Backbone-backbone Hydrogen Bonds Research Articles

Elucidating the molecular basis of spontaneous activation in an engineered mechanosensitive channel.

Mechanosensitive channel of large conductance (MscL) protein detect and respond to changes in the pressure profile of cellular membranes and transduce the mechanical energy into electrical and/or chemical signals. MscL can be activated using ultrasonic or chemical activation methods to improve the absorption of medicines and bioactive compounds into cells. However, by re-engineering the MscL, chemical signals such as pH change can trigger the channel's activation. This paper elucidates the activation mechanism of an engineered and wild-type (WT) MscL at an atomic level through a combination of equilibrium and non-equilibrium (NE) molecular dynamics (MD) simulations, the String method with swarms of trajectories (SMwST), and Expectation maximization (EM) methods. Compared to the WT and engineered MscL activation processes, it suggests that the two systems are likely associated with different active states and different transition pathways. These findings indicate that (1) periplasmic loops play a key role in the activation process of MscL; (2) the loss of various backbone-backbone hydrogen bonds and salt bridge interactions in the engineered MscL channel causes the spontaneous opening of the channel; and (3) the most significant interactions that were lost during the activation process were those between the transmembrane helices 1 and 2 in the engineered MscL channel. In this work, the orientation-based biasing approach for producing and optimizing an open MscL model is a promising way to systematically characterize unknown protein functional states and to research the activation processes in ion channels. This work makes it possible to study ways to design pH-triggered channel-functionalized liposomes for drug delivery.

Elucidating the molecular basis of spontaneous activation in an engineered mechanosensitive channel

Mechanosensitive channel of large conductance (MscL) detects and responds to changes in the pressure profile of cellular membranes and transduces the mechanical energy into electrical and/or chemical signals. MscL can be activated using ultrasonic or chemical activation methods to improve the absorption of medicines and bioactive compounds into cells. However, re-engineering chemical signals such as pH change can trigger channel activation in MscL. This study elucidates the activation mechanism of an engineered MscL at an atomic level through a combination of equilibrium and non-equilibrium (NE) molecular dynamics (MD) simulations. Comparing the wild-type (WT) and engineered MscLactivation processes suggests that the two systems are likely associated with different active states and different transition pathways. These findings indicate that (1) periplasmic loops play a key role in the activation process of MscL, (2) the loss of various backbone-backbone hydrogen bonds and salt bridge interactions in the engineered MscLchannel causes the spontaneous opening of the channel, and (3) the most significant interactions lost during the activation process are between the transmembrane helices 1 and 2 in engineered MscLchannel. The orientation-based biasing approach for producing and optimizing an open MscL model used in this work is a promising way to characterize unknown protein functional states and investigate the activation processes in ion channels and transmembrane proteins in general. This work paves the way for a computational framework for engineering more efficient pH-sensing mechanosensitive channels.

A tensegrity model for hydrogen bond networks in proteins.

Hydrogen-bonding networks in proteins considered as structural tensile elements are in balance separately from any other stabilising interactions that may be in operation. The hydrogen bond arrangement in the network is reminiscent of tensegrity structures in architecture and sculpture. Tensegrity has been discussed before in cells and tissues and in proteins. In contrast to previous work only hydrogen bonds are studied here. The other interactions within proteins are either much stronger − covalent bonds connecting the atoms in the molecular skeleton or weaker forces like the so-called hydrophobic interactions. It has been demonstrated that the latter operate independently from hydrogen bonds. Each category of interaction must, if the protein is to have a stable structure, balance out. The hypothesis here is that the entire hydrogen bond network is in balance without any compensating contributions from other types of interaction. For sidechain-sidechain, sidechain-backbone and backbone-backbone hydrogen bonds in proteins, tensegrity balance (“closure”) is required over the entire length of the polypeptide chain that defines individually folding units in globular proteins (“domains”) as well as within the repeating elements in fibrous proteins that consist of extended chain structures. There is no closure to be found in extended structures that do not have repeating elements. This suggests an explanation as to why globular domains, as well as the repeat units in fibrous proteins, have to have a defined number of residues. Apart from networks of sidechain-sidechain hydrogen bonds there are certain key points at which this closure is achieved in the sidechain-backbone hydrogen bonds and these are associated with demarcation points at the start or end of stretches of secondary structure. Together, these three categories of hydrogen bond achieve the closure that is necessary for the stability of globular protein domains as well as repeating elements in fibrous proteins.

