Incorporating physics to overcome data scarcity in predictive modeling of protein function: A case study of BK channels.

Machine learning has played transformative roles in numerous chemical and biophysical problems such as protein folding where large amount of data exists. Nonetheless, many important problems remain challenging for data-driven machine learning approaches due to the limitation of data scarcity. One approach to overcome data scarcity is to incorporate physical principles such as through molecular modeling and simulation. Here, we focus on the big potassium (BK) channels that play important roles in cardiovascular and neural systems. Many mutants of BK channel are associated with various neurological and cardiovascular diseases, but the molecular effects are unknown. The voltage gating properties of BK channels have been characterized for 473 site-specific mutations experimentally over the last three decades; yet, these functional data by themselves remain far too sparse to derive a predictive model of BK channel voltage gating. Using physics-based modeling, we quantify the energetic effects of all single mutations on both open and closed states of the channel. Together with dynamic properties derived from atomistic simulations, these physical descriptors allow the training of random forest models that could reproduce unseen experimentally measured shifts in gating voltage, ∆V1/2, with a RMSE ~ 32 mV and correlation coefficient of R ~ 0.7. Importantly, the model appears capable of uncovering nontrivial physical principles underlying the gating of the channel, including a central role of hydrophobic gating. The model was further evaluated using four novel mutations of L235 and V236 on the S5 helix, mutations of which are predicted to have opposing effects on V1/2 and suggest a key role of S5 in mediating voltage sensor-pore coupling. The measured ∆V1/2 agree quantitatively with prediction for all four mutations, with a high correlation of R = 0.92 and RMSE = 18 mV. Therefore, the model can capture nontrivial voltage gating properties in regions where few mutations are known. The success of predictive modeling of BK voltage gating demonstrates the potential of combining physics and statistical learning for overcoming data scarcity in nontrivial protein function prediction.

Incorporating physics to overcome data scarcity in predictive modeling of protein function: a case study of BK channels.

Machine learning has played transformative roles in numerous chemical and biophysical problems such as protein folding where large amount of data exists. Nonetheless, many important problems remain challenging for data-driven machine learning approaches due to the limitation of data scarcity. One approach to overcome data scarcity is to incorporate physical principles such as through molecular modeling and simulation. Here, we focus on the big potassium (BK) channels that play important roles in cardiovascular and neural systems. Many mutants of BK channel are associated with various neurological and cardiovascular diseases, but the molecular effects are unknown. The voltage gating properties of BK channels have been characterized for 473 site-specific mutations experimentally over the last three decades; yet, these functional data by themselves remain far too sparse to derive a predictive model of BK channel voltage gating. Using physics-based modeling, we quantify the energetic effects of all single mutations on both open and closed states of the channel. Together with dynamic properties derived from atomistic simulations, these physical descriptors allow the training of random forest models that could reproduce unseen experimentally measured shifts in gating voltage, ΔV 1/2 , with a RMSE ∼ 32 mV and correlation coefficient of R ∼ 0.7. Importantly, the model appears capable of uncovering nontrivial physical principles underlying the gating of the channel, including a central role of hydrophobic gating. The model was further evaluated using four novel mutations of L235 and V236 on the S5 helix, mutations of which are predicted to have opposing effects on V 1/2 and suggest a key role of S5 in mediating voltage sensor-pore coupling. The measured ΔV 1/2 agree quantitatively with prediction for all four mutations, with a high correlation of R = 0.92 and RMSE = 18 mV. Therefore, the model can capture nontrivial voltage gating properties in regions where few mutations are known. The success of predictive modeling of BK voltage gating demonstrates the potential of combining physics and statistical learning for overcoming data scarcity in nontrivial protein function prediction. Author Summary: Deep machine learning has brought many exciting breakthroughs in chemistry, physics and biology. These models require large amount of training data and struggle when the data is scarce. The latter is true for predictive modeling of the function of complex proteins such as ion channels, where only hundreds of mutational data may be available. Using the big potassium (BK) channel as a biologically important model system, we demonstrate that a reliable predictive model of its voltage gating property could be derived from only 473 mutational data by incorporating physics-derived features, which include dynamic properties from molecular dynamics simulations and energetic quantities from Rosetta mutation calculations. We show that the final random forest model captures key trends and hotspots in mutational effects of BK voltage gating, such as the important role of pore hydrophobicity. A particularly curious prediction is that mutations of two adjacent residues on the S5 helix would always have opposite effects on the gating voltage, which was confirmed by experimental characterization of four novel mutations. The current work demonstrates the importance and effectiveness of incorporating physics in predictive modeling of protein function with scarce data.

