing an integration step of 2 fs after an initial energy minimization using Steepest Descent algorithm.Electrostatic interactions were calculated using the Particle Mesh Ewald method (Darden et al. 1993).Temperature and pressure were kept constant by coupling proteins, ions, and solvent to external temperature and pressure baths with coupling constants of τ = 0.1 and 0.5 ps (Berendsen et al. 1984), respectively.The reference temperature was set to 310 K.The systems were slowly heated from 50 to 310 K, in steps of 5 ps, each one increasing the reference temperature by 50 K, allowing a progressive thermalization of the systems.Each simulation was then performed for 0.1 μs, without any restraint.

SIMULATION STABILITY
The overall progression of the performed simulations was first assessed by root mean square deviation (RMSD).Each system had their RMSD evaluated taking the respective crystallographic EDSON V.F.FAUTH, EDUARDO M. CILLI, RODRIGO LIGABUE-BRAUN and HUGO VERLI structure as reference in function of time (Figure 2).The MD simulations of StnII in physiologic and acidic conditions (Systems 3 and 4, black and blue, respectively) were stable up to 0.04 μs (Figure 2B).After this period, the deviation from starting structure increased under acidic condition.This increase seems to be mainly related to partial unfolding of C-terminal α-helix (Figure 2D and Figure 3F).On the other hand, this C-terminal helix on EqtxII seems to be more stable in acidic conditions (Figures 2A,2C and 3E).There was secondary structure loss during the trajectory of StnII at pH 3.0, as observed via pseudo dihedral ξ (Figure 3F).Considering this observation and aiming to further explore the influence of pH on StnII conformational stability, the last frame from the acidic simulation of StnII was used to start a new simulation in neutral pH.During this simulation, the evaluation of the C-terminal α-helix RMSD indicated a reoccurrence of the conformation adopted by this helix in pH 7.0 (System 5, red curve in Figure 2D and Figure 3F for the pseudo dihedral ξ).
This observation, points to the reversibility of the conformational modifications observed for StnII between acidic and neutral solutions.
The highly concentrated urea solution caused an increase of hydrophilic solvent-accessible surface (SAS) for StnII (Figure 4A).Nonetheless, no major difference could be observed for its RMSD (Figure 2A), indicating that, at least in the simulated time scale, these conditions do not significantly affect the protein structure stability, despite the increased C-terminal α-helix (Figure 2C).DIFFERENTIAL DYNAMICS OF ACTINOPORINS IN SOLUTION STRUCTURAL FLUCTUATION While the RMSD data presented in Figure 2 present a global perspective on the protein conformational behavior during the performed simulations, they lack resolution at the residue level.To overcome this limitation, the flexibility and helical content of EqtxII and StnII as a function of both time and residue numbers were evaluated by root mean square fluctuation (RMSF) and ellipticity at 222 nm (Figure 5), as previously described (Pol-Fachin et al. 2009, Terra et al. 2007, Verli et al. 2007, Verli and Guimarães 2005).The respective plots were drawn from the compilation of average data at every 200 ps of trajectory.
These analyses revealed that upon acidification the loop regions connecting strands β5 and β6 and strands β8 and β9, presented increased flexibility in EqtxII and decreased flexibility in StnII (Figures 5B and 5D, respectively).On the other hand, the flexibility observed for the protein N-terminal region for both EqtxII and StnII (Figure 5) was similar during their trajectories in acidic and physiological solutions.
Such similar flexibility in acidic condition was also observed when comparing the two proteins, including increased flexibility of the loop between strands β6 and β7 (Figures 5B and 5D,respectively).It is noteworthy that the loop between strands β6 and β7 and the beginning of strand β7, correspond to the Experimental data support that acidic solutions are capable of increasing the interaction between StnII and EqtxII with hydrophobic probes (Álvarez et al. 2001, Martínez et al. 2001, Malavasic et al. 1996), and that some pore-forming proteins may undergo partial unfolding triggered by membrane surfaces (Parker and Feil 2005, Bychkova et al. 1996, van der Goot et al. 1991).Previous works (Ulrih et al. 2004, Mancheño et al. 2001) also indicate that acidic conditions may be employed in the study of EqtxII and StnII to mimic the conformational modifications induced by the membrane surface, suggesting that such conditions may induce conformational changes suitable for molecular target recognition.
Considering that the analyses of RMSD and RMSF for EqtxII (Systems 1-2) and StnII (Systems 3-4) pointed to modifications in the secondary structure content and flexibility on the C-terminal α-helix of these toxins, its α-helical content was characterized through ellipticity.Ellipticity is defined as the percentage of time in which the residues present a helical nature as a function of time and residue numbers (Hirst and Brooks 1994) (Figure 5).Accordingly, the increased flexibility in this region seems to be related to the stability of its helical content over the simulation time for EqtxII (Figures 5B and 5H, Figures 6A and 6B) and for StnII (Figures 5D and 5J).In addition, our results corroborate synchrotron radiation CD analysis for EqtxII (Miles et al. 2008), indicating that this secondary structure may change in solution (Figures 5G and 5H), although in acidic solutions a high degree of conformational instability in the C-terminal α-helix may be observed (Figure 6B).
When the last frame of StnII, simulated at pH 3.0 is modified to pH 7.0 (i.e.is, a return of StnII to physiological pH), the beginning of the reversion  process is observed, with a consequent decrease in flexibility and increase in helical content, close to 0.1 μs of simulation (Figures 5E, 5K and Figure 7).Such behavior suggests that acidic environments may be able to induce partially unfolded states for StnII and EqtxII, mainly in the C-terminal α-helix.This observation agrees with the NMR structure of EqtxII, which was obtained in acidic solution and is partially unfolded when compared to the crystal structures of actinoporins (Figure 1).Considering the physicalchemical characteristic (e.g.dielectric constant, pH) and complex composition (e.g.lipids, proteins, carbohydrates) at the membrane surface (Konkul 2009, Parker and Feil 2005, Sui 2000), our results may be complementary to a previous publication (Lam et al. 2009) that presented MD simulations for EqtxII in DMPC membrane environment.
In opposition to its behavior in acidic medium, the simulation of StnII in a 10.0 M urea solution (System 6), presented lower flexibility for the entire protein (Figure 5F).This may be the reason for the small deviation in RMSD and radius of gyration from the starting structure (green curve in Figure 2A and Figure 4B).Furthermore, a comparison of EDSON V.F.FAUTH, EDUARDO M. CILLI, RODRIGO LIGABUE-BRAUN and HUGO VERLI radial distribution function of the solvent around the protein, when in neutral condition and in 10.0 M urea solution (Figures 2E and 2F,respectively), indicates that the water solvation shells of StnII are influenced by the high urea concentration (black curve in Figure 2F).Such process is accompanied by an increase of its hydrophilic SAS in relation to the starting X-ray-derived structure (Figure 4A), as well as by an increased number of hydrogen bond interactions between hydrophilic residues and the urea molecules that accumulate around them (green curve in Figures 2F, 4D, and 8).Such increased protein-solvent interaction may explain the lower flexibility on System 6 in comparison to the no-urea Systems.

