Evaluation of transmembrane helix predictions in 2014

Experimental structure determination continues to be challenging for membrane proteins. Computational prediction methods are therefore needed and widely used to supplement experimental data. Here, we re‐examined the state of the art in transmembrane helix prediction based on a nonredundant dataset with 190 high‐resolution structures. Analyzing 12 widely‐used and well‐known methods using a stringent performance measure, we largely confirmed the expected high level of performance. On the other hand, all methods performed worse for proteins that could not have been used for development. A few results stood out: First, all methods predicted proteins in eukaryotes better than those in bacteria. Second, methods worked less well for proteins with many transmembrane helices. Third, most methods correctly discriminated between soluble and transmembrane proteins. However, several older methods often mistook signal peptides for transmembrane helices. Some newer methods have overcome this shortcoming. In our hands, PolyPhobius and MEMSAT‐SVM outperformed other methods. Proteins 2015; 83:473–484. © 2014 Wiley Periodicals, Inc.     

Distinct protein interfaces in transmembrane domains suggest an in vivo folding model

Given the known high-resolution structures of alpha-helical transmembrane domains, we show that there are statistically distinct classes of transmembrane interfaces which relate to the folding and oligomerization of transmembrane domains. Distinct types of interfaces have been categorized and refer to those between: the same polypeptide chain, different polypeptide chains, helices that are sequential neighbors, and those that are nonsequential. These different interfaces may reflect different phases in the mechanism of transmembrane domain folding and are consistent with the current experimental evidence pertaining to the folding and oligomerization of transmembrane domains. The classes of helix-helix interfaces have been identified in terms of the numbers and different types of pairwise amino acid interactions. The specific measures used are interaction entropy, the information content of interacting partners compared to a random set of contacts, the amino acid composition of the classes and the abundances of specific amino acid pairs in close contact. Knowledge of the clear differences in the types of helix-helix contacts helps with the derivation of knowledge-based constraints which until now have focused on only the interiors of transmembrane domains as compared to the exterior. Taken together, an in vivo model for membrane protein folding is presented, which is distinct from the familiar two-stage model. The model takes into account the different interfaces of membrane helices defined herein, and the available data regarding folding in the translocation channel.     

Bicelle size modulates the rate of bacteriorhodopsin folding

The conformational equilibria of integral membrane proteins have proven extremely difficult to characterize within native lipid bilayers. To circumvent technical issues, investigations of the structure and stability of α‐helical membrane proteins are often carried out in mixed micelle or bicelle solvents that mimic the membrane and facilitate measurements of reversible folding. Under these conditions, the energetics of membrane protein folding are typically proportional to the mole fraction of an anionic detergent in the micelle. However, investigations of the folding and unfolding of bacteriorhodopsin (bR) surprisingly revealed that the folding rate is also highly sensitive to the bulk molar concentration of lipids and detergents. We show here that this rate enhancement coincides with changes in bicelle size and suggest this effect arises through restriction of the conformational search space during folding. In conjunction with previous mutagenic studies, these results provide additional evidence that a topological search limits the rate of bR folding. Furthermore, this finding provides insights into the manner by which micellar and bicellar environments influence the conformational stability of polytopic membrane proteins.     

Revisiting the folding kinetics of bacteriorhodopsin

The elucidation of the physical principles that govern the folding and stability of membrane proteins is one of the greatest challenges in protein science. Several insights into the folding of α-helical membrane proteins have come from the investigation of the conformational equilibrium of H. halobium bacteriorhodopsin (bR) in mixed micelles using SDS as a denaturant. In an effort to confirm that folded bR and SDS-denatured bR reach the same conformational equilibrium, we found that bR folding is significantly slower than has been previously known. Interrogation of the effect of the experimental variables on folding kinetics reveals that the rate of folding is dependent not only on the mole fraction of SDS but also on the molar concentrations of mixed micelle components, a variable that was not controlled in the previous study of bR folding kinetics. Moreover, when the molar concentrations of mixed micelle components are fixed at the concentrations commonly employed for bR equilibrium studies, conformational relaxation in the transition zone is slower than hydrolysis of the retinal Schiff base. As a result, the conformational equilibrium between folded bR and SDS-denatured bR cannot be achieved under the conventional condition. Our finding suggests that the molar concentrations of mixed micelle components are important experimental variables in the investigation of the kinetics and thermodynamics of bR folding and should be accounted for to ensure the accurate assessment of the conformational equilibrium of bR without the interference of retinal hydrolysis.     

