Separation of protein corona from nanoparticles under intracellular acidic conditions: effect of protonation on nanoparticle-protein and protein-protein interactions.

Protein coronas separate from nanoparticles under intracellular acidic conditions however, competitive adsorption of multiple proteins and their protein network formation under different pH conditions have not yet been systematically studied at the atomic scale. Herein, we report all-atom molecular dynamics simulations of plasma proteins (human serum albumin and immunoglobulin gamma-1 chain C) adsorbed to 10 nm-sized carboxyl-terminated polystyrene (PS) nanoparticles at different protonation states that mimic extracellular and intracellular pH conditions of 7, 6-5, and 4.5. Binding free energies are calculated from umbrella sampling simulations, showing the significantly weakened binding between PS particles and proteins at the protonation state at pH 4.5, in agreement with experiments showing the separation of protein corona from nanoparticles at pH 4.5. Mixtures of multiple proteins and PS particles are also simulated, showing much less protein adsorption and protein cluster formation at the protonation state at pH 4.5 than that at higher pH values, which are further confirmed by calculating the diffusivities and hydrodynamic radii of individual proteins. In particular, electrostatic particle-protein and protein-protein interactions are significantly weakened by a combination of particle and protein protonation rather than by particle protonation alone, to an extent dependent on different proteins. These findings help explain the experimental observations regarding separation of protein corona from nanoparticles under intracellular acidic conditions at pH 4.5 but not at higher pH, supporting that acidification cannot be the only reason for this separation during the process of endosome maturation.

Differences in protein distribution, conformation, and dynamics in hard and soft coronas: dependence on protein and particle electrostatics.

We perform all-atom molecular dynamics simulations of a 9 nm-thick protein layer, which consists of serum albumin (SA) or a mixture of SA and immunoglobulin gamma-1, formed on 10 nm-sized cationic, anionic, and neutral polystyrene particles. More than half of the proteins are densely concentrated within a distance of ∼3 nm from the particle surface, while fewer proteins are broadly distributed in the range of 3-9 nm from the particle. This compares favorably with the experimental observations of a hard corona as the first layer adjacent to the particle and a soft corona as a loose protein-network. The conformation and diffusivity of the proteins vary in different positions of the layer, and are to an extent dependent on the protein and particle electrostatics. These, combined with free energy calculations, show that the protein and particle charges do not significantly modify the strength of protein-particle binding but do influence the distribution of proteins in the layer. In particular, a free protein more strongly binds to the complex of a protein and particle than to either one, showing the synergistic effect of already adsorbed proteins and a particle. This helps explain the experimental observation regarding the formation of a denser protein layer and the stronger protein-protein interaction in the hard corona than the soft corona.

Transition-Metal Dichalcogenide Artificial Antibodies with Multivalent Polymeric Recognition Phases for Rapid Detection and Inactivation of Pathogens.

Antibodies are recognition molecules that can bind to diverse targets ranging from pathogens to small analytes with high binding affinity and specificity, making them widely employed for sensing and therapy. However, antibodies have limitations of low stability, long production time, short shelf life, and high cost. Here, we report a facile approach for the design of luminescent artificial antibodies with nonbiological polymeric recognition phases for the sensitive detection, rapid identification, and effective inactivation of pathogenic bacteria. Transition-metal dichalcogenide (TMD) nanosheets with a neutral dextran phase at the interfaces selectively recognized S. aureus, whereas the nanosheets bearing a carboxymethylated dextran phase selectively recognized E. coli O157:H7 with high binding affinity. The bacterial binding sites recognized by the artificial antibodies were thoroughly identified by experiments and molecular dynamics simulations, revealing the significance of their multivalent interactions with the bacterial membrane components for selective recognition. The luminescent WS2 artificial antibodies could rapidly detect the bacteria at a single copy from human serum without any purification and amplification. Moreover, the MoSe2 artificial antibodies selectively killed the pathogenic bacteria in the wounds of infected mice under light irradiation, leading to effective wound healing. This work demonstrates the potential of TMD artificial antibodies as an alternative to antibodies for sensing and therapy.