Inner pore hydration free energy controls the activation of big potassium channels.

Hydrophobic gating is an emerging mechanism in regulation of protein ion channels where the pore remains physically open but becomes dewetted to block ion permeation. Atomistic molecular dynamics simulations have played a crucial role in understanding hydrophobic gating by providing the molecular details to complement mutagenesis and structural studies. However, existing studies rely on direct simulations and do not quantitatively describe how the sequence and structural changes may control the delicate liquid-vapor equilibrium of confined water in the pore of the channel protein. To address this limitation, we explore two enhanced sampling methods, namely metadynamics and umbrella sampling, to derive free-energy profiles of pore hydration in both the closed and open states of big potassium (BK) channels, which are important in cardiovascular and neural systems. It was found that metadynamics required substantially longer sampling times and struggled to generate stably converged free-energy profiles due to the slow dynamics of cooperative pore water diffusion even in the barrierless limit. Using umbrella sampling, well-converged free-energy profiles can be readily generated for the wild-type BK channels as well as three mutants with pore-lining mutations experimentally known to dramatically perturb the channel gating voltage. The results show that the free energy of pore hydration faithfully reports the gating voltage of the channel, providing further support for hydrophobic gating in BK channels. Free-energy analysis of pore hydration should provide a powerful approach for quantitative studies of how protein sequence, structure, solution conditions, and/or drug binding may modulate hydrophobic gating in ion channels.

An allosteric modulator activates BK channels by perturbing coupling between Ca2+ binding and pore opening.

BK type Ca2+-activated K+ channels activate in response to both voltage and Ca2+. The membrane-spanning voltage sensor domain (VSD) activation and Ca2+ binding to the cytosolic tail domain (CTD) open the pore across the membrane, but the mechanisms that couple VSD activation and Ca2+ binding to pore opening  are not clear. Here we show that a compound, BC5, identified from in silico screening, interacts with the CTD-VSD interface and specifically modulates the Ca2+ dependent activation mechanism. BC5 activates the channel in the absence of Ca2+ binding but Ca2+ binding inhibits BC5 effects. Thus, BC5 perturbs a pathway that couples Ca2+ binding to pore opening to allosterically affect both, which is further supported by atomistic simulations and mutagenesis. The results suggest that the CTD-VSD interaction makes a major contribution to the mechanism of Ca2+ dependent activation and is an important site for allosteric agonists to modulate BK channel activation.

Activation of TMEM16F by inner gate charged mutations and possible lipid/ion permeation mechanisms.

Transmembrane protein 16F (TMEM16F) is a ubiquitously expressed Ca2+-activated phospholipid scramblase that also functions as a largely non-selective ion channel. Though recent structural studies have revealed the closed and intermediate conformations of mammalian TMEM16F (mTMEM16F), the open and conductive state remains elusive. Instead, it has been proposed that an open hydrophilic pathway may not be required for lipid scrambling. We previously identified an inner activation gate, consisting of F518, Y563, and I612, and showed that charged mutations of the inner gate residues led to constitutively active mTMEM16F scrambling. Herein, atomistic simulations show that lysine substitution of F518 and Y563 can indeed lead to spontaneous opening of the permeation pore in the Ca2+-bound state of mTMEM16F. Dilation of the pore exposes hydrophilic patches in the upper pore region, greatly increases the pore hydration level, and enables lipid scrambling. The putative open state of mTMEM16F resembles the active state of fungal scramblases and is a meta-stable state for the wild-type protein in the Ca2+-bound state. Therefore, mTMEM16F may be capable of supporting the canonical in-groove scrambling mechanism in addition to the out-of-groove one. Further analysis reveals that the in-groove phospholipid and ion transduction pathways of mTMEM16F overlap from the intracellular side up to the inner gate but diverge from each other with different exits to the extracellular side of membrane.

Specific PIP2 binding promotes calcium activation of TMEM16A chloride channels.

