Pleckstrin homology (PH) domains and phosphoinositides

PH (pleckstrin homology) domains represent the 11th most common domain in the human proteome. They are best known for their ability to bind phosphoinositides with high affinity and specificity, although it is now clear that less than 10% of all PH domains share this property. Cases in which PH domains bind specific phosphoinositides with high affinity are restricted to those phosphoinositides that have a pair of adjacent phosphates in their inositol headgroup. Those that do not [PtdIns3P, PtdIns5P and PtdIns(3,5)P2] are instead recognized by distinct classes of domains including FYVE domains, PX (phox homology) domains, PHD (plant homeodomain) fingers and the recently identified PROPPINs (b-propellers that bind polyphosphoinositides). Of the 90% of PH domains that do not bind strongly and specifically to phosphoinositides, few are well understood. One group of PH domains appears to bind both phosphoinositides (with little specificity) and Arf (ADP-ribosylation factor) family small G-proteins, and are targeted to the Golgi apparatus where both phosphoinositides and the relevant Arfs are both present. Here, the PH domains may function as coincidence detectors. A central challenge in understanding the majority of PH domains is to establish whether the very low affinity phosphoinositide binding reported in many cases has any functional relevance. For PH domains from dynamin and from Dbl family proteins, this weak binding does appear to be functionally important, although its precise mechanistic role is unclear. In many other cases, it is quite likely that alternative binding partners are more relevant, and that the observed PH domain homology represents conservation of structural fold rather than function.     

Binding of copper(I) by the Wilson disease protein and its copper chaperone

The Wilson disease protein (WND) is a transport ATPase involved in copper delivery to the secretory pathway. Mutations in WND and its homolog, the Menkes protein, lead to genetic disorders of copper metabolism. The WND and Menkes proteins are distinguished from other P-type ATPases by the presence of six soluble N-terminal metal-binding domains containing a conserved CXXC metal-binding motif. The exact roles of these domains are not well established, but possible functions include exchanging copper with the metallochaperone Atox1 and mediating copper-responsive cellular relocalization. Although all six domains can bind copper, genetic and biochemical studies indicate that the domains are not functionally equivalent. One way the domains could be tuned to perform different functions is by having different affinities for Cu(I). We have used isothermal titration calorimetry to measure the association constant (K(a)) and stoichiometry (n) values of Cu(I) binding to the WND metal-binding domains and to their metallochaperone Atox1. The association constants for both the chaperone and target domains are approximately 10(5) to 10(6) m(-1), suggesting that the handling of copper by Atox1 and copper transfer between Atox1 and WND are under kinetic rather than thermodynamic control. Although some differences in both n and K(a) values are observed for variant proteins containing less than the full complement of six metal-binding domains, the data for domains 1-6 were best fitted with a single site model. Thus, the individual functions of the six WND metal-binding domains are not conferred by different Cu(I) affinities but instead by fold and electrostatic surface properties.     

The GYF domain

GYF domains are small, versatile adaptor domains that recognize proline-rich sequences (PRS). They are present in most eukaryotic species sequenced so far, but in contrast to other PRS-recognition domains (PRD), GYF domains have not experienced the same amplification in metazoa during evolution. Mutational and structural analysis has shown the conserved signature W-X-Y-X(6-11)-GPF-X(4)-M-X(2)-W-X(3)-GYF to be the site of interaction with proline-rich peptides. In contrast, composition and length of the C-terminal half of GYF domains are not conserved. Similar to other PRD, GYF domains bind to many different PRS that converge on a minimal consensus sequence. All GYF domains analyzed so far selected for the core motif PPG, whereas amino-acid preferences adjacent to this motif vary. As a result of this analysis, two subfamilies have been identified: CD2BP2-type and SMY2-type GYF domains. The latter subfamily comprises most GYF domains and is characterized by a shorter beta(1)-beta(2) loop and an aspartate instead of the tryptophan found at position 8 in CD2BP2-type GYF domains. Recent analysis of binding specificities for GYF domains allowed identification of novel interaction partners. Thereby proteomics has contributed to a functional understanding of GYF domain-containing proteins and sets the stage for a more systematic investigation of their functions in vivo.     