The activation energy for insertion of transmembrane alpha-helices is dependent on membrane composition

The physical mechanisms that govern the folding and assembly of integral membrane proteins are poorly understood. It appears that certain properties of the lipid bilayer affect membrane protein folding in vitro, either by modulating helix insertion or packing. In order to begin to understand the origin of this effect, we investigate the effect of lipid forces on the insertion of a transmembrane alpha-helix using a water-soluble, alanine-based peptide, KKAAAIAAAAAIAAWAAIAAAKKKK-amide. This peptide binds to preformed 1,2-dioleoyl-l-alpha-phosphatidylcholine (DOPC) vesicles at neutral pH, but spontaneous transmembrane helix insertion directly from the aqueous phase only occurs at high pH when the Lys residues are de-protonated. These results suggest that the translocation of charge is a major determinant of the activation energy for insertion. Time-resolved measurements of the insertion process at high pH indicate biphasic kinetics with time constants of ca 30 and 430 seconds. The slower phase seems to correlate with formation of a predominantly transmembrane alpha-helical conformation, as determined from the transfer of the tryptophan residue to the hydrocarbon region of the membrane. Temperature-dependent measurements showed that insertion can proceed only above a certain threshold temperature and that the Arrhenius activation energy is of the order of 90 kJ mol(-1). The kinetics, threshold temperature and the activation energy change with the mole fraction of 1,2-dioleoyl-l-alpha-phosphatidylethanolamine (DOPE) introduced into the DOPC membrane. The activation energy increases with increasing DOPE content, which could reflect the fact that this lipid drives the bilayer towards a non-bilayer transition and increases the lateral pressure in the lipid chain region. This suggests that folding events involving the insertion of helical segments across the bilayer can be controlled by lipid forces.     

The stability of transmembrane helix interactions measured in a biological membrane

Despite some promising progress in the understanding of membrane protein folding and assembly, there is little experimental information regarding the thermodynamic stability of transmembrane helix interactions and even less on the stability of transmembrane helix-helix interactions in a biological membrane. Here we describe an approach that allows quantitative measurement of transmembrane helix interactions in a biological membrane, and calculation of changes in the interaction free energy resulting from substitution of single amino acids. Dimerization of several variants of the glycophorin A transmembrane domain are characterized and compared to the wild-type (wt) glycophorin A transmembrane helix dimerization. The calculated DeltaDeltaG(app) values are further compared with values found in the literature. In addition, we compare interactions between the wt glycophorin A transmembrane domain and helices in which critical glycine residues are replaced by alanine or serine, respectively. The data demonstrate that replacement of the glycine residues by serine is less destabilizing than replacement by alanine with a DeltaDeltaG(app) value of about 0.4 kcal/mol. Our study comprises the first measurement of a transmembrane helix interaction in a biological membrane, and we are optimistic that it can be further developed and applied.     

