An empirical energy function for structural assessment of protein transmembrane domains

Connexin channels and hemichannels are modulated differently by charge reversal at residues forming the intracellular pocket

BACKGROUND

Members of the β-subfamily of connexins contain an intracellular pocket surrounded by amino acid residues from the four transmembrane helices. The presence of this pocket has not previously been investigated in members of the α-, γ-, δ-, and ε-subfamilies. We studied connexin50 (Cx50) as a representative of the α-subfamily, because its structure has been determined and mutations of Cx50 are among the most common genetic causes of congenital cataracts.

METHODS

To investigate the presence and function of the intracellular pocket in Cx50 we used molecular dynamics simulation, site-directed mutagenesis, gap junction tracer intercellular transfer, and hemichannel activity detected by electrophysiology and by permeation of charged molecules.

RESULTS

Employing molecular dynamics, we determined the presence of the intracellular pocket in Cx50 hemichannels and identified the amino acids participating in its formation. We utilized site-directed mutagenesis to alter a salt-bridge interaction that supports the intracellular pocket and occurs between two residues highly conserved in the connexin family, R33 and E162. Substitution of opposite charges at either position decreased formation of gap junctional plaques and cell-cell communication and modestly reduced hemichannel currents. Simultaneous charge reversal at these positions produced plaque-forming non-functional gap junction channels with highly active hemichannels.

CONCLUSIONS

These results show that interactions within the intracellular pocket influence both gap junction channel and hemichannel functions. Disruption of these interactions may be responsible for diseases associated with mutations at these positions.

Expression of KID syndromic mutation Cx26S17F produces hyperactive hemichannels in supporting cells of the organ of Corti

Some mutations in gap junction protein Connexin 26 (Cx26) lead to syndromic deafness, where hearing impairment is associated with skin disease, like in Keratitis Ichthyosis Deafness (KID) syndrome. This condition has been linked to hyperactivity of connexin hemichannels but this has never been demonstrated in cochlear tissue. Moreover, some KID mutants, like Cx26S17F, form hyperactive HCs only when co-expressed with other wild-type connexins. In this work, we evaluated the functional consequences of expressing a KID syndromic mutation, Cx26S17F, in the transgenic mouse cochlea and whether co-expression of Cx26S17F and Cx30 leads to the formation of hyperactive HCs. Indeed, we found that cochlear explants from a constitutive knock-in Cx26S17F mouse or conditional in vitro cochlear expression of Cx26S17F produces hyperactive HCs in supporting cells of the organ of Corti. These conditions also produce loss of hair cells stereocilia. In supporting cells, we found high co-localization between Cx26S17F and Cx30. The functional properties of HCs formed in cells co-expressing Cx26S17F and Cx30 were also studied in oocytes and HeLa cells. Under the recording conditions used in this study Cx26S17F did not form functional HCs and GJCs, but cells co-expressing Cx26S17F and Cx30 present hyperactive HCs insensitive to HCs blockers, Ca²⁺ and La³⁺, resulting in more Ca²⁺ influx and cellular damage. Molecular dynamic analysis of putative heteromeric HC formed by Cx26S17F and Cx30 presents alterations in extracellular Ca²⁺ binding sites. These results support that in KID syndrome, hyperactive HCs are formed by the interaction between Cx26S17F and Cx30 in supporting cells probably causing damage to hair cells associated to deafness.

BRANEart: Identify Stability Strength and Weakness Regions in Membrane Proteins

Understanding the role of stability strengths and weaknesses in proteins is a key objective for rationalizing their dynamical and functional properties such as conformational changes, catalytic activity, and protein-protein and protein-ligand interactions. We present BRANEart, a new, fast and accurate method to evaluate the per-residue contributions to the overall stability of membrane proteins. It is based on an extended set of recently introduced statistical potentials derived from membrane protein structures, which better describe the stability properties of this class of proteins than standard potentials derived from globular proteins. We defined a per-residue membrane propensity index from combinations of these potentials, which can be used to identify residues which strongly contribute to the stability of the transmembrane region or which would, on the contrary, be more stable in extramembrane regions, or vice versa. Large-scale application to membrane and globular proteins sets and application to tests cases show excellent agreement with experimental data. BRANEart thus appears as a useful instrument to analyze in detail the overall stability properties of a target membrane protein, to position it relative to the lipid bilayer, and to rationally modify its biophysical characteristics and function. BRANEart can be freely accessed from http://babylone.3bio.ulb.ac.be/BRANEart.