In silico study of amphiphilic nanotubes based on cyclic peptides in polar and non-polar solvent.

The stability of cyclic peptide assemblies (CPs) forming a macromolecular nanotube structure was investigated in solvents of different polarity using computational methods. The stability and structure of the complexes were studied using traditional molecular dynamics (MD). Energy of dissociation was estimated from steered MD in combination with umbrella sampling simulations. A cyclic peptide nanotube (CPNT) was constructed by stacking of eight cyclo[(D-Trp-L-Gln-D-Trp-L-Glu)2], and hereafter is referred to as (WQWE)8. Its dissociation was studied by pulling 1, 2, or 3 subunits from the nanotube. The crucial point in the dissociation event of the CP subunit(s) is the breaking of backbone-backbone hydrogen bonds and consecutive annihilation of side chain interactions. Gibbs free energy calculations to estimate the binding affinity of CP subunit(s) reveal that the (WQWE)8 nanotube is significantly more stable in non-polar environments than in polar environments. The presently investigated nanotube, (WQWE)8, displays a higher stability in polar solvent than the previously studied nanotube, (QAEA)8. It appears that tryptophan contributes favorable to the improved stability by forming side chain-side chain hydrogen bonds.

Dissecting Molecular Interactions Involved in Recognition of Target Disulfides by the Barley Thioredoxin System

Thioredoxin reduces disulfide bonds, thus regulating activities of target proteins in various biological systems, e.g., inactivation of inhibitors of starch hydrolases and proteases in germinating plant seeds. In the three-dimensional structure of a complex with barley α-amylase/subtilisin inhibitor (BASI), two loops in barley thioredoxin h2 (HvTrxh2), containing an invariant cis-proline ((86)EAMP(89)) and a conserved glycine ((104)VGA(106)), surround the active site cysteines ((45)WCGPC(49)) and contribute to binding of BASI through backbone-backbone hydrogen bonds [Maeda, K., Hägglund, P., Finnie, C., Svensson, B., and Henriksen, A. (2006) Structure 14, 1701-1710]. This study involves mutational analysis of key amino acid residues from these two loops in reactions with three protein disulfide substrates, BASI, barley glutathione peroxidase, and bovine insulin as well as with NADPH-dependent barley thioredoxin reductase. HvTrxh2 M88G and M88A adjacent to the invariant cis-proline lost efficiency in both BASI disulfide reduction and recycling by thioredoxin reductase. These effects were further pronounced in M88P lacking a backbone NH group. Remarkably, HvTrxh2 E86R in the same loop displayed overall retained catalytic properties, with the exception of a 3-fold increased activity toward BASI. From the (104)VGA(106) loop, a backbone hydrogen bond donated by A106 appears to be important for target disulfide recognition as A106P lost 90% activity toward BASI but was efficiently recycled by thioredoxin reductase. The findings support important roles in target recognition of backbone-backbone hydrogen bond and electrostatic interactions and are discussed in relation to earlier structural and functional studies of thioredoxins and related proteins.

Biochemical Basis of the Interaction between Cystic Fibrosis Transmembrane Conductance Regulator and Immunoglobulin-like Repeats of Filamin

Mutations in the chloride channel cystic fibrosis transmembrane regulator (CFTR) cause cystic fibrosis, a genetic disorder characterized by defects in CFTR biosynthesis, localization to the cell surface, or activation by regulatory factors. It was discovered recently that surface localization of CFTR is stabilized by an interaction between the CFTR N terminus and the multidomain cytoskeletal protein filamin. The details of the CFTR-filamin interaction, however, are unclear. Using x-ray crystallography, we show how the CFTR N terminus binds to immunoglobulin-like repeat 21 of filamin A (FlnA-Ig21). CFTR binds to beta-strands C and D of FlnA-Ig21 using backbone-backbone hydrogen bonds, a linchpin serine residue, and hydrophobic side-chain packing. We use NMR to determine that the CFTR N terminus also binds to several other immunoglobulin-like repeats from filamin A in vitro. Our structural data explain why the cystic fibrosis-causing S13F mutation disrupts CFTR-filamin interaction. We show that FlnA-Ig repeats transfected into cultured Calu-3 cells disrupt CFTR-filamin interaction and reduce surface levels of CFTR. Our findings suggest that filamin A stabilizes surface CFTR by anchoring it to the actin cytoskeleton through interactions with multiple filamin Ig repeats. Such an interaction mode may allow filamins to cluster multiple CFTR molecules and to promote colocalization of CFTR and other filamin-binding proteins in the apical plasma membrane of epithelial cells.