Comprehensive analysis of the factors contributing to the stability and solubility of autonomous human VH domains

We report a comprehensive analysis of sequence features that allow for the production of autonomous human heavy chain variable (V(H)) domains that are stable and soluble in the absence of a light chain partner. Using combinatorial phage-displayed libraries and conventional biophysical methods, we analyzed the entire former light chain interface and the third complementarity determining region (CDR3). Unlike the monomeric variable domains of camelid heavy chain antibodies (V(H)H domains), in which autonomous behavior depends on interactions between the hydrophobic former light chain interface and CDR3, we find that the stability of many in vitro evolved V(H) domains is essentially independent of the CDR3 sequence and instead derives from mutations that increase the hydrophilicity of the former light chain interface by replacing exposed hydrophobic residues with structurally compatible hydrophilic substitutions. The engineered domains can be expressed recombinantly at high yield, are predominantly monomeric at high concentrations, unfold reversibly, and are even more thermostable than typical camelid V(H)H domains. Many of the stabilizing mutations are rare in natural V(H) and V(H)H domains and thus could not be predicted by studying natural sequences and structures. The results demonstrate that autonomous V(H) domains with structural properties beyond the scope of natural frameworks can be derived by using non-natural mutations, which differ from those found in camelid V(H)H domains. These findings should enable the development of libraries of synthetic V(H) domains with CDR3 diversities unconstrained by structural demands.     

Coevolution of the domains of cytoplasmic tyrosine kinases

Many signaling molecules are multidomain proteins that have other domains in addition to the catalytic kinase domain. Protein tyrosine kinases almost without exception contain Src homology 2 (SH2) and/or SH3 domains that can interact with other signaling proteins. Here, we studied evolution of the tyrosine kinases containing SH2 and/or SH3 and kinase domains. The three domains seem to have duplicated together, since the phylogenetic analysis using parsimony gave almost identical evolutionary trees for the separate domains and the multidomain complexes. The congruence analysis of the sequences for the separate domains also suggested that the domains have coevolved. There are several reasons for the domains to appear in a cluster. Kinases are regulated in many ways, and the presence of SH2 and SH3 domains at proper positions is crucial. Because all three domains can recognize different parts of ligands and substrates, their evolution has been interconnected. The reasons for the clustering and coevolution of the three domains in protein tyrosine kinases (PTKs) are discussed.     

The geometry of domain combination in proteins.

Most proteins in genomes are the result of the recombination of two or more domains. It has been found that if proteins are formed by a combination of domains from superfamilies A and B, then the domains may occur in the sequential order AB or BA but only in about 2% of cases do they occur in both sequential orders. The classical Rossmann domains of known structure are combined with catalytic domains from seven different superfamilies. In addition, there are eight cases where structures with both AB and BA domain combinations are known. For these two sets of structures, we analysed: (i) the relative orientation of the domains; (ii) the type of domain connection; (iii) the structure of the interdomain links; and (iv) domain function. The results of this analysis indicate that in most cases domain order is conserved because recombination of the domains has only occurred once during the course of evolution. Functional reasons become important when the domain connections are short. In seven out of the eight known cases where domains are combined in the AB and BA sequential orders they have different geometrical relationships that give them different functional properties.     

Domains in the fibrinogen molecule

Intramolecular melting of fibrinogen and its degradation products has been studied by a scanning microcalorimetric method in various solution environments (especially variations in pH), and inferences are made about the features that seemed to be independently folding segments (“domains”), as evidenced by their independent resistance to thermal denaturation. It was shown that there are 12 more or less independent co-operative regions of ordered compact structure in fibrinogen, which can be considered as structural domains of this macromolecule. Of these 12 domains, two are in the central part of the molecule, corresponding to the E fragment, four are in each terminal part, corresponding to the D fragments, and two are formed by the carboxy-terminal portions of the α-chains. All fibrinogen domains can be divided into two groups according to their thermodynamic properties: (1) thermolabile domains, to which belong three domains from each terminal part of the molecule and the domains formed by the carboxy-terminal portions of the α-chains; (2) thermostable domains, to which belong both domains from the central part and one domain from each terminal part of the molecule. This division seems to reflect the structural differences between the domains.     