TMEM16A is a widely expressed Ca2+-activated Cl- channel that regulates crucial physiological functions including fluid secretion, neuronal excitability, and smooth muscle contraction. There is a critical need to understand the molecular mechanisms of TMEM16A gating and regulation. However, high-resolution TMEM16A structures have failed to reveal an activated state with an unobstructed permeation pathway even with saturating Ca2+. This has been attributed to the requirement of PIP2 for preventing TMEM16A desensitization. Here, atomistic simulations show that specific binding of PIP2 to TMEM16A can lead to spontaneous opening of the permeation pathway in the Ca2+-bound state. The predicted activated state is highly consistent with a wide range of mutagenesis and functional data. It yields a maximal Cl- conductance of ~1 pS, similar to experimental estimates, and recapitulates the selectivity of larger SCN- over Cl-. The resulting molecular mechanism of activation provides a basis for understanding the interplay of multiple signals in controlling TMEM16A channel function.

Hydrophobic dewetting in gating and regulation of transmembrane protein ion channels.

Water is at the heart of almost all biological phenomena, without which no life that we know of would have been possible. It is a misleadingly complex liquid that exists in near coexistence with the vapor phase under ambient conditions. Confinement within a hydrophobic cavity can tip this balance enough to drive a cooperative dewetting transition. For a nanometer-scale pore, the dewetting transition leads to a stable dry state that is physically open but impermeable to ions. This phenomenon is often referred to as hydrophobic gating. Numerous transmembrane protein ion channels have now been observed to utilize hydrophobic gating in their activation and regulation. Here, we review recent theoretical, simulation, and experimental studies that together have started to establish the principles of hydrophobic gating and discuss how channels of various sizes, topologies, and biological functions can utilize these principles to control the thermodynamic properties of water within their interior pores for gating and regulation. Exciting opportunities remain in multiple areas, particularly on direct experimental detection of hydrophobic dewetting in biological channels and on understanding how the cell may control the hydrophobic gating in regulation of ion channels.

Publisher Correction: Hydrophobic gating in BK channels.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Aromatic interactions with membrane modulate human BK channel activation.

Large-conductance potassium (BK) channels are transmembrane (TM) proteins that can be synergistically and independently activated by membrane voltage and intracellular Ca2+. The only covalent connection between the cytosolic Ca2+ sensing domain and the TM pore and voltage sensing domains is a 15-residue 'C-linker'. To determine the linker's role in human BK activation, we designed a series of linker sequence scrambling mutants to suppress potential complex interplay of specific interactions with the rest of the protein. The results revealed a surprising sensitivity of BK activation to the linker sequence. Combining atomistic simulations and further mutagenesis experiments, we demonstrated that nonspecific interactions of the linker with membrane alone could directly modulate BK activation. The C-linker thus plays more direct roles in mediating allosteric coupling between BK domains than previously assumed. Our results suggest that covalent linkers could directly modulate TM protein function and should be considered an integral component of the sensing apparatus.

Amyloid assembly is dominated by misregistered kinetic traps on an unbiased energy landscape.

Atomistic description of protein fibril formation has been elusive due to the complexity and long time scales of the conformational search. Here, we develop a multiscale approach combining numerous atomistic simulations in explicit solvent to construct Markov State Models (MSMs) of fibril growth. The search for the in-register fully bound fibril state is modeled as a random walk on a rugged two-dimensional energy landscape defined by ß-sheet alignment and hydrogen-bonding states, whereas transitions involving states without hydrogen bonds are derived from kinetic clustering. The reversible association/dissociation of an incoming peptide and overall growth kinetics are then computed from MSM simulations. This approach is applied to derive a parameter-free, comprehensive description of fibril elongation of Aß16-22 and how it is modulated by phenylalanine-to-cyclohexylalanine (CHA) mutations. The trajectories show an aggregation mechanism in which the peptide spends most of its time trapped in misregistered ß-sheet states connected by weakly bound states twith short lifetimes. Our results recapitulate the experimental observation that mutants CHA19 and CHA1920 accelerate fibril elongation but have a relatively minor effect on the critical concentration for fibril growth. Importantly, the kinetic consequences of mutations arise from cumulative effects of perturbing the network of productive and nonproductive pathways of fibril growth. This is consistent with the expectation that nonfunctional states will not have evolved efficient folding pathways and, therefore, will require a random search of configuration space. This study highlights the importance of describing the complete energy landscape when studying the elongation mechanism and kinetics of protein fibrils.

Molecular basis of PIP2-dependent regulation of the Ca2+-activated chloride channel TMEM16A.