STRUCTURAL REASONS FOR THE LACK OF STNII HEMOLYTIC ACTIVITY IN UREA SOLUTION
The conformational changes caused by different denaturation conditions (i.e.pre-incubation and incubation in extremely acidic and alkaline solutions, increasing concentration of strong denaturant agents like guanidine hydrochloride and urea) were used in the search for partially unfolded states of EqtxII and StnII.Such states were used as proxies for the pore formation in lipid membranes and red blood cells (Álvarez et al. 1998, 2001, Martínez et al. 2001, Malavasic et al. 1996, Belmonte et al. 1993).Accordingly, experimental results indicate that the conformational modifications induced by changing the pH are less preeminent than those caused by urea (Álvarez et al. 2001(Álvarez et al. , Malavasic et al. 1996)).In addition, most of the experimental data associated with these modifications were presented by fluorescent measurements, for which both EqtxII and StnII had their structures affected by an increased exposure in their Trp residues to aqueous solution (Álvarez et al. 2001(Álvarez et al. , Malavasic et al. 1996)).Such exposure reduced the hemolytic activity of StnII without major modifications in its secondary structure content (Martínez et al. 2001).
Our results in the MD simulation of StnII in 10.0 M urea solution (System 6) are in agreement with previous far-UV CD spectrum (Martínez et al. 2001), since the obtained trajectory did not present major changes in the protein secondary structure content.Our results also agree with fluorescence data, as the interaction between theses residues and solution molecules in System 6 was increased almost two-fold when compared to others systems (Table I).Furthermore, when this analysis was performed for residues encompassing the phosphocholine binding site, a similar behavior was observed for Pro105, Tyr111, Tyr135 and Tyr136 (Table I).Since Tyr135 and Tyr136 residues compose the C-terminal α-helix of StnII, this observation may be related to the increased deviation previously present in Figure 2C (green curve).
Analyses to verify if protein residues were preferentially interacting with urea molecules were also performed by measuring the interaction energy between protein residues and solution molecules.The results obtained (Table II) are in agreement with the data presented in Figure 2F and Figure 4D, as illustrated in Figure 8, suggesting that the molecular target recognition by StnII may not be favored in this condition.
Taking into account the necessary interaction with the membrane surface and the need for flexibility in the action mechanism of actinoporins, it seems that urea may modulate StnII hemolytic activity by hampering its ability to interact with membrane surfaces.It may also cause molecular rigidity, which may impair the conformational changes associated with pore formation.