Multiple roles of pleckstrin homology domains in phospholipase Cbeta function

Since their discovery almost 10 years ago pleckstrin homology (PH) domains have been identified in a wide variety of proteins. Here, we focus on two proteins whose PH domains play a defined functional role, phospholipase C (PLC)-beta(2) and PLCdelta(1). While the PH domains of both proteins are responsible for membrane targeting, their specificity of membrane binding drastically differs. However, in both these proteins the PH domains work to modulate the activity of their catalytic core upon interaction with either phosphoinositol lipids or G protein activators. These observations show that these PH domains are not simply binding sites tethered onto their host enzyme but are intimately associated with their catalytic core. This property may be true for other PH domains.     

Computer modeling of the membrane interaction of FYVE domains

FYVE domains are membrane targeting domains that are found in proteins involved in endosomal trafficking and signal transduction pathways. Most FYVE domains bind specifically to phosphatidylinositol 3-phosphate (PI(3)P), a lipid that resides mainly in endosomal membranes. Though the specific interactions between FYVE domains and the headgroup of PI(3)P have been well characterized, principally through structural studies, the available experimental structures suggest several different models for FYVE/membrane association. Thus, the manner in which FYVE domains adsorb to the membrane surface remains to be elucidated. Towards this end, recent experiments have shown that FYVE domains bind PI(3)P in the context of phospholipid bilayers and that hydrophobic residues on a conserved loop are able to penetrate the membrane interface in a PI(3)P-dependent manner.Here, the finite difference Poisson-Boltzmann (FDPB) method has been used to calculate the energetic interactions of FYVE domains with phospholipid membranes. Based on the computational analysis, it is found that (1) recruitment to membranes is facilitated by non-specific electrostatic interactions that occur between basic residues on the domains and acidic phospholipids in the membrane, (2) the energetic analysis can quantitatively differentiate among the modes of membrane association proposed by the experimentally determined structures, (3) FDPB calculations predict energetically feasible models for the membrane-associated states of FYVE domains, (4) these models are consistent with the observation that conserved hydrophobic residues insert into the membrane interface, and (5) the calculations provide a molecular model for the hydrophobic partitioning: binding of PI(3)P significantly neutralizes positive potential in the region of the hydrophobic residues, which acts as an "electrostatic switch" by reducing the energetic barrier for membrane penetration. Finally, the computational results are extended to FYVE domains of unknown structure through the construction of high quality homology models for human FYVE sequences.     

C1 domains exposed: from diacylglycerol binding to protein-protein interactions

C1 domains, cysteine-rich modules originally identified in protein kinase C (PKC) isozymes, are present in multiple signaling families, including PKDs, chimaerins, RasGRPs, diacylglycerol kinases (DGKs) and others. Typical C1 domains bind the lipid second messenger diacylglycerol (DAG) and DAG-mimetics such as phorbol esters, and are critical for governing association to membranes. On the contrary, atypical C1 domains possess structural determinants that impede phorbol ester/DAG binding. C1 domains are generally expressed as twin modules (C1A and C1B) or single domains. Biochemical and cellular studies in PKC and PKD isozymes revealed that C1A and C1B domains are non-equivalent as lipid-binding motifs or translocation modules. It has been recently determined that individual C1 domains have unique patterns of ligand recognition, driven in some cases by subtle structural differences. Insights from recent 3-D studies on beta2-chimaerin and Munc13-1 revealed that their single C1 domains are sterically blocked by intramolecular interactions, suggesting that major conformational changes would be required for exposing the site of DAG interaction. Thus, it is clear that the protein context plays a major role in determining whether binding of DAG to the C1 domain would lead to enzyme activation or merely serves as an anchoring mechanism.     