Effect of Protein Corona on Nanoparticle-Lipid Membrane Binding: The Binding Strength and Dynamics.

All-atom molecular dynamics simulations of the 10 nm-sized anionic polystyrene (PS) particle complexed with plasma proteins (human serum albumin, immunoglobulin gamma-1 chain-C, and apolipoprotein A-I) adsorbed onto lipid bilayers [asymmetrically composed of extracellular (zwitterionic) and cytosolic (anionic) leaflets] are performed. Free energies calculated from umbrella sampling simulations show that proteins on the particle more weakly bind to the zwitterionic leaflet than do bare particles, in agreement with experiments showing the suppression of the particle-bilayer binding by protein corona. Proteins on the particle interact more strongly with the anionic leaflet than with the zwitterionic leaflet because of charge interactions between cationic protein residues and anionic lipid headgroups, to an extent dependent on various plasma proteins. In particular, hydrogen bonds between proteins and zwitterionic leaflets restrict the motion of lipids and thus reduce the lateral dynamics of bilayers, while the tight binding between proteins and anionic leaflets disrupts the helical structure of proteins and disorders lipids, leading to an increase in the lateral dynamics of bilayers. These findings help explain the experimental observation regarding the fact that the bilayer dynamics decreases when interacting with protein corona and suggest that the effect of protein corona on the binding strength and bilayer dynamics depends on protein types and bilayer charges.

A simple strategy for signal enhancement in lateral flow assays using superabsorbent polymers.

To enhance the sensitivity of lateral flow assays (LFAs), a simple strategy is proposed using a nitrocellulose membrane modified with a superabsorbent polymer (SAP). SAP was incorporated into a nitrocellulose membrane for the flow control of detection probes. When absorbing aqueous solutions, SAP promoted the formation of biomolecule complexes to achieve up to a tenfold sensitivity improvement for the detection of human IgG. The assay time was optimized experimentally and numerically to within 20 min using this strategy. Moreover, fluid saturation in LFAs modified with SAP was mathematically simulated to better understand the underlying process, and molecular dynamics simulations were carried out to determine the effect of SAP. The proposed design was also applied to samples spiked with human immunoglobulin-depleted serum to test its applicability. The strategy presented is unique in that it preserves the characteristics of conventional LFAs, as it minimizes user intervention and is simple to manufacture at scale.

Effects of Nanoparticle Electrostatics and Protein-Protein Interactions on Corona Formation: Conformation and Hydrodynamics.

All-atom molecular dynamics simulations of plasma proteins (human serum albumin, fibrinogen, immunoglobulin gamma-1 chain-C, complement C3, and apolipoprotein A-I) adsorbed onto 10 nm sized cationic, anionic, and neutral polystyrene (PS) particles in water are performed. In simulations of a single protein with a PS particle, proteins eventually bind to all PS particles, regardless of particle charge, in agreement with experiments showing the binding between anionic proteins and particles, which is further confirmed by calculating the binding free energies from umbrella sampling simulations. Simulations of mixtures of multiple proteins and a PS particle show the formation of the protein layer on the surface via the adsorption competition between proteins, which influences the binding affinity and structure of adsorbed proteins. In particular, diffusivities are much higher for proteins bound to the particle surface or to the boundary of the protein layer than for those bound to both the particle surface and other proteins, indicating the dependence of protein mobility on their positions in the layer. These findings help to explain in detail experimental observations regarding the replacement of plasma proteins at the early stage of corona formation and the difference in the binding strength of proteins in inner and outer protein-layers.

Effect of lipid shape on toroidal pore formation and peptide orientation in lipid bilayers.