A comprehensive computational study of amino acid interactions in membrane proteins

Transmembrane proteins play a fundamental role in a wide series of biological processes but, despite their importance, they are less studied than globular proteins, essentially because their embedding in lipid membranes hampers their experimental characterization. In this paper, we improved our understanding of their structural stability through the development of new knowledge-based energy functions describing amino acid pair interactions that prevail in the transmembrane and extramembrane regions of membrane proteins. The comparison of these potentials and those derived from globular proteins yields an objective view of the relative strength of amino acid interactions in the different protein environments, and their role in protein stabilization. Separate potentials were also derived from α-helical and β-barrel transmembrane regions to investigate possible dissimilarities. We found that, in extramembrane regions, hydrophobic residues are less frequent but interactions between aromatic and aliphatic amino acids as well as aromatic-sulfur interactions contribute more to stability. In transmembrane regions, polar residues are less abundant but interactions between residues of equal or opposite charges or non-charged polar residues as well as anion-π interactions appear stronger. This shows indirectly the preference of the water and lipid molecules to interact with polar and hydrophobic residues, respectively. We applied these new energy functions to predict whether a residue is located in the trans- or extramembrane region, and obtained an AUC score of 83% in cross validation, which demonstrates their accuracy. As their application is, moreover, extremely fast, they are optimal instruments for membrane protein design and large-scale investigations of membrane protein stability.

Repository of Enriched Structures of Proteins Involved in the Red Blood Cell Environment (RESPIRE)

The Red Blood Cell (RBC) is a metabolically-driven cell vital for processes such a gas transport and homeostasis. RBC possesses at its surface exposing antigens proteins that are critical in blood transfusion. Due to their importance, numerous studies address the cell function as a whole but more and more details of RBC structure and protein content are now studied using massive state-of-the art characterisation techniques. Yet, the resulting information is frequently scattered in many scientific articles, in many databases and specialized web servers. To provide a more compendious view of erythrocytes and of their protein content, we developed a dedicated database called RESPIRE that aims at gathering a comprehensive and coherent ensemble of information and data about proteins in RBC. This cell-driven database lists proteins found in erythrocytes. For a given protein entry, initial data are processed from external portals and enriched by using state-of-the-art bioinformatics methods. As structural information is extremely useful to understand protein function and predict the impact of mutations, a strong effort has been put on the prediction of protein structures with a special treatment for membrane proteins. Browsing the database is available through text search for reference gene names or protein identifiers, through pre-defined queries or via hyperlinks. The RESPIRE database provides valuable information and unique annotations that should be useful to a wide audience of biologists, clinicians and structural biologists.

Database URL:http://www.dsimb.inserm.fr/respire

MyPMFs: a simple tool for creating statistical potentials to assess protein structural models

Evaluating the model quality of protein structures that evolve in environments with particular physicochemical properties requires scoring functions that are adapted to their specific residue compositions and/or structural characteristics. Thus, computational methods developed for structures from the cytosol cannot work properly on membrane or secreted proteins. Here, we present MyPMFs, an easy-to-use tool that allows users to train statistical potentials of mean force (PMFs) on the protein structures of their choice, with all parameters being adjustable. We demonstrate its use by creating an accurate statistical potential for transmembrane protein domains. We also show its usefulness to study the influence of the physical environment on residue interactions within protein structures. Our open-source software is freely available for download at https://github.com/bibip-impmc/mypmfs.

The syndromic deafness mutation G12R impairs fast and slow gating in Cx26 hemichannels

Mutations in connexin 26 hemichannels that cause syndromic deafness have a gain-of-function phenotype that is poorly understood. García et al. show that one such mutation impairs fast and slow gating in these hemichannels because of an interaction between the N terminus and intracellular loop.

Mutations in connexin 26 (Cx26) hemichannels can lead to syndromic deafness that affects the cochlea and skin. These mutations lead to gain-of-function hemichannel phenotypes by unknown molecular mechanisms. In this study, we investigate the biophysical properties of the syndromic mutant Cx26G12R (G12R). Unlike wild-type Cx26, G12R macroscopic hemichannel currents do not saturate upon depolarization, and deactivation is faster during hyperpolarization, suggesting that these channels have impaired fast and slow gating. Single G12R hemichannels show a large increase in open probability, and transitions to the subconductance state are rare and short-lived, demonstrating an inoperative fast gating mechanism. Molecular dynamics simulations indicate that G12R causes a displacement of the N terminus toward the cytoplasm, favoring an interaction between R12 in the N terminus and R99 in the intracellular loop. Disruption of this interaction recovers the fast and slow voltage-dependent gating mechanisms. These results suggest that the mechanisms of fast and slow gating in connexin hemichannels are coupled and provide a molecular mechanism for the gain-of-function phenotype displayed by the syndromic G12R mutation.

