Conservation analysis and structure prediction of the SH2 family of phosphotyrosine binding domains

Modulation of potassium channels by protein tyrosine kinase inhibitors

Exposure of non-excitatory cells to the tyrosine kinase (PTK) inhibitors, genistein, herbimycin A, and tyrphostin, induced at least two families of K+ currents. The first, a TEA-insensitive slow-inactivating K+ current, is induced within 3 min following treatment with 140 mM genistein or 100 nM herbimycin A. The second current, a TEA-sensitive delayed rectifier, is induced within 30 min following treatment with 50 mM genistein or 10 nM herbimycin A. Currents with similar biophysical and pharmacological characteristics are induced in these cells following exposure to ionizing radiation. The radiation-induced currents are inhibited by pretreatment with the free radical scavenger, N-Acetyl L-Cysteine, or by pretreatment with the protein kinase C inhibitor, staurosporine; those induced by PTK inhibitors are not. The latter, therefore, do not appear to be mediated through free radicals or require serine/threonine phosphorylation for activation. Once the channels are activated by the PTK inhibitors, phosphorylation of the channel at serine/threonine residues results in slower inactivation of the induced current. We propose that protein tyrosine phosphorylation of the K+ channel protein itself or of a factor that interacts with it maintains the K+ channels of non-excitatory cells in a closed state. Following exposure to ionizing radiation, free radical-induced activation of serine/threonine kinase(s) results in phosphorylation of the channel and/or inactivation of a tyrosine kinase that in turn leads to activation of the K+ channels.

Secondary structure prediction from multiple sequence data: blood clotting factor XIII and Yersinia protein-tyrosine phosphatase

Predictions of protein structure are best tested without prior knowledge of the protein three-dimensional structure. Three-dimensional atomic models will soon be determined by X-ray crystallography for the alpha-subunit of human blood clotting factor XIII and members of the family of protein tyrosine specific phosphatases. Accordingly, we here present secondary structure predictions for each of these proteins. The secondary structure predictions were generated from aligned sets of protein sequences. This technique has previously provided reliable predictions for the Annexins and the SH2 domains. The factor XIII alpha prediction contains 39 regions predicted in strand conformation (34% of the protein) with only 3 helices (4%). The protein tyrosine phosphatases have 12 predicted strands and 5 helices (30 and 17%, respectively). We expect greater reliability from regions of alignments that show clear patterns of residue conservation (61% of factor XIII alpha and 57% of the protein tyrosine phosphatases). The aligned protein tyrosine phosphatases show two regions (L39-L80 and I138-E253) with clear patterns of residue conservation separated by a region of variable amino acid composition. We suggest this indicates that the tyrosine phosphatase fold comprises two domains separated by an exposed linker. Potential phosphate binding sites are identified in the protein tyrosine phosphatases.

A prediction of the secondary structure of the pleckstrin homology domain

A consensus prediction for the secondary structure of the pleckstrin homology (PH) domain is presented. The prediction is based on an analysis of patterns of conservation and variation of homologous protein sequences. The structure is predicted to be formed largely from beta strands with a single alpha helix.

Retrospective analysis of a secondary structure prediction: the catalytic domain of matrix metalloproteinases

Secondary structure prediction of the catalytic domain of matrix metalloproteinases is evaluated in the light of recently published experimentally determined structures. The prediction was made by combining conformational propensity, surface probability, and residue conservation calculated for an alignment of 19 sequences. The position of each observed secondary structure element was correctly predicted with a high degree of accuracy, with a single beta-strand falsely predicted. The domain fold was also anticipated from the prediction by analogy with the structural elements found in the distantly related metalloproteinases thermolysin, astacin, and adamalysin.

Combining evolutionary information and neural networks to predict protein secondary structure

Using evolutionary information contained in multiple sequence alignments as input to neural networks, secondary structure can be predicted at significantly increased accuracy. Here, we extend our previous three-level system of neural networks by using additional input information derived from multiple alignments. Using a position-specific conservation weight as part of the input increases performance. Using the number of insertions and deletions reduces the tendency for overprediction and increases overall accuracy. Addition of the global amino acid content yields a further improvement, mainly in predicting structural class. The final network system has sustained overall accuracy of 71.6% in a multiple cross-validation test on 126 unique protein chains. A test on a new set of 124 recently solved protein structures that have no significant sequence similarity to the learning set confirms the high level of accuracy. The average cross-validated accuracy for all 250 sequence-unique chains is above 72%. Using various data sets, the method is compared to alternative prediction methods, some of which also use multiple alignments: the performance advantage of the network system is at least 6 percentage points in three-state accuracy. In addition, the network estimates secondary structure content from multiple sequence alignments about as well as circular dichroism spectroscopy on a single protein and classifies 75% of the 250 proteins correctly into one of four protein structural classes. Of particular practical importance is the definition of a position-specific reliability index. For 40% of all residues the method has a sustained three-state accuracy of 88%, as high as the overall average for homology modelling. A further strength of the method is greatly increased accuracy in predicting the placement of secondary structure segments.