Free‐energy function based on an all‐atom model for proteins

We have developed a free-energy function based on an all-atom model for proteins. It comprises two components, the hydration entropy (HE) and the total dehydration penalty (TDP). Upon a transition to a more compact structure, the number of accessible configurations arising from the translational displacement of water molecules in the system increases, leading to a water-entropy gain. To fully account for this effect, the HE is calculated using a statistical-mechanical theory applied to a molecular model for water. The TDP corresponds to the sum of the hydration energy and the protein intramolecular energy when a fully extended structure, which possesses the maximum number of hydrogen bonds with water molecules and no intramolecular hydrogen bonds, is chosen as the standard one. When a donor and an acceptor (e.g., N and O, respectively) are buried in the interior after the break of hydrogen bonds with water molecules, if they form an intramolecular hydrogen bond, no penalty is imposed. When a donor or an acceptor is buried with no intramolecular hydrogen bond formed, an energetic penalty is imposed. We examine all the donors and acceptors for backbone-backbone, backbone-side chain, and side chain-side chain intramolecular hydrogen bonds and calculate the TDP. Our free-energy function has been tested for three different decoy sets. It is better than any other physics-based or knowledge-based potential function in terms of the accuracy in discriminating the native fold from misfolded decoys and the achievement of high Z-scores.

New angle‐dependent potential energy function for backbone–backbone hydrogen bond in protein–protein interactions

Backbone-backbone hydrogen bonds (BBHBs) are one of the most abundant interactions at the interface of protein-protein complex. Here, we propose an angle-dependent potential energy function for BBHB based on density functional theory (DFT) calculations and the operation of a genetic algorithm to find the optimal parameters in the potential energy function. The angular part of the energy function is assumed to be the product of the power series of sine and cosine functions with respect to the two angles associated with BBHB. Two radial functions are taken into account in this study: Morse and Leonard-Jones 12-10 potential functions. Of these two functions under consideration, the former is found to be more accurate than the latter in terms of predicting the binding energies obtained from DFT calculations. The new HB potential function also compares well with the knowledge-based potential derived by applying Boltzmann statistics for a variety of protein-protein complexes in protein data bank.

Probing the Folding Transition State Structure of the Villin Headpiece Subdomain via Side Chain and Backbone Mutagenesis

Backbone-backbone hydrogen bonds are a common feature of native protein structures, yet their thermodynamic and kinetic influence on folding has long been debated. This is reflected by the disparity between current protein folding models, which place hydrogen bond formation at different stages along the folding trajectory. For example, previous studies have suggested that the denatured state of the villin headpiece subdomain contains a residual helical structure that may provide a bias toward the folded state by confining the conformational search associated with its folding. Although helical hydrogen bonds clearly stabilize the folded state, here we show, using an amide-to-ester mutation strategy, that the formation of backbone hydrogen bonds within helices is not rate-limiting in the folding of the subdomain, thereby suggesting that such hydrogen bonds are unlikely to be formed en route from the denatured to the transition state. On the other hand, elimination of hydrogen bonds within the turn region elicits a slower folding rate, consistent with the hypothesis that these residues are involved in the formation of a folding nucleus. While illustrating a potentially conserved aspect of helix-turn-helix folding, our results further underscore the inherent importance of turns in protein supersecondary structure formation.

Structure and Energetics of the Hydrogen-Bonded Backbone in Protein Folding

We seek to understand the link between protein thermodynamics and protein structure in molecular detail. A classical approach to this problem involves assessing changes in protein stability resulting from added cosolvents. Under any given conditions, protein molecules in aqueous buffer are in equilibrium between unfolded and folded states, U(nfolded) <==> N(ative). Addition of organic osmolytes, small uncharged compounds found throughout nature, shift this equilibrium. Urea, a denaturing osmolyte, shifts the equilibrium toward U; trimethylamine N-oxide (TMAO), a protecting osmolyte, shifts the equilibrium toward N. Using the Tanford Transfer Model, the thermodynamic response to many such osmolytes has been dissected into groupwise free energy contributions. It is found that the energetics involving backbone hydrogen bonding controls these shifts in protein stability almost entirely, with osmolyte cosolvents simply dialing between solvent-backbone versus backbone-backbone hydrogen bonds, as a function of solvent quality. This reciprocal relationship establishes the essential link between protein thermodynamics and the protein's hydrogen-bonded backbone structure.