Amphiphilic peptides of different lengths were simulated with lipid bilayers composed of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and 1-myristoyl-2-hydroxy-sn-glycero-3-phosphocholine (lysoMPC) in different ratios. Simulations of lipid bilayers without peptides show that the bilayers with more lysoMPC become more disordered and thinner. Amphiphilic peptides added to this simulation do not insert into the DMPC bilayer at a low peptide/lipid ratio (P/L ≤ 1/50), while they do insert into the DMPC/lysoMPC bilayer and form a toroidal pore even at such a low P/L ratio, where the pore edge is surrounded by lysoMPC rather than by DMPC. In particular, upon pore formation, peptides migrate toward the edge of a pore and become tilted, showing transmembrane alignment regardless of the peptide length, in qualitative agreement with experiments. This pore formation occurs more frequently in larger bilayers that allow greater curvature, indicating that bilayer curvature is important for pore formation. These results indicate that the addition of lysoMPC induces a thinner bilayer with greater curvature, and thus the bilayer with lysoMPC can be more easily penetrated by peptides, leading to the formation of a toroidal pore stabilized by peptides and lysoMPC. These findings help explain experimental observations of the effect of the inverted cone-shaped lyso-lipid on pore formation and peptide orientation, and also support the experimental suggestion regarding the formation of an iris-like ring of helices lining a toroidal pore.

Effects of the asphaltene structure and the tetralin/heptane solvent ratio on the size and shape of asphaltene aggregates.

Asphaltene molecules, which consist of differently hydrogenated polyaromatic cores grafted with side alkyl chains of different sizes and grafting densities, were simulated with a solvent mixture of heptane and tetralin using coarse-grained force fields. Starting with the initial configuration of randomly distributed asphaltene molecules and solvents, the asphaltene molecules aggregate because of the attractive force between their polyaromatic cores, but their sizes and shapes differ. The average aggregate size decreases with an increase in the hydrogenated polycyclic core, side-chain length, and tetralin concentration, which agree with experimental observations in the hydrocracking process. The number of side chains also influences the aggregate size but only in the presence of tetralin. In particular, the effect of tetralin addition occurs more significantly for asphaltene molecules with more side chains because side chains sterically block the intermolecular interactions between polyaromatic cores, which makes it easier for the aromatic ring of tetralin to bind to the polyaromatic core of asphaltene. These steric effects of side chains yield different shapes of aggregates, showing parallel stacking (face-to-face) for aromatic cores with many side chains, and T-shape (edge-to-face) or offset-parallel stacking for those with fewer side chains. These findings agree with the experimental observation regarding the effect of tetralin on the solubility of asphaltene, and indicate that the extent of the tetralin effect depends on the number of side alkyl chains, implying that tetralin solvents need to be added with consideration for the structural change of asphaltene under hydrogenation or dealkylation conditions.

Aggregation and insertion of melittin and its analogue MelP5 into lipid bilayers at different concentrations: effects on pore size, bilayer thickness and dynamics.

Melittin and its analogue MelP5 (five mutations T10A, R22A, K23A, R24Q, and Q26L of melittin) were simulated with lipid bilayers at different peptide/lipid molar ratios using all-atom and coarse-grained (CG) force fields. In CG simulations, both melittin and MelP5 insert into the bilayer at high concentration, while at low concentration only MelP5 can do so, showing the increased membrane permeability of MelP5 because five mutations weaken the electrostatic repulsion between peptides and strengthen the hydrophobic interactions between peptides and lipid tails, in quantitative agreement with experiments. In particular, aggregation of 6-8 MelP5 leads to pore formation, as also suggested by experiments. All-atom simulations, starting with atomic coordinates converted from the final configurations of CG simulations, show that MelP5 peptides bring more water molecules into the pores than do melittin peptides, indicating that MelP5 peptides form larger pores. Also, MelP5 peptides more effectively disorder lipids and thus increase the lateral mobility of lipids than do melittin peptides, leading to thinner bilayers. These findings indicate that differences of only five sequences can influence peptide aggregation and insertion, and bilayer thickness and dynamics, which helps explain experimental observations of the higher extent of antimicrobial activity and macromolecular leakage for MelP5 than for melittin, and also support experimental suggestions regarding the number of aggregated MelP5 and different effects of melittin and MelP5 on pore formation.