Influence of proline residues in transmembrane helix packing

Integral membrane proteins often contain proline residues in their alpha-helical transmembrane (TM) fragments, which may strongly influence their folding and association. Pro-scanning mutagenesis of the helical domain of glycophorin A (GpA) showed that replacement of the residues located at the center abrogates helix packing while substitution of the residues forming the ending helical turns allows dimer formation. Synthetic TM peptides revealed that a point mutation of one of the residues of the dimerization motif (L75P) located at the N-terminal helical turn of the GpA TM fragment, adopts a secondary structure and oligomeric state similar to the wild-type sequence in detergents. In addition, both glycosylation mapping in biological membranes and molecular dynamics showed that the presence of a proline residue at the lipid/water interface has as an effect the extension of the helical end. Thus, helix packing can be an important factor that determines appearance of proline in TM helices. Membrane proteins might accumulate proline residues at the two ends of their TM segments in order to modulate the exposition of key amino acid residues at the interface for molecular recognition events while allowing stable association and native folding.     

The control of transmembrane helix transverse position in membranes by hydrophilic residues

The ability of hydrophilic residues to shift the transverse position of transmembrane (TM) helices within bilayers was studied in model membrane vesicles. Transverse shifts were detected by fluorescence measurements of the membrane depth of a Trp residue at the center of a hydrophobic sequence. They were also estimated from the effective length of the TM-spanning sequence, derived from the stability of the TM configuration under conditions of negative hydrophobic mismatch. Hydrophilic residues (at the fifth position in a 21-residue hydrophobic sequence composed of alternating Leu and Ala residues and flanked on both ends by two Lys) induced transverse shifts that moved the hydrophilic residue closer to the membrane surface. At pH 7, the dependence of the extent of shift upon the identity of the hydrophilic residue increased in the order: L < GYT < RH < S < P < K < EQ < N < D. By varying pH, shifts with ionizable residues fully charged or uncharged were measured, and the extent of shift increased in the order: L < GYH^oT < E^oR < S < P < K⁺< QD^oH⁺ < NE^– < D^–. The dependence of transverse shifts upon hydrophilic residue identity was consistent with the hypothesis that shift magnitude is largely controlled by the combination of side chain hydrophilicity, ionization state, and ability to position polar groups near the bilayer surface (snorkeling). Additional experiments showed that shift was also modulated by the position of the hydrophilic residue in the sequence and the hydrophobicity of the sequence moved out of the bilayer core upon shifting. Combined, these studies show that the insertion boundaries of TM helices are very sensitive to sequence, and can be altered even by weakly hydrophilic residues. Thus, many TM helices may have the capacity to exist in more than one transverse position. Knowledge of the magnitudes of transverse shifts induced by different hydrophilic residues should be useful for design of mutagenesis studies measuring the effect of transverse TM helix position upon function.     

Assembly of light-harvesting bacteriochlorophyll in a model transmembrane helix in its natural environment

The transmembrane, bacteriochlorophyll-binding region of a bacterial light-harvesting complex, (LH2-alpha from the photosynthetic bacterium Rhodobacter sphaeroides) was redesigned and overexpressed in a mutant of Rb. sphaeroides lacking LH2. Bacteriochlorophyll served as internal probe for the fitness of this new region for the assembly and energy transfer function of the LH2 complex. The ability to absorb and transfer light energy is practically undisturbed by the exchange of the transmembrane segment, valine -7 to threonine +6, of LH2-alpha with a 14 residue Ala-Leu sequence. This stretch makes up the residues of the transmembrane helix that are in close contact (< or =4.5 A) with the bacteriochlorophyll molecules that are coordinated through His of both the alpha and beta-subunits. In this Ala-Leu stretch, neither alpha-His0, which binds the bacteriochlorophyll, nor the adjacent alpha-Ile-1, were replaced. Novel LH2 complexes composed of LH2-alpha with a model transmembrane sequence and a normal LH2-beta are assembled in vivo into a complex, the biochemical and spectroscopic properties of which closely resemble the native one. In contrast, the additional insertion of four residues just outside the C-terminal end of the model transmembrane helix leads to complete loss of functional antenna complex. The results suggest that light energy can be harvested and transferred efficiently by bacteriochlorophyll molecules attached to only few key residues distributed over the polypeptide, while residues at the bacteriochlorophyll-helix interface seem to be largely dispensable for the functional assembly of this membrane protein complex. This novel antenna with a simplified transmembrane domain and a built-in probe for assembly and function provides a powerful model system for investigation of the factors that contribute to the assembly of chromophores in membrane-embedded proteins.     