Structural Basis for Target Protein Recognition by the Protein Disulfide Reductase Thioredoxin

Thioredoxin is ubiquitous and regulates various target proteins through disulfide bond reduction. We report the structure of thioredoxin (HvTrxh2 from barley) in a reaction intermediate complex with a protein substrate, barley alpha-amylase/subtilisin inhibitor (BASI). The crystal structure of this mixed disulfide shows a conserved hydrophobic motif in thioredoxin interacting with a sequence of residues from BASI through van der Waals contacts and backbone-backbone hydrogen bonds. The observed structural complementarity suggests that the recognition of features around protein disulfides plays a major role in the specificity and protein disulfide reductase activity of thioredoxin. This novel insight into the function of thioredoxin constitutes a basis for comprehensive understanding of its biological role. Moreover, comparison with structurally related proteins shows that thioredoxin shares a mechanism with glutaredoxin and glutathione transferase for correctly positioning substrate cysteine residues at the catalytic groups but possesses a unique structural element that allows recognition of protein disulfides.

Stability and Structure of Oligomers of the Alzheimer Peptide A β16–22: From the Dimer to the 32-Mer

Several neurodegenerative diseases such as Alzheimer’s, Parkinson’s, and Huntington’s diseases are associated with amyloid fibrils formed by different polypeptides. We probe the structure and stability of oligomers of different sizes of the fragment A β 16–22 of the Alzheimer β-amyloid peptide using atomic-detail molecular dynamics simulations with explicit solvent. We find that only large oligomers form a stable β-sheet aggregate, the minimum nucleus size being of the order of 8–16 peptides. This effect is attributed to better hydrophobic contacts and a better shielding of backbone-backbone hydrogen bonds from the solvent in bigger assemblies. Moreover, the observed stability of β-sheet aggregates with a different number of layers can be explained on the basis of their solvent-accessible surface area. Depending on the stacking interface between the sheets, we observe straight or twisted structures, which could be linked to the experimentally observed polymorphism of amyloid fibrils. To compare our 32-mer structure to experimental data, we calculate its x-ray diffraction pattern. Good agreement is found between experimentally and theoretically determined reflections, suggesting that our model indeed closely resembles the structures found in vitro.

Structural and Biophysical Characterization of the EphB4·EphrinB2 Protein-Protein Interaction and Receptor Specificity

Increasing evidence implicates the interaction of the EphB4 receptor with its preferred ligand, ephrinB2, in pathological forms of angiogenesis and in tumorigenesis. To identify the molecular determinants of the unique specificity of EphB4 for ephrinB2, we determined the crystal structure of the ligand binding domain of EphB4 in complex with the extracellular domain of ephrinB2. This structural analysis suggested that one amino acid, Leu-95, plays a particularly important role in defining the structural features that confer the ligand selectivity of EphB4. Indeed, all other Eph receptors, which promiscuously bind many ephrins, have a conserved arginine at the position corresponding to Leu-95 of EphB4. We have also found that amino acid changes in the EphB4 ligand binding cavity, designed based on comparison with the crystal structure of the more promiscuous EphB2 receptor, yield EphB4 variants with altered binding affinity for ephrinB2 and an antagonistic peptide. Isothermal titration calorimetry experiments with an EphB4 Leu-95 to arginine mutant confirmed the importance of this amino acid in conferring high affinity binding to both ephrinB2 and the antagonistic peptide ligand. Isothermal titration calorimetry measurements also revealed an interesting thermodynamic discrepancy between ephrinB2 binding, which is an entropically driven process, and peptide binding, which is an enthalpically driven process. These results provide critical information on the EphB4*ephrinB2 protein interfaces and their mode of interaction, which will facilitate development of small molecule compounds inhibiting the EphB4*ephrinB2 interaction as novel cancer therapeutics.