The calcium-activated chloride channel (CaCC) TMEM16A plays crucial roles in regulating neuronal excitability, smooth muscle contraction, fluid secretion and gut motility. While opening of TMEM16A requires binding of intracellular Ca2+, prolonged Ca2+-dependent activation results in channel desensitization or rundown, the mechanism of which is unclear. Here we show that phosphatidylinositol (4,5)-bisphosphate (PIP2) regulates TMEM16A channel activation and desensitization via binding to a putative binding site at the cytosolic interface of transmembrane segments (TMs) 3-5. We further demonstrate that the ion-conducting pore of TMEM16A is constituted of two functionally distinct modules: a Ca2+-binding module formed by TMs 6-8 and a PIP2-binding regulatory module formed by TMs 3-5, which mediate channel activation and desensitization, respectively. PIP2 dissociation from the regulatory module results in ion-conducting pore collapse and subsequent channel desensitization. Our findings thus provide key insights into the mechanistic understanding of TMEM16 channel gating and lipid-dependent regulation.

An inner activation gate controls TMEM16F phospholipid scrambling.

Transmembrane protein 16F (TMEM16F) is an enigmatic Ca2+-activated phospholipid scramblase (CaPLSase) that passively transports phospholipids down their chemical gradients and mediates blood coagulation, bone development and viral infection. Despite recent advances in the structure and function understanding of TMEM16 proteins, how mammalian TMEM16 CaPLSases open and close, or gate their phospholipid permeation pathways remains unclear. Here we identify an inner activation gate, which is established by three hydrophobic residues, F518, Y563 and I612, in the middle of the phospholipid permeation pathway of TMEM16F-CaPLSase. Disrupting the inner gate profoundly alters TMEM16F phospholipid permeation. Lysine substitutions of F518 and Y563 even lead to constitutively active CaPLSases that bypass Ca2+-dependent activation. Strikingly, an analogous lysine mutation to TMEM16F-F518 in TMEM16A (L543K) is sufficient to confer CaPLSase activity to the Ca2+-activated Cl- channel (CaCC). The identification of an inner activation gate can help elucidate the gating and permeation mechanism of TMEM16 CaPLSases and channels.

Covalent labeling and mass spectrometry reveal subtle higher order structural changes for antibody therapeutics.

Monoclonal antibodies are among the fastest growing therapeutics in the pharmaceutical industry. Detecting higher-order structure changes of antibodies upon storage or mishandling, however, is a challenging problem. In this study, we describe the use of diethylpyrocarbonate (DEPC)-based covalent labeling (CL) - mass spectrometry (MS) to detect conformational changes caused by heat stress, using rituximab as a model system. The structural resolution obtained from DEPC CL-MS is high enough to probe subtle conformation changes that are not detectable by common biophysical techniques. Results demonstrate that DEPC CL-MS can detect and identify sites of conformational changes at the temperatures below the antibody melting temperature (e.g., 55 á´¼C). The observed labeling changes at lower temperatures are validated by activity assays that indicate changes in the Fab region. At higher temperatures (e.g., 65 á´¼C), conformational changes and aggregation sites are identified from changes in CL levels, and these results are confirmed by complementary biophysical and activity measurements. Given the sensitivity and simplicity of DEPC CL-MS, this method should be amenable to the structural investigations of other antibody therapeutics.

Hydrophobic gating in BK channels.

The gating mechanism of transmembrane ion channels is crucial for understanding how these proteins control ion flow across membranes in various physiological processes. Big potassium (BK) channels are particularly interesting with large single-channel conductance and dual regulation by membrane voltage and intracellular Ca2+. Recent atomistic structures of BK channels failed to identify structural features that could physically block the ion flow in the closed state. Here, we show that gating of BK channels does not seem to require a physical gate. Instead, changes in the pore shape and surface hydrophobicity in the Ca2+-free state allow the channel to readily undergo hydrophobic dewetting transitions, giving rise to a large free energy barrier for K+ permeation. Importantly, the dry pore remains physically open and is readily accessible to quaternary ammonium channel blockers. The hydrophobic gating mechanism is also consistent with scanning mutagenesis studies showing that modulation of pore hydrophobicity is correlated with activation properties.

Developments in Glycopeptide Antibiotics.