Mutagenesis data in the automated prediction of transmembrane helix dimers

We present a molecular modeling protocol that selects modeled protein structures based on experimental mutagenesis results. The computed effect of a point mutation should be consistent with its experimental effect for correct models; mutations that do not affect protein stability and function should not affect the computed energy of a correct model while destabilizing mutations should have unfavorable computed energies. On the other hand, an incorrect model will likely display computed energies that are inconsistent with experimental results. We added terms to our energy function which penalize models that are inconsistent with experimental results. This creates a selective advantage for models that are consistent with experimental results in the Monte Carlo simulated annealing protocol we use to search conformational space. We calibrated our protocol to predict the structure of transmembrane helix dimers using glycophorin A as a model system. Inclusion of mutational data in this protocol compensates for the limitations of our force field and the limitations of our conformational search. We demonstrate an application of this structure prediction protocol by modeling the transmembrane region of the BNIP3 apoptosis factor.     

Marginally hydrophobic transmembrane α‐helices shaping membrane protein folding

Cells have developed an incredible machinery to facilitate the insertion of membrane proteins into the membrane. While we have a fairly good understanding of the mechanism and determinants of membrane integration, more data is needed to understand the insertion of membrane proteins with more complex insertion and folding pathways. This review will focus on marginally hydrophobic transmembrane helices and their influence on membrane protein folding. These weakly hydrophobic transmembrane segments are by themselves not recognized by the translocon and therefore rely on local sequence context for membrane integration. How can such segments reside within the membrane? We will discuss this in the light of features found in the protein itself as well as the environment it resides in. Several characteristics in proteins have been described to influence the insertion of marginally hydrophobic helices. Additionally, the influence of biological membranes is significant. To begin with, the actual cost for having polar groups within the membrane may not be as high as expected; the presence of proteins in the membrane as well as characteristics of some amino acids may enable a transmembrane helix to harbor a charged residue. The lipid environment has also been shown to directly influence the topology as well as membrane boundaries of transmembrane helices—implying a dynamic relationship between membrane proteins and their environment.     

TMKink: a method to predict transmembrane helix kinks

A hallmark of membrane protein structure is the large number of distorted transmembrane helices. Because of the prevalence of bends, it is important to not only understand how they are generated but also to learn how to predict their occurrence. Here, we find that there are local sequence preferences in kinked helices, most notably a higher abundance of proline, which can be exploited to identify bends from local sequence information. A neural network predictor identifies over two-thirds of all bends (sensitivity 0.70) with high reliability (specificity 0.89). It is likely that more structural data will allow for better helix distortion predictors with increased coverage in the future. The kink predictor, TMKink, is available at http://tmkinkpredictor.mbi.ucla.edu/.     

Kinetics of folding and unfolding of goat alpha-lactalbumin

The equilibrium unfolding and the kinetic folding and unfolding of goat alpha-lactalbumin (GLA) were studied by near- and far-ultraviolet circular dichroism (CD) and by stopped-flow fluorescence spectroscopy. Specifically, the influence of environmental conditions such as pH and Ca2+ binding was examined. Compared to the apo-form, the Ca2+-bound form was found to be strongly stabilized in equilibrium conditions at pH 7.5 and 25 degrees C. The kinetics of the refolding of apo-GLA show a major change of fluorescence intensity during the experimental dead-time, but this unresolved effect is strongly diminished in holo-GLA. In both cases, however, the chevron plots can adequately be fitted to a three-state model. Moreover, double-mix stopped-flow experiments showed that the native state (N) is reached through one major pathway without the occurrence of alternative tracks. In contrast to the homologous bovine alpha-lactalbumin (BLA), the compactness of GLA is strongly influenced by the presence of Ca2+ ions. Unlike the two-state transition observed in guanidine hydrochloride (GdnHCl)-induced equilibrium denaturation experiments at higher pH, an equilibrium intermediate state (I) is involved in denaturation at pH 4.5. In the latter case, analysis of the kinetic data makes clear that the intermediate and the unfolded states (U) show practically no Gibbs free energy difference and that they are in rapid equilibrium with each other. A possible explanation for these variations in stability and in folding characteristics with pH could be the degree of protonation of His107 that directly influences non-native interactions. Variation of environmental conditions and even small differences in sequence, therefore, can result in important effects on thermodynamic and folding parameters.     