Structural Stability and Dynamics of an Amyloid-Forming Peptide GNNQQNY from the Yeast Prion Sup-35

A seven amino acid yeast prion sup-35 fragment (GNNQQNY) forms amyloid fibrils. The availability of its detailed atomic oligomeric structure makes it a good model for studying the early stage of aggregation. Here we perform long all-atom explicit solvent molecular simulations of various sizes and arrangements of oligomer seeds of the wild-type and its mutants to study its stability and dynamics. Previous studies have suggested that the early stage rate-limiting step of oligomer formation occurs in high-order oligomers. Our simulations show that with the increase in the number of strands even from a dimer to a trimer, oligomer stability increases dramatically. This suggests that the minimal nucleus seed for GNNQQNY fibril formation could be small and is likely three or four peptides, in agreement with experiment, and that higher-order oligomers do not dissociate quickly since they have small diffusion coefficients and thus slow kinetics. Further, for the hydrophilic polar GNNQQNY, there are no hydrogen bonds and no hydrophobic interactions between adjacent β-sheets. Simulations suggest that within the sheet, the driving forces to associate and stabilize are interstrand backbone-backbone and side chain-side chain hydrogen bonds, whereas between the sheets, shape-complementary by the dry polar steric zipper via the side chains of Asn-2, Gln-4, and Asn-6 holds the sheets together, as proposed in an earlier study. Since the polar side chains of Asn-2, Gln-4, and Asn-6 act as a hook to bind two neighboring sheets together, these geometric restraints reduce the conformational search for the correct side chain packing to a two-dimensional problem of intersheet side chain interactions. Mutant simulations show that substitution of Asn-2, Gln-4, or Asn-6 by Ala would disrupt this steric zipper, leading to unstable oligomers.

Two-rung Model of a Left-handed β-Helix for Prions Explains Species Barrier and Strain Variation in Transmissible Spongiform Encephalopathies

In this study, a new β-helical model is proposed that explains the species barrier and strain variation in transmissible spongiform encephalopathies. The left-handed β-helix serves as a structural model that can explain the seeded growth characteristics of β-sheet structure in PrP Sc fibrils. Molecular dynamics simulations demonstrate that the left-handed β-helix is structurally more stable than the right-handed β-helix, with a higher β-sheet content during the simulation and a better distributed network of inter-strand backbone-backbone hydrogen bonds between parallel β-strands of different rungs. Multiple sequence alignments and homology modelling of prion sequences with different rungs of left-handed β-helices illustrate that the PrP region with the highest β-helical propensity (residues 105–143) can fold in just two rungs of a left-handed β-helix. Even if no other flanking sequence participates in the β-helix, the two rungs of a β-helix can give the growing fibril enough elevation to accommodate the rest of the PrP protein in a tight packing at the periphery of a trimeric β-helix. The folding of β-helices is driven by backbone–backbone hydrogen bonding and stacking of side-chains in adjacent rungs. The sequence and structure of the last rung at the fibril end with unprotected β-sheet edges selects the sequence of a complementary rung and dictates the folding of the new rung with optimal backbone hydrogen bonding and side-chain stacking. An important side-chain stack that facilitates the β-helical folding is between methionine residues 109 and 129, which explains their importance in the species barrier of prions. Because the PrP sequence is not evolutionarily optimised to fold in a β-helix, and because the β-helical fold shows very little sequence preference, alternative alignments are possible that result in a different rung able to select for an alternative complementary rung. A different top rung results in a new strain with different growth characteristics. Hence, in the present model, sequence variation and alternative alignments clarify the basis of the species barrier and strain specificity in PrP-based diseases.

Solution Structure of Human Peptidyl Prolyl Isomerase-like Protein 1 and Insights into Its Interaction with SKIP

The human PPIL1 (peptidyl prolyl isomerase-like protein 1) is a specific component of human 35 S U5 small nuclear ribonucleoprotein particle and 45 S activated spliceosome. It is recruited by SKIP, another essential component of 45 S activated spliceosome, into spliceosome just before the catalytic step 1. It stably associates with SKIP, which also exists in 35 S and activated spliceosome as a nuclear matrix protein. We report here the solution structure of PPIL1 determined by NMR spectroscopy. The structure of PPIL1 resembles other members of the cyclophilin family and exhibits PPIase activity. To investigate its interaction with SKIP in vitro, we identified the SKIP contact region by GST pulldown experiments and surface plasmon resonance. We provide direct evidence of PPIL1 stably associated with SKIP. The dissociation constant is 1.25 x 10(-7) M for the N-terminal peptide of SKIP-(59-129) with PPIL1. We also used chemical shift perturbation experiments to show the possible SKIP binding interface on PPIL1. These results illustrated that a novel cyclophilin-protein contact mode exists in the PPIL1-SKIP complex during activation of the spliceosome. The biological implication of this binding with spliceosome rearrangement during activation is discussed.