Adsorption of Plasma Proteins onto PEGylated Lipid Bilayers: The Effect of PEG Size and Grafting Density.

Lipid bilayers grafted with polyethylene glycol (PEG) of different sizes (Mw = 750, 2000, and 5000) and grafting densities (1.6-25 mol % of PEGylated lipid in dipalmitoylphosphatidylcholine (DPPC) lipid molecules) were simulated with human serum albumin (HSA) using coarse-grained force fields. At low enough grafting density, the PEG has a conformation similar to that of an isolated chain in water, and its Flory radius RF is smaller than the distance between the grafting points (d), which is the so-called "mushroom" regime. In contrast, densely grafted PEG chains (RF > d) extend like brushes, with brush layer thickness given by the Alexander-de Gennes theory. A nearly spherical HSA added to this simulation migrates to the bilayer surface because of the charge interactions between anion residues of HSA and cationic cholines of DPPC, but this HSA-bilayer binding can be sterically suppressed by the PEG chains to an extent that depends on the PEG size and grafting density. In particular, regardless of the extent of the coverage of the PEG on the bilayer, the binding between HSAs and bilayers is suppressed by the PEG layer in a brush but not in a mushroom, indicating that the attractive force between proteins and bilayers can overcome the steric effect of the PEG layer in the mushroom state or in the transition region from mushroom to brush. This helps explain and clarify experiments that show much less adsorption of plasma proteins onto the particle or membrane surface when PEGs are in the brush rather than in the mushroom state.

Effects of temperature, salt concentration, and the protonation state on the dynamics and hydrogen-bond interactions of polyelectrolyte multilayers on lipid membranes.

Polyelectrolyte multilayers, which consist of poly-l-lysines (PLL) and hyaluronic acids (HA), are simulated on phospholipid membranes with explicit water at different temperatures, salt concentrations, and protonation states of PLL that correspond to pH 7 or higher. PLL and HA polymers, which are initially sequentially deposited as three HA/PLL bilayers above the membrane, partially intermix with each other within 300 ns, and with a significant amount of water at almost half of its bulk density. With reduced protonation of amine groups of PLL, the polymers diffuse faster, especially at higher temperatures, and for 0%-protonation, disperse into the water, due to the many fewer hydrogen bonds between PLL and HA polymers. When PLL is protonated, the addition of salt ions weakens electrostatic interactions between PLL and HA and, at 0.5 M NaCl, eventually reduces the number of hydrogen bonds, which in experiments leads to hole formation inside the PLL/HA film. Multilayers are stabilized by hydrogen bonds, primarily between charged groups and to a lesser extent between uncharged groups. PLL and HA also electrostatically interact with lipid head groups of membranes which reduces the lateral mobility of membrane lipids, to an extent dependent on the salt concentration. These findings help quantitate the effects of temperature, salt, and the protonation state (or pH) on the stability and dynamics of multilayers and membranes, and show trends that compare favorably with the experimental observations of the swelling of multilayers.

The binding and insertion of imidazolium-based ionic surfactants into lipid bilayers: the effects of the surfactant size and salt concentration.