Predicting repeat protein folding kinetics from an experimentally determined folding energy landscape

The Notch ankyrin domain is a repeat protein whose folding has been characterized through equilibrium and kinetic measurements. In previous work, equilibrium folding free energies of truncated constructs were used to generate an experimentally determined folding energy landscape (Mello and Barrick, Proc Natl Acad Sci USA 2004;101:14102–14107). Here, this folding energy landscape is used to parameterize a kinetic model in which local transition probabilities between partly folded states are based on energy values from the landscape. The landscape‐based model correctly predicts highly diverse experimentally determined folding kinetics of the Notch ankyrin domain and sequence variants. These predictions include monophasic folding and biphasic unfolding, curvature in the unfolding limb of the chevron plot, population of a transient unfolding intermediate, relative folding rates of 19 variants spanning three orders of magnitude, and a change in the folding pathway that results from C‐terminal stabilization. These findings indicate that the folding pathway(s) of the Notch ankyrin domain are thermodynamically selected: the primary determinants of kinetic behavior can be simply deduced from the local stability of individual repeats.     

TMSEG: Novel prediction of transmembrane helices

Transmembrane proteins (TMPs) are important drug targets because they are essential for signaling, regulation, and transport. Despite important breakthroughs, experimental structure determination remains challenging for TMPs. Various methods have bridged the gap by predicting transmembrane helices (TMHs), but room for improvement remains. Here, we present TMSEG, a novel method identifying TMPs and accurately predicting their TMHs and their topology. The method combines machine learning with empirical filters. Testing it on a non‐redundant dataset of 41 TMPs and 285 soluble proteins, and applying strict performance measures, TMSEG outperformed the state‐of‐the‐art in our hands. TMSEG correctly distinguished helical TMPs from other proteins with a sensitivity of 98 ± 2% and a false positive rate as low as 3 ± 1%. Individual TMHs were predicted with a precision of 87 ± 3% and recall of 84 ± 3%. Furthermore, in 63 ± 6% of helical TMPs the placement of all TMHs and their inside/outside topology was correctly predicted. There are two main features that distinguish TMSEG from other methods. First, the errors in finding all helical TMPs in an organism are significantly reduced. For example, in human this leads to 200 and 1600 fewer misclassifications compared to the second and third best method available, and 4400 fewer mistakes than by a simple hydrophobicity‐based method. Second, TMSEG provides an add‐on improvement for any existing method to benefit from. Proteins 2016; 84:1706–1716. © 2016 Wiley Periodicals, Inc.     

Sequence-based study of two related proteins with different folding behaviors

Z(SPA-1) is an engineered protein that binds to its parent, the three-helix-bundle Z domain of staphylococcal protein A. Uncomplexed Z(SPA-1) shows a reduced helix content and a melting behavior that is less cooperative, compared with the wild-type Z domain. Here we show that the difference in folding behavior between these two sequences can be partly understood in terms of an off-lattice model with 5-6 atoms per amino acid and a minimalistic potential, in which folding is driven by backbone hydrogen bonding and effective hydrophobic attraction.     