Crystal structures of the two membrane-proximal Ig-like domains (D3D4) of LILRB1/B2: alternative models for their involvement in peptide-HLA binding

Leukocyte immunoglobulin-like receptors (LILRs), also called CD85s, ILTs, or LIRs, are important mediators of immune activation and tolerance that contain tandem immunoglobulin (Ig)-like folds. There are 11 (in addition to two pseudogenes) LILRs in total, two with two Ig-like domains (D1D2) and the remaining nine with four Ig-like domains (D1D2D3D4). Thus far, the structural features of the D1D2 domains of LILR proteins are well defi ned, but no structures for the D3D4 domains have been reported. This is a very important fi eld to be studied as it relates to the unknown functions of the D3D4 domains, as well as their relative orientation to the D1D2 domains on the cell surface. Here, we report the crystal structures of the D3D4 domains of both LILRB1 and LILRB2. The two Iglike domains of both LILRB1-D3D4 and LILRB2-D3D4 are arranged at an acute angle (~60°) to form a bent structure, resembling the structures of natural killer inhibitory receptors. Based on these two D3D4 domain structures and previously reported D1D2/HLA I complex structures, two alternative models of full-length (four Ig-like domains) LILR molecules bound to HLA I are proposed.     

The thermostabilizing domain of the modular xylanase XynA of Thermotoga maritima represents a novel type of binding domain with affinity for soluble xylan and mixed-linkage beta-1,3/beta-1, 4-glucan

Thermotoga maritima XynA is an extremely thermostable modular enzyme with five domains (A1-A2-B-C1-C2). Its catalytic domain (-B-) is flanked by duplicated non-catalytic domains. The C-terminal repeated domains represent cellulose-binding domains (CBDs). Xylanase domains related to the N-terminal domains of XynA (A1-A2) are called thermostabilizing domains because their deletion normally leads to increased thermosensitivity of the enzymes. It was found that a glutathione-S-transferase (GST) hybrid protein (GST-A1A2) containing both A-domains of XynA can interact with various soluble xylan preparations and with mixed-linkage beta-1,3/beta-1,4-glucans. GST-A1A2 showed no affinity for insoluble microcrystalline cellulose, whereas, vice versa, GST-C2, which contains the C-terminal CBD of XynA, did not interact with soluble xylan. Another hybrid protein, GST-A2, displayed the same binding properties as GST-A1A2, indicating that A2 alone can also promote xylan binding. The dissociation constants for the binding of xylose, xylobiose, xylotriose, xylotetraose and xylopentaose by GST-A2, as determined at 20 degrees C by fluorescence quench experiments, were 8.1 x 10(-3) M, 2.3 x 10(-4) M, 2.3 x 10(-5) M, 2.5 x 10(-6)M and 1.1 x 10(-6) M respectively. The A-domains of XynA, which are designated as xylan binding domains (XBD), are, from the structural as well as the functional point of view, prototypes of a novel class of binding domains. More than 50 related protein segments with hitherto unknown function were detected in about 30 other multidomain beta-glycanases, among them putative plant (Arabidopsis thaliana) xylanases. It is argued that polysaccharide binding and not thermostabilization is the main function of A-like domains.     

Autonomous Tetramerization Domains in the Glycan-binding Receptors DC-SIGN and DC-SIGNR

Multivalent binding of glycans on pathogens and on mammalian cells by the receptors DC-SIGN (CD209) and DC-SIGNR (L-SIGN, CD299) is dependent on correct disposition of the C-type carbohydrate-recognition domains projected at the C-terminal ends of necks at the cell surface. In the work reported here, neck domains of DC-SIGN and DC-SIGNR expressed in isolation are shown to form tetramers in the absence of the CRDs. Stability analysis indicates that interactions between the neck domains account fully for the stability of the tetrameric extracellular portions of the receptors. The neck domains are approximately 40% α-helical based on circular dichroism analysis. However, in contrast to other glycan-binding receptors in which fully helical neck regions are intimately associated with C-terminal C-type CRDs, the neck domains in DC-SIGN and DC-SIGNR act as autonomous tetramerization domains and the neck domains and CRDs are organized independently. Neck domains from polymorphic forms of DC-SIGNR that lack some of the repeat sequences show modestly reduced stability, but differences near the C-terminal end of the neck domains lead to significantly enhanced stability of DC-SIGNR tetramers compared to DC-SIGN.     