Glycopeptide antibiotics (GPAs) are a key weapon in the fight against drug resistant bacteria, with vancomycin still a mainstream therapy against serious Gram-positive infections more than 50 years after it was first introduced. New, more potent semisynthetic derivatives that have entered the clinic, such as dalbavancin and oritavancin, have superior pharmacokinetic and target engagement profiles that enable successful treatment of vancomycin-resistant infections. In the face of resistance development, with multidrug resistant (MDR) S. pneumoniae and methicillin-resistant Staphylococcus aureus (MRSA) together causing 20-fold more infections than all MDR Gram-negative infections combined, further improvements are desirable to ensure the Gram-positive armamentarium is adequately maintained for future generations. A range of modified glycopeptides has been generated in the past decade via total syntheses, semisynthetic modifications of natural products, or biological engineering. Several of these have undergone extensive characterization with demonstrated in vivo efficacy, good PK/PD profiles, and no reported preclinical toxicity; some may be suitable for formal preclinical development. The natural product monobactam, cephalosporin, and ß-lactam antibiotics all spawned multiple generations of commercially and clinically successful semisynthetic derivatives. Similarly, next-generation glycopeptides are now technically well positioned to advance to the clinic, if sufficient funding and market support returns to antibiotic development.

Enhanced Sampling of Intrinsic Structural Heterogeneity of the BH3-Only Protein Binding Interface of Bcl-xL.

Antiapoptotic Bcl-xL plays central roles in regulating programed cell death. Partial unfolding of Bcl-xL has been observed at the interface upon specific binding to the pro-apoptotic BH3-only protein PUMA, which in turn disrupts the interaction of Bcl-xL with tumor suppressor p53 and promotes apoptosis. Previous analysis of existing Bcl-xL structures and atomistic molecular dynamics (MD) simulations have suggested that substantial intrinsic structure heterogeneity exists at the BH3-only protein binding interface of Bcl-xL to facilitate its conformational transitions upon binding. In this study, enhanced sampling is applied to further characterize the interfacial conformations of unbound Bcl-xL in explicit solvent. Extensive replica exchange with solute tempering (REST) simulations, with a total accumulated time of 16 µs, were able to cover much wider conformational spaces for the interfacial region of Bcl-xL. The resulting structural ensembles are much better converged, with local and long-range structural features that are highly consistent with existing NMR data. These simulations further demonstrate that the BH3-only protein binding interface of Bcl-xL is intrinsically disordered and samples many rapidly interconverting conformations. Intriguingly, all previously observed conformers are well represented in the unbound structure ensemble. Such intrinsic structural heterogeneity and flexibility may be critical for Bcl-xL to undergo partial unfolding induced by PUMA binding, and likely provide a robust basis that allows Bcl-xL to respond sensitively to binding of various ligands in cellular signaling and regulation.

The Levinthal Problem in Amyloid Aggregation: Sampling of a Flat Reaction Space.

The formation of amyloid fibrils has been associated with many neurodegenerative disorders, yet the mechanism of aggregation remains elusive, partly because aggregation time scales are too long to probe with atomistic simulations. A microscopic theory of fibril elongation was recently developed that could recapitulate experimental results with respect to the effects of temperature, denaturants, and protein concentration on fibril growth kinetics (Schmit, J. D. J. Chem. Phys. 2013, 138 (18), 185102). The theory identifies the conformational search over H-bonding states as the slowest step in the aggregation process and suggests that this search can be efficiently modeled as a random walk on a rugged one-dimensional energy landscape. This insight motivated the multiscale computational algorithm for simulating fibril growth presented in this paper. Briefly, a large number of short atomistic simulations are performed to compute the system diffusion tensor in the reaction coordinate space predicted by the analytic theory. Ensemble aggregation pathways and growth kinetics are then computed from Markov state model (MSM) trajectories. The algorithm is deployed here to understand the fibril growth mechanism and kinetics of Aß16-22 and three of its mutants. The order of growth rates of the wild-type and two single mutation peptides (CHA19 and CHA20) predicted by the MSM trajectories is consistent with experimental results. The simulation also correctly predicts that the double mutation (CHA19/CHA20) would reduce the fibril growth rate, even though the degree of rate reduction with respect to either single mutation is overestimated. This artifact may be attributed to the simplistic implicit solvent model. These trends in the growth rate are not apparent from inspection of the rate constants of individual bonds or the lifetimes of the mis-registered states that are the primary kinetic traps but only emerge in the ensemble of trajectories generated by the MSM.