Early events during folding of wild-type staphylococcal nuclease and a single-tryptophan variant studied by ultrarapid mixing

A continuous-flow mixing device with a dead time of 100 micros coupled with intrinsic tryptophan and 1-anilinonaphthalene-8-sulfonate (ANS) fluorescence was used to monitor structure formation during early stages of the folding of staphylococcal nuclease (SNase). A variant with a unique tryptophan fluorophore in the N-terminal beta-barrel domain (Trp76 SNase) was obtained by replacing the single Trp140 in wild-type SNase with His in combination with Trp substitution of Phe76. A common background of P47G, P117G and H124L mutations was chosen in order to stabilize the protein and prevent accumulation of cis proline isomers under native conditions. In contrast to WT(*) SNase, which shows no changes in tryptophan fluorescence prior to the rate-limiting folding step ( approximately 100 ms), the F76W/W140H variant shows additional changes (enhancement) during an early folding phase with a time constant of 75 micros. Both proteins exhibit a major increase in ANS fluorescence and identical rates for this early folding event. These findings are consistent with the rapid accumulation of an ensemble of states containing a loosely packed hydrophobic core involving primarily the beta-barrel domain while the specific interactions in the alpha-helical domain involving Trp140 are formed only during the final stages of folding. The fact that both variants exhibit the same number of kinetic phases with very similar rates confirms that the folding mechanism is not perturbed by the F76W/W140H mutations. However, the Trp at position 76 reports on the rapid formation of a hydrophobic cluster in the N-terminal beta-sheet region while the wild-type Trp140 is silent during this early stage of folding. Quantitative modeling of the (un)folding kinetics and thermodynamics of these two proteins versus urea concentration revealed that the F76W/W140H mutation selectively destabilizes the native state relative to WT(*) SNase while the stability of transient intermediates remains unchanged, leading to accumulation of intermediates under equilibrium conditions at moderate denaturant concentrations.     

Controlling the folding efficiency of an integral membrane protein

Research into the folding mechanisms of integral membrane proteins lags far behind that of water-soluble proteins, to the extent that the term protein folding is synonymous with water-soluble proteins. Hydrophobic membrane proteins, and particularly those with transmembrane alpha-helical motifs, are frequently considered too difficult to work with. We show that the stored curvature elastic stress of lipid bilayers can be used to guide the design of efficient folding systems for these integral membrane proteins. The curvature elastic stress of synthetic phosphatidylcholine/phosphatidylethanolamine lipid bilayers can be used to control both the rate of folding and the yield of folded protein. The use of a physical bilayer property generalises this approach beyond the particular chemistry of the lipids involved.     

Sequence and conformational preferences at termini of α‐helices in membrane proteins: Role of the helix environment

α‐helices are amongst the most common secondary structural elements seen in membrane proteins and are packed in the form of helix bundles. These α‐helices encounter varying external environments (hydrophobic, hydrophilic) that may influence the sequence preferences at their N and C‐termini. The role of the external environment in stabilization of the helix termini in membrane proteins is still unknown. Here we analyze α‐helices in a high‐resolution dataset of integral α‐helical membrane proteins and establish that their sequence and conformational preferences differ from those in globular proteins. We specifically examine these preferences at the N and C‐termini in helices initiating/terminating inside the membrane core as well as in linkers connecting these transmembrane helices. We find that the sequence preferences and structural motifs at capping (Ncap and Ccap) and near‐helical (N' and C') positions are influenced by a combination of features including the membrane environment and the innate helix initiation and termination property of residues forming structural motifs. We also find that a large number of helix termini which do not form any particular capping motif are stabilized by formation of hydrogen bonds and hydrophobic interactions contributed from the neighboring helices in the membrane protein. We further validate the sequence preferences obtained from our analysis with data from an ultradeep sequencing study that identifies evolutionarily conserved amino acids in the rat neurotensin receptor. The results from our analysis provide insights for the secondary structure prediction, modeling and design of membrane proteins. Proteins 2014; 82:3420–3436. © 2014 Wiley Periodicals, Inc.