Imidazolium-based ionic surfactants with hydrocarbon tails of different sizes were simulated with lipid bilayers at different salt concentrations. Starting with the random position of ionic surfactants outside the bilayer, surfactants with long tails mostly insert into the bilayer, while those with short tails show the insertion of fewer surfactant molecules, indicating the effect of the tail length. In particular, surfactants with a tail of two or four hydrocarbons insert and reversibly detach from the bilayer, while the inserted longer surfactants cannot be reversibly detached because of the strong hydrophobic interaction with lipid tails, in quantitative agreement with experiments. Longer surfactants insert more deeply and irreversibly into the bilayer and thus increase lateral diffusivities of the bilayer, indicating that longer surfactants more significantly disorder lipid bilayers, which also agrees with experiments regarding the effect of the tail length of ionic surfactants on membrane permeability and toxicity. Addition of NaCl ions weakens the electrostatic interactions between headgroups of surfactants and lipids, leading to the binding of fewer surfactants into the bilayer. In particular, our simulation findings indicate that insertion of ionic surfactants can be initiated by either the hydrophobic interaction between tails of surfactants and lipids or the electrostatic binding between imidazolium heads and lipid heads, and the strength of hydrophobic and electrostatic interactions depends on the tail length of surfactants.

The centrosomal protein nephrocystin-6 is mutated in Joubert syndrome and activates transcription factor ATF4.

The molecular basis of nephronophthisis, the most frequent genetic cause of renal failure in children and young adults, and its association with retinal degeneration and cerebellar vermis aplasia in Joubert syndrome are poorly understood. Using positional cloning, we here identify mutations in the gene CEP290 as causing nephronophthisis. It encodes a protein with several domains also present in CENPF, a protein involved in chromosome segregation. CEP290 (also known as NPHP6) interacts with and modulates the activity of ATF4, a transcription factor implicated in cAMP-dependent renal cyst formation. NPHP6 is found at centrosomes and in the nucleus of renal epithelial cells in a cell cycle-dependent manner and in connecting cilia of photoreceptors. Abrogation of its function in zebrafish recapitulates the renal, retinal and cerebellar phenotypes of Joubert syndrome. Our findings help establish the link between centrosome function, tissue architecture and transcriptional control in the pathogenesis of cystic kidney disease, retinal degeneration, and central nervous system development.

Molecular dynamics studies of PEGylated α-helical coiled coils and their self-assembled micelles.

We performed coarse-grained (CG) molecular dynamics simulations of trimeric α-helical coiled coils grafted with poly(ethylene glycol) (PEG) of different sizes and conjugate positions and the self-assembled micelle of amphiphilic trimers. The CG model for the trimeric coiled coil is verified by comparing the α-helical structure and interhelical distance with those calculated from all-atom simulations. In CG simulations of PEGylated trimers, the end-to-end distances and radii of gyration of PEGs grafted to the sides of peptides become shorter than those of free PEGs in water, which agrees with experiments. This shorter size of the grafted PEGs is also confirmed by calculating the thickness of the PEG layer, which is less than the size of the mushroom. These indicate the adsorption of PEG chains onto coiled coils since hydrophobic residues in the exterior sites of coiled coils tend to be less exposed to water and thus interact with PEGs, leading to the compact conformation of adsorbed PEGs. Simulations of the self-assembly of amphiphilic trimers show that the randomly distributed trimers self-assemble to micelles. The outer radius and hydrodynamic radius of the micelle, which were calculated respectively from radial densities and diffusion coefficients, are ∼7 nm, in agreement with the experimental value of ∼7.5 nm, while the aggregation number of amphiphilic molecules per micelle is lower than the experimentally proposed value. These simulations predict the experimentally measured size of PEGs grafted to the trimeric coiled coils and their self-assembled amphiphilic micelles and suggest that the aggregation number of the micelle may be lower, which needs to be confirmed by experiments.

Effect of the structural difference between Bax-α5 and Bcl-xL-α5 on their interactions with lipid bilayers.