Organization and Structure of Branched Amphipathic Oligopeptide Bilayers.

A class of self-assembling branched amphiphilic peptide capsules (BAPCs) was recently developed that could serve as a new drug delivery vehicle. BAPCs can encapsulate solutes up to ∼12 kDa during assembly, are unusually stable, and are readily taken up by cells with low cytotoxicity. Coarse-grained simulations have supported that BAPCs are defined by bilayers that resemble those formed by diacyl phospholipids. Here, atomistic simulations were performed to characterize the structure and organization of bilayers formed by three branched amphiphilic peptides (BAPs): bis(Ac-FLIVIGSII)-K-K4-CO-NH2, bis(Ac-CHA-LIVIGSII)-K-K4-CO-NH2, and bis(Ac-FLIVI)-K-K4-CO-NH2. The results show BAPs form a network of intra- and intermolecular backbone hydrogen bonds within the same leaflet in addition to hydrophobic side-chain interactions. The terminal residues of two leaflets form an interdigitation region locking two leaflets together. The phenyl groups in bis(Ac-FLIVIGSII)-K-K4-CO-NH2 and bis(Ac-FLIVI)-K-K4-CO-NH2 are tightly packed near the bilayer center but do not formed ordered structures with specific π-π stacking. Replacing phenyl groups with the cyclohexane side chain only slightly increases the level of disorder in bilayer structures and thus should not significantly affect the stability, consistent with experimental results on bis(Ac-CHA-LIVIGSII)-K-K4-CO-NH2 BAPCs. Self-assembly simulations further suggest that leaflet interdigitation likely occurs at early stages of BAPC formation. Atomistic simulations also reveal that the BAPC bilayers are highly permeable to water. This prediction was validated using fluorescence measurements of encapsulated self-quenching dye upon transferring BAPCs to buffers with different salt concentrations. Improved understanding of the organization and structure of BAPC bilayers at the atomic level will provide a basis for future rational modifications of BAP sequence to improve BAPC properties as a new class of delivery vehicle.

Necessity of high-resolution for coarse-grained modeling of flexible proteins.

The popular MARTINI coarse-grained (CG) force field requires the protein structure to be fixed, and is unsuitable for simulating dynamic processes such as protein folding. Here, we examine the feasibility of developing a flexible protein model within the MARTINI framework. The results demonstrate that the MARTINI CG scheme does not properly describe the volume and packing of protein backbone and side chains and leads to excessive collapse without structural restraints in explicit CG water. Combining atomistic protein representation with the MARTINI CG solvent, such as in the PACE model, dramatically improves description of flexible protein conformations. Yet, the CG solvent is insufficient to capture the conformational dependence of protein-solvent interactions, and PACE is unable to properly model context dependent conformational transitions. Taken together, high physical resolution at or near the atomistic level is likely necessary for flexible protein models with explicit, microscopic solvents, and the coarse-graining needs to focus on possible simplification in interaction potentials. © 2016 Wiley Periodicals, Inc.

The recognition of membrane-bound PtdIns3P by PX domains.

Phox-homology (PX) domains target proteins to the organelles of the secretary and endocytic systems by binding to phosphatidylinositol phospholipids (PIPs). Among all the structures of PX domains that have been solved, only three have been solved in a complex with the main physiological ligand: PtdIns3P. In this work, molecular dynamic simulations have been used to explore the structure and dynamics of the p40(phox) -PX domain and the SNX17-PX domain and their interaction with membrane-bound PtdIns3P. In the simulations, both PX domains associated spontaneously with the membrane-bound PtdIns3P and formed stable complexes. The interaction between the p40(phox) -PX domain and PtdIns3P in the membrane was found to be similar to the crystal structure of the p40(phox) -PX-PtdIns3P complex that is available. The interaction between the SNX17-PX domain and PtdIns3P was similar to that observed in the p40(phox) -PX-PtdIns3P complex; however, some residues adopted different orientations. The simulations also showed that nonspecific interactions between the ß1-ß2 loop and the membrane play an important role in the interaction of membrane bound PtdIns3P and different PX domains. The behaviour of unbound PtdIns3P within a 2-oleoyl-1-palmitoyl-sn-glycero-3-phosphocholine (POPC) membrane environment was also examined and compared to the available experimental data and simulation studies of related molecules.