On Biophysical Properties and Sensitivity to Gap Junction Blockers of Connexin 39 Hemichannels Expressed in HeLa Cells

Although connexins (Cxs) are broadly expressed by cells of mammalian organisms, Cx39 has a very restricted pattern of expression and the biophysical properties of Cx39-based channels [hemichannels (HCs) and gap junction channels (GJCs)] remain largely unknown. Here, we used HeLa cells transfected with Cx39 (HeLa-Cx39 cells) in which intercellular electrical coupling was not detected, indicating the absence of GJCs. However, functional HCs were found on the surface of cells exposed to conditions known to increase the open probability of other Cx HCs (e.g., extracellular divalent cationic-free solution (DCFS), extracellular alkaline pH, mechanical stimulus and depolarization to positive membrane potentials). Cx39 HCs were blocked by some traditional Cx HC blockers, but not by others or a pannexin1 channel blocker. HeLa-Cx39 cells showed similar resting membrane potentials (RMPs) to those of parental cells, and exposure to DCFS reduced RMPs in Cx39 transfectants, but not in parental cells. Under these conditions, unitary events of ~75 pS were frequent in HeLa-Cx39 cells and absent in parental cells. Real-time cellular uptake experiments of dyes with different physicochemical features, as well as the application of a machine-learning approach revealed that Cx39 HCs are preferentially permeable to molecules characterized by six categories of descriptors, namely: (1) electronegativity, (2) ionization potential, (3) polarizability, (4) size and geometry, (5) topological flexibility and (6) valence. However, Cx39 HCs opened by mechanical stimulation or alkaline pH were impermeable to Ca²⁺. Molecular modeling of Cx39-based channels suggest that a constriction present at the intracellular portion of the para helix region co-localizes with an electronegative patch, imposing an energetic and steric barrier, which in the case of GJCs may hinder channel function. Results reported here demonstrate that Cx39 form HCs and add to our understanding of the functional roles of Cx39 HCs under physiological and pathological conditions in cells that express them.

TMPL: a database of experimental and theoretical transmembrane protein models positioned in the lipid bilayer

Knowing the position of protein structures within the membrane is crucial for fundamental and applied research in the field of molecular biology. Only few web resources propose coordinate files of oriented transmembrane proteins, and these exclude predicted structures, although they represent the largest part of the available models. In this article, we present TMPL (http://www.dsimb.inserm.fr/TMPL/), a database of transmembrane protein structures (α-helical and β-sheet) positioned in the lipid bilayer. It is the first database to include theoretical models of transmembrane protein structures, making it a large repository with more than 11 000 entries. The TMPL database also contains experimentally solved protein structures, which are available as either atomistic or coarse-grained models. A unique feature of TMPL is the possibility for users to update the database by uploading, through an intuitive web interface, the membrane assignments they can obtain with our recent OREMPRO web server.

OREMPRO web server: orientation and assessment of atomistic and coarse-grained structures of membrane proteins

: The experimental determination of membrane protein orientation within the lipid bilayer is extremely challenging, such that computational methods are most often the only solution. Moreover, obtaining all-atom 3D structures of membrane proteins is also technically difficult, and many of the available data are either experimental low-resolution structures or theoretical models, whose structural quality needs to be evaluated. Here, to address these two crucial problems, we propose OREMPRO, a web server capable of both (i) positioning α-helical and β-sheet transmembrane domains in the lipid bilayer and (ii) assessing their structural quality. Most importantly, OREMPRO uses the sole alpha carbon coordinates, which makes it the only web server compatible with both high and low structural resolutions. Finally, OREMPRO is also interesting in its ability to process coarse-grained protein models, by using coordinates of backbone beads in place of alpha carbons.

AVAILABILITY AND IMPLEMENTATION

http://www.dsimb.inserm.fr/OREMPRO/ CONTACT: : guillaume.postic@univ-paris-diderot.fr or jean-christophe.gelly@univ-paris-diderot.fr

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Membrane positioning for high- and low-resolution protein structures through a binary classification approach

The critical importance of algorithms for orienting proteins in the lipid bilayer stems from the extreme difficulty in obtaining experimental data about the membrane boundaries. Here, we present a computational method for positioning protein structures in the membrane, based on the sole alpha carbon coordinates and, therefore, compatible with both high and low structural resolutions. Our algorithm follows a new and simple approach, by treating the membrane assignment problem as a binary classification. Compared with the state-of-the-art algorithms, our method achieves similar accuracy, while being faster. Finally, our open-source software is also capable of processing coarse-grained models of protein structures.