Bax-α5 and Bcl-xL-α5, which are shorter versions of apoptosis-regulating proteins Bax and Bcl-xL, were simulated with lipid bilayers composed of pure dioleoylglycerophosphocholine (DOPC) lipids or a mixture of DOPCs and cholesterols. Starting with the initial peptide position near the bilayer surface, both Bax-α5 and Bcl-xL-α5 bind to the bilayer because of their charge interactions with lipid head groups. After binding to the bilayer surface, Bax-α5 inserts into the pure DOPC bilayer, but not into the DOPC-cholesterol bilayer, showing the effect of cholesterols on the peptide-bilayer interaction. Despite the similar peptide structure, Bcl-xL-α5 does not insert into the bilayer, in contrast to the interaction of Bax-α5 with the bilayer. Bcl-xL-α5 predominantly has the random-coil structure in both aqueous and membrane environments, while Bax-α5 shows a higher extent of α-helical structure in the bilayer than in water, in quantitative agreement with experiment. In particular, although Bax-α5 and Bcl-xL-α5 have the same extent of the electrostatic interaction with lipid head groups, Bax-α5 has stronger hydrophobic interaction with lipid tails than does Bcl-xL-α5. These indicate that Bax-α5 retains α-helical structure, where hydrophobic residues on one side of the α-helix interact with lipid tails and thus can easily attract the peptide into the lipid-tail region, while Bcl-xL-α5 forms a random coil that tends to spread on the bilayer surface and thus has weaker hydrophobic interaction with lipid tails. Our findings help explain the experimental observation that showed that Bax-α5 disorders lipids and induces pore formation, but Bcl-xL-α5 does not.

Dynamics and stability of lipid bilayers modulated by thermosensitive polypeptides, cholesterols, and PEGylated lipids.

Lipid bilayers, which consist of dipalmitoylglycerophosphocholines (DPPCs), PEGylated lipids, cholesterols, and elastin-like polypeptides (ELPs; [VPGVG]3) at different molar ratios, were simulated. Simulations were carried out for 2 µs using the coarse-grained (CG) model that had captured the experimentally observed phase behavior of PEGylated lipids and lateral diffusivity of DPPC bilayers. Starting with the initial position of ELPs on the bilayer surface, ELPs insert into the hydrophobic region of the bilayer because of their interaction with lipid tails, consistent with previous all-atom simulations. Lateral diffusion coefficients of DPPCs significantly increase in the bilayer composed of more ELPs and less cholesterols, showing their opposite effects on the bilayer dynamics. In particular, ELPs modulate the dynamics and phase for the disordered liquid bilayer, but not for the ordered gel bilayer, indicating that ELPs can destabilize only the disordered bilayer. In the ordered bilayer, ELP chains tend to have a spherical shape and slowly diffuse, while they are extended and diffuse faster in the disordered bilayer, indicating the effect of the bilayer phase on the conformation and diffusivity of ELPs. These findings explain the experimental observation that the ELP-conjugated liposomes are stable at 310 K (ordered phase) but become unstable and release the encapsulated drugs at 315 K (disordered phase), which suggests the effects of ELPs and cholesterols. Since the cholesterol-stabilized bilayer can be destabilized by the extended shaped ELPs only in the disordered phase (not in the ordered phase), the inclusion of cholesterols is required to safely shield drugs at 310 K as well as allow ELPs to disrupt lipids and destabilize the liposomes at 315 K.

Effects of PEGylation on the binding interaction of magainin 2 and tachyplesin I with lipid bilayer surface.

Poly(ethylene glycol) (PEG)-grafted magainin 2 and tachyplesin I were simulated with lipid bilayers. In the simulations of PEGylated magainin 2 and tachyplesin I in water, both peptides are wrapped by PEG chains. The α-helical structure of PEGylated magainin 2 is broken in water, while the ß-sheet of PEGylated tachyplesin I keeps stable, similar to the structural behavior of unPEGylated peptides, in agreement with experiments. Simulations of PEGylated peptides with lipid bilayers show that PEG chains block the electrostatic interaction between cationic residues of peptides and anionic phosphates of lipids, leading to the less binding of the peptide to the bilayer surface, which is observed more significantly for magainin 2 than for tachyplesin I. Since the random-coiled magainin 2 can be more completely covered by PEGs than does the ß-sheet tachyplesin I, the PEGylation effect on the decreased binding is larger for magainin 2, showing the dependence of PEGylation on the peptide structure. These simulation findings qualitatively support the experimental observation of the different extents of the reduced membrane-permeabilizing activity for PEGylated magainin 2 and tachyplesin I.