Favorable scaffolds: proteins with different sequence, structure and function may associate in similar ways

Proteins with similar structures may have different functions. Here, using a non-redundant two-chain protein-protein interface dataset containing 103 clusters, we show that this paradigm extends to interfaces. Whereas usually similar interfaces are obtained from globally similar chains, this is not always the case. Remarkably, in some interface clusters, although the interfaces are similar, the overall structures and functions of the chains are different. Hence, our work suggests that different folds may combinatorially assemble to yield similar local interface motifs. The preference of different folds to associate in similar ways illustrates that the paradigm is universal, whether for single chains in folding or for protein-protein association in binding. We analyze and compare the two types of clusters. Type I, with similar interfaces, similar global structures and similar functions, is better packed, less planar, has larger total and non-polar buried surface areas, better complementarity and more backbone-backbone hydrogen bonds than Type II (similar interfaces, different global structures and different functions). The dataset clusters may provide rich data for protein-protein recognition, cellular networks and drug design. In particular, they should be useful in addressing the difficult question of what the favorable ways for proteins to interact are.

Tandem LIM domains provide synergistic binding in the LMO4:Ldb1 complex.

Nuclear LIM-only (LMO) and LIM-homeodomain (LIM-HD) proteins have important roles in cell fate determination, organ development and oncogenesis. These proteins contain tandemly arrayed LIM domains that bind the LIM interaction domain (LID) of the nuclear adaptor protein LIM domain-binding protein-1 (Ldb1). We have determined a high-resolution X-ray crystal structure of LMO4, a putative breast oncoprotein, in complex with Ldb1-LID, providing the first example of a tandem LIM:Ldb1-LID complex and the first structure of a type-B LIM domain. The complex possesses a highly modular structure with Ldb1-LID binding in an extended manner across both LIM domains of LMO4. The interface contains extensive hydrophobic and electrostatic interactions and multiple backbone-backbone hydrogen bonds. A mutagenic screen of Ldb1-LID, assessed by yeast two-hybrid and competition ELISA analysis, identified key features at the interface and revealed that the interaction is tolerant to mutation. These combined properties provide a mechanism for the binding of Ldb1 to numerous LMO and LIM-HD proteins. Furthermore, the modular extended interface may form a general mode of binding to tandem LIM domains.

N-acetyl-L-aspartic acid-N'-methylamide with side-chain orientation capable of external hydrogen bonding

In this study, we generated and analyzed the side-chain conformational potential energy hypersurfaces for each of the nine possible backbone conformers for N-acetyl-L-aspartic acid-N' methylamide. We found a total of 27 out of the 81 possible conformers optimized at the B3LYP/6-31G(d) level of theory. The relative energies, as well as the stabilization energies exerted by the side-chain on the backbone, have been calculated for each of the 27 optimized conformers at this level of theory. Various backbone-backbone (N-H . . . O=C) and backbone-side-chain (N-H . . . O=C; N-H . . . OH) hydrogen bonds were analyzed. The appearance of the notoriously absent backbone conformer may be attributed to such side-chain-backbone (SC/BB) and backbone-backbone (BB/BB) hydrogen bonds.

NMR structure of Escherichia coli glutaredoxin 3-glutathione mixed disulfide complex: implications for the enzymatic mechanism

Glutaredoxins (Grxs) catalyze reversible oxidation/reduction of protein disulfide groups and glutathione-containing mixed disulfide groups via an active site Grx-glutathione mixed disulfide (Grx-SG) intermediate. The NMR solution structure of the Escherichia coli Grx3 mixed disulfide with glutathione (Grx3-SG) was determined using a C14S mutant which traps this intermediate in the redox reaction. The structure contains a thioredoxin fold, with a well-defined binding site for glutathione which involves two intermolecular backbone-backbone hydrogen bonds forming an antiparallel intermolecular β-bridge between the protein and glutathione. The solution structure of E. coli Grx3-SG also suggests a binding site for a second glutathione in the reduction of the Grx3-SG intermediate, which is consistent with the specificity of reduction observed in Grxs. Molecular details of the structure in relation to the stability of the intermediate and the activity of Grx3 as a reductant of glutathione mixed disulfide groups are discussed. A comparison of glutathione binding in Grx3-SG and ligand binding in other members of the thioredoxin superfamily is presented, which illustrates the highly conserved intermolecular interactions in this protein family.