CONCLUSIONS
In the current work, we performed a series of MD simulations of actinoporins EqtxII and StnII in different solution conditions in order to support the characterization of protein structure, conformation and dynamics, according to environmental conditions at the atomic level.
The solvation effects caused by urea may have a role in modulating hemolytic activity of StnII, as suggested by the calculations presented in this work.Such effects seem to be brought about with little to no loss in secondary structure content.Additionally, some pore-forming proteins underwent partial unfolding triggered by a local acid pH (Parker and Feil 2005, Bychkova et al. 1996, van der Goot et al. 1991).Thus, our results agree with previous studies (Ulrih et al. 2004, Mancheño et al. 2001) indicating that an acid condition can be employed in the study of EqtxII and StnII to mimic the effect of such acidic condition.In this way, our results show that during the simulations in acidic condition the folded structure of StnII undergoes a conformational modification, resembling the unfolded NMR conformation of EqtxII.Such partially unfolding was observed to be reversible and centered in functionally important regions of these proteins.There are indications that a more intricate behavior is required for these proteins to be active, including loop disorder and charge interactions for membrane contact, along with motif conservation and long-distance conformational rearrangements for proper pore formation and activity (Bakrac and Anderluh 2010, Pardo-Cea et al. 2011, García-Linares et al. 2013, 2014).Nonetheless, acidic medium models are still employed to study conformational changes in actinoporins, and our work presented atom-level observations regarding the conformational behavior of these proteins.Altogether, it may be expected that a deeper comprehension of these phenomena could contribute to future explorations of biological properties of actinoporins and to the rational design of new bioactive compounds.Palavras-chave: Actinoporinas, GROMACS, dinâmica molecular, proteínas formadoras de poros.

Figure 1 -
Figure 1 -Comparison between all atomic structures of actinoporins available.Crystal structure of StnII (PDB ID 1GWY), crystal structure of StnII complex with glycerol (PDB ID 1O71), crystal structure of StnII complex with phosphorylcholine (PDB ID 1O72), crystal structure of EqtxII (PDB ID 1IAZ) and NMR structure of EqtxII (PDB ID 1KD6).

Figure 2 -
Figure 2 -MD analyses as a function of time for Systems 1 (brown), 2 (indigo), 3 (black), 4 (blue), 5 (red), and 6 (green).All-atom RMSD from crystal structure is shown in (A) and (B) for entire protein, and in (C) and (D) for residues from the C-terminal α-helix.A comparative radial distribution function (RDF) of water molecules around StnII in System 3 and System 6 (black curve in E and F, respectively) and urea molecules around StnII in System 6 (green curve in F).

Figure 3 -
Figure 3 -MD analyses as a function of time in relation to the starting X-ray-derived model for System 1 (brown), System 2 (indigo), System 3 (black), System 4 (blue), System 5 (red), and System 6 (green) considering the following properties: root mean square deviation (RMSD) for entire protein (A) and (B), and for residues of C-terminal α-helix (C) and (D).Average pseudo dihedral ξ (i.e. the dihedral formed by 4 consecutive α-carbon atoms Cαi, Cαi +1 , Cαi +2 and Cαi +3 ) over residues of C-terminal α-helix (E) and (F), for which the reference value for a α-helical content is around 45° (see Section 3.2 for further details).

Figure 4 -
Figure 4 -MD analyses as a function of time in relation to the starting X-ray-derived structure considering the: (A) hydrophilic solventaccessible surface (SAS) and (B) gyration radious (Rg) for Systems 3 (black curve) and 6 (green curve).Additionally, the number of protein-protein (black curve) and protein-water (light blue curve) hydrogen bonds is presented for Systems 3 (C) and 6 (D), as well as the protein-urea hydrogen bonds (green curve) in System 6 (D).

Figure 6 -
Figure 6 -(A) All-atom root mean square fluctuation (RMSF) from 0.06 to 0.14 nm for EqtxII in acidic solution (System 2) over the 0.25 μs simulation as a function of both time and residue numbers are presented together with its corresponding (B) ellipticity from 0 to 80 % at 222 nm (percentage of time in which the residues present a helical nature).Both analyses range from blue to red according with the increase in flexibility or helical content.

Figure 7 -
Figure 7 -Conformational transitions of StnII over 0.3 μs MD simulation as a function of solution pH.The helical content was analyzed employing the progression of the average pseudo dihedral ξ, that is, the dihedral formed by 4 consecutive α-carbon atoms Cαi, Cαi +1 , Cαi +2 and Cαi +3 (see text and supplementary material for details).
Trp and phosphocoline binding site residues of StnII and solution molecules present in a 10.0 M urea solution.a Average values from the sum of Coulomb and Lennard-Jones components over the 0.1 μs MD simulation.

Figure 8 -
Figure 8 -Illustration of the of hydrogen bond interactions between hydrophilic residues in the C-terminal α-helix of StnII (i.e.Tyr135 and Tyr136) and urea molecules that accumulates around it.