Membrane penetration and curvature induced by single-walled carbon nanotubes: the effect of diameter, length, and concentration.

We performed coarse-grained (CG) molecular dynamics (MD) simulations of single-walled carbon nanotubes (SWNTs) with lipid bilayers to understand the effect of the SWNT diameter, length, and concentration on membrane curvature and penetration. Starting with different orientations of multiple SWNTs near lipid bilayers, simulations show that SWNTs insert into the bilayer and induce membrane curvature, which is much larger than that observed from previous simulations of a single SWNT. Longer and thicker SWNTs at higher concentration cause larger membrane curvature, indicating the effect of the SWNT size and concentration, in qualitative agreement with experiments. In particular, thicker SWNTs significantly increase the bilayer height and the difference of the projected and contour bilayer areas, decrease the area compressibility, and disorder lipids. When inserted into the bilayer, thinner SWNTs mainly contact the entire tails of lipids, while thicker SWNTs are wrapped mainly by the ending tail-carbons, leading to the larger membrane curvature. This indicates the effect of SWNT diameter on the SWNT-lipid interaction, yielding different extents of membrane curvature. These findings imply that the SWNT-induced membrane penetration and curvature are modulated by a combination of SWNT length, diameter, and concentration.

Isolating a trimer intermediate in the self-assembly of E2 protein cage.

Understanding the self-assembly mechanism of caged proteins provides clues to develop their potential applications in nanotechnology, such as a nanoscale drug delivery system. The E2 protein from Bacillus stearothermophilus , with a virus-like caged structure, has drawn much attention for its potential application as a nanocapsule. To investigate its self-assembly process from subunits to a spherical protein cage, we truncate the C-terminus of the E2 subunit. The redesigned protein subunit shows dynamic transition between monomer and trimer, but not the integrate 60-mer. The results indicate the role of the trimer as the intermediate and building block during the self-assembly of the E2 protein cage. In combination with the molecular dynamics simulations results, we conclude that the C-terminus modulates the self-assembly of the E2 protein cage from trimer to 60-mer. This investigation elucidates the role of the intersubunit interactions in engineering other functionalities in other caged structure proteins.

Effects of Lipid Shape and Interactions on the Conformation, Dynamics, and Curvature of Ultrasound-Responsive Liposomes.

We perform coarse-grained molecular dynamics simulations of bilayers composed of various lipids and cholesterol at their different ratios. Simulations show that cholesterol-lipid interactions restrict the lateral dynamics of bilayers but also promote bilayer curvature, indicating that these opposite effects simultaneously occur and thus cannot significantly influence bilayer stability. In contrast, lyso-lipids effectively pack the vacancy in the bilayer composed of cone-shaped lipids and thus reduce bilayer dynamics and curvature, showing that bilayers are more significantly stabilized by lyso-lipids than by cholesterol, in agreement with experiments. In particular, the bilayer composed of cone-shaped lipids shows higher dynamics and curvature than does the bilayer composed of cylindrical-shaped lipids. To mimic ultrasound, a high external pressure was applied in the direction of bilayer normal, showing the formation of small pores that are surrounded by hydrophilic lipid headgroups, which can allow the release of drug molecules encapsulated into the liposome. These findings help to explain experimental observations regarding that liposomes are more significantly stabilized by lyso-lipids than by cholesterol, and that the liposome with cone-shaped lipids more effectively releases drug molecules upon applying ultrasound than does the liposome with cylindrical-shaped lipids.