Structure of the split PH domain and distinct lipid-binding properties of the PH-PDZ supramodule of alpha-syntrophin

Pleckstrin homology (PH) domains play diverse roles in cytoskeletal dynamics and signal transduction. Split PH domains represent a unique subclass of PH domains that have been implicated in interactions with complementary partial PH domains 'hidden' in many proteins. Whether partial PH domains exist as independent structural units alone and whether two halves of a split PH domain can fold together to form an intact PH domain are not known. Here, we solved the structure of the PH(N)-PDZ-PH(C) tandem of alpha-syntrophin. The split PH domain of alpha-syntrophin adopts a canonical PH domain fold. The isolated partial PH domains of alpha-syntrophin, although completely unfolded, remain soluble in solution. Mixing of the two isolated domains induces de novo folding and yields a stable PH domain. Our results demonstrate that two complementary partial PH domains are capable of binding to each other to form an intact PH domain. We further showed that the PH(N)-PDZ-PH(C) tandem forms a functionally distinct supramodule, in which the split PH domain and the PDZ domain function synergistically in binding to inositol phospholipids.     

All phox homology (PX) domains from Saccharomyces cerevisiae specifically recognize phosphatidylinositol 3-phosphate

Phox homology (PX) domains are named for a 130-amino acid region of homology shared with part of two components of the phagocyte NADPH oxidase (phox) complex. They are found in proteins involved in vesicular trafficking, protein sorting, and lipid modification. It was recently reported that certain PX domains specifically recognize phosphatidylinositol 3-phosphate (PtdIns-3-P) and drive recruitment of their host proteins to the cytoplasmic leaflet of endosomal and/or vacuolar membranes where this phosphoinositide is enriched. We have analyzed phosphoinositide binding by all 15 PX domains encoded by the Saccharomyces cerevisiae genome. All yeast PX domains specifically recognize PtdIns-3-P in protein-lipid overlay experiments, with just one exception (a significant sequence outlier). In surface plasmon resonance studies, four of the yeast PX domains bind PtdIns-3-P with high (micromolar range) affinity. Although the remaining PX domains specifically recognize PtdIns-3-P, they bind this lipid with only low affinity. Interestingly, many proteins with "low affinity" PX domains are known to form large multimeric complexes, which may increase the overall avidity for membranes. Our results establish that PtdIns-3-P, and not other phosphoinositides, is the target of all PX domains in S. cerevisiae and suggest a role for PX domains in assembly of multiprotein complexes at specific membrane surfaces.     

Utility of epimerization domains for the redesign of nonribosomal peptide synthetases

Many pharmacologically important agents are assembled on multimodular nonribosomal peptide synthetases (NRPSs) whose modules comprise a set of core domains with all essential catalytic functions necessary for the incorporation and modification of one building block. Very often, d-amino acids are found in such products which, with few exceptions, are generated by the action of NRPS integrated epimerization (E) domains that alter the stereochemistry of the corresponding peptidyl carrier protein (PCP) bound l-intermediate. In this study we present a quantitative investigation of substrate specificity of four different E domains (two 'peptidyl-' and two 'aminoacyl-'E domains) derived from different NRPSs towards PCP bound peptides. The respective PCP-E bidomain apo-proteins (TycB(3)-, FenD(2)-, TycA- and GrsA-PCP-E) were primed with various peptidyl-CoA precursors by utilizing the promiscuous phosphopantetheinyl transferase Sfp. PCP bound peptidyl-S-Ppant epimerization products were chemically cleaved and analyzed for their l/d-ratios by LCMS. We were able to show that all four E domains tolerate a broad variety of peptidyl-S-Ppant-substrates as evaluated by k(obs) values and final l/d-product equilibria determined for each reaction. The two C-terminal amino acids of the substrate seem to be recognized by 'peptidyl-'E domains. Interestingly, the 'aminoacyl-'E domains GrsA- and TycA-E were also able to convert the elongated intermediates. All four E domains accepted an N-methylated precursor as well and epimerized this substrate with high efficiency. Finally, we could demonstrate that the condensation (C) domain of TycB(1) is also able to process peptidyl substrates transferred by TycA. In conclusion, these findings are of great impact on future engineering attempts.     

Structural classification of biotin carboxyl carrier proteins

We gathered primary and tertiary structures of acyl-CoA carboxylases from public databases, and established that members of their biotin carboxylase (BC) and biotin carboxyl carrier protein (BCCP) domains occur in one family each and that members of their carboxyl transferase (CT) domains occur in two families. Protein families have members similar in primary and tertiary structure that probably have descended from the same protein ancestor. The BCCP domains complexed with biotin in acyl and acyl-CoA carboxylases transfer bicarbonate ions from BC domains to CT domains, enabling the latter to carboxylate acyl and acyl-CoA moieties. We separated the BCCP domains into four subfamilies based on more subtle primary structure differences. Members of different BCCP subfamilies often are produced by different types of organisms and are associated with different carboxylases.     

Structure and plasticity of Endophilin and Sorting Nexin 9

Endophilin and Sorting Nexin 9 (Snx9) play key roles in endocytosis by membrane curvature sensing and remodeling via their Bin/Amphiphysin/Rvs (BAR) domains. BAR and the related F-BAR domains form dimeric, crescent-shaped units that occur N- or C-terminally to other lipid-binding, adaptor, or catalytic modules. In crystal structures, the PX-BAR unit of Snx9 (Snx9(PX-BAR)) adopts an overall compact, moderately curved conformation. SAXS-based solution studies revealed an alternative, more curved state of Snx9(PX-BAR) in which the PX domains are flexibly connected to the BAR domains, providing a model for how Snx9 exhibits lipid-dependent curvature preferences. In contrast, Endophilin appears to be rigid in solution, and the SH3 domains are located at the distal tips of a BAR domain dimer with fixed curvature. We also observed tip-to-tip interactions between the BAR domains in a trigonal crystal form of Snx9(PX-BAR) reminiscent of functionally important interactions described for F-BAR domains.     

Hair-forming activity of human lymphocyte specific protein 1 requires cooperation between its caldesmon-like domains and the villin headpiece-like domains.

LSP1 is an F-actin binding with multiple F-actin binding domains. Overexpression of LSP1 in NAD 47/89 patient's neutrophils created hair-like projections on the patient's neutrophil cell surfaces and inhibited neutrophil cell motility and transfection of LSP1 in serial cell lines recreate the NAD 47/89 phenotype and produce branching hair-like surface projections. Although LSP1 contains hair-forming ability and LSP1 F-actin binding domains have been defined, the LSP1 domains responsible for its hair-forming activity, the relationship to the F-actin binding domains, and the required domain interactions, if any, for hair formation are not well understood. To define the hair-forming domains of LSP1, the relationship to the known F-actin binding domains, and binding domain interactions, LSP1 truncates, which include or exclude the different F-actin binding domains, were created by PCR. LSP1 mutants were created by site-directed mutagenesis to define the amino acids important for hair formation. Sf9 cells were infected with recombinant baculovirus expressing the cDNA of LSP1 truncates and mutants, and the morphology of infected Sf9 cells was documented by DIC optics. Results show that (1) the hair-forming activity of LSP1 is localized to the basic C-terminal half of the molecule, which contains all of the F-actin binding domains; (2) both the caldesmon-like domains and the villin headpiece-like domains are required for the hair-forming activity of LSP1; (3) basic amino acids in the villin headpiece regions are crucial for the hair-forming activity of LSP1 molecule. The results suggest cooperation between the caldesmon-like domains and the villin headpiece-like domains are required for the hair-forming activity of human LSP1 in cells.