Expanding the genetic lexicon: incorporating non-standard amino acids into proteins by ribosome-based synthesis

Only 20 amino acids are normally incorporated into proteins synthesized in living cells, and this has limited the structural range of proteins that can be prepared. New methods that allow the incorporation of amino acids that are not normally encoded by natural genes are being developed: these include reassigning functions within the existing genetic code, and expanding the genetic code by constructing additional, non-natural codons. Used in conjunction with recent major advances in understanding protein structure-function relationships, these approaches should extend the range of de novo protein designs that are possible.

Oncogenes, Protein Tyrosine Kinases, and Signal Transduction

Many oncogenes encode protein tyrosine kinases (PTKs). Oncogenic mutations of these genes invariably result in constitutive activation of these PTKs. Autophosphorylation of the PTKs and tyrosine phosphorylation of their cellular substrates are essential events for transmission of the mitogenic signal into cells. The recent discovery of the characteristic amino acid sequences, of the src homology domains 2 and 3 (SH2 and SH3), and extensive studies on proteins containing the SH2 and SH3 domains have revealed that protein tyrosine-phosphorylation of PTKs provides phosphotyrosine sites for SH2 binding and allows extracellular signals to be relayed into the nucleus through a chain of protein-protein interactions mediated by the SH2 and SH3 domains. Studies on oncogenes, PTKs and SH2/SH3-containing proteins have made a tremendous contribution to our understanding of the mechanisms for the control of cell growth, oncogenesis, and signal transduction. This review is intended to provide an outline of the most recent progress in the study of signal transduction by PTKs. Copyright 1994 S. Karger AG, Basel

Conservation analysis and structure prediction of the protein serine/threonine phosphatases. Sequence similarity with diadenosine tetraphosphatase from Escherichia coli suggests homology to the protein phosphatases

A multiple sequence alignment of 44 serine/threonine-specific protein phosphatases has been performed. This reveals the position of a common conserved catalytic core, the location of invariant residues, insertions and deletions. The multiple alignment has been used to guide and improve a consensus secondary-structure prediction for the common catalytic core. The location of insertions and deletions has aided in defining the positions of surface loops and turns. The prediction suggests that the core protein phosphatase structure comprises two domains: the first has a single, beta sheet flanked by alpha helices, while the second is predominantly alpha helical. Knowledge of the core secondary structures provides a guide for the design of site-directed-mutagenesis experiments that will not disrupt the native phosphatase fold. A sequence similarity between eukaryotic serine/threonine protein phosphatases and the Escherichia coli diadenosine tetraphosphatase has been identified. This extends over the N-terminal 100 residues of bacteriophage phosphatases and E. coli diadenosine tetraphosphatase. Residues which are invariant amongst these classes are likely to be important in catalysis and protein folding. These include Arg92, Asn138, Asp59, Asp88, Gly58, Gly62, Gly87, Gly93, Gly137, His61, His139 and Val90 and fall into three clusters with the consensus sequences GD(IVTL)HG, GD(LYF)V(DA)RG and GNH, where brackets surround alternative amino acids. The first two consensus sequences are predicted to fall in the beta-alpha and beta-beta loops of a beta-alpha-beta-beta secondary-structure motif. This places the predicted phosphate-binding site at the N-terminus of the alpha helix, where phosphate binding may be stabilised by the alpha-helix dipole.

Comparison of calcium-dependent conformational changes in the N-terminal SH2 domains of p85 and GAP defines distinct properties for SH2 domains

Src-homology region 2 (SH2) domains are stretches of about 100 amino acids which are found to be structurally conserved in a number of signaling molecules. These regions have been shown to bind with high affinity to phosphotyrosine residues within activated receptor tyrosine kinases. Here we report the bacterial expression and purification of individual N-terminal SH2 (NSH2) domains of phosphatidylinositol 3-kinase (PI-3K) binding subunit (p85) and Ras GTPase activating protein (GAP) in amounts suitable for structure-function studies. The p85NSH2 domain stains dark purple and absorbs around 620-640 nm with Stains-all, a dye known to bind to calcium binding proteins. This effect was not observed for the GAPNSH2 domain. Circular dichroism analysis of the N-terminal SH2 domain of these proteins shows that p85NSH2, but not GAPNSH2, undergoes a significant dose-dependent change in conformation in the presence of increasing calcium concentrations. Moreover, the conformational change of p85NSH2 induced by calcium could be replicated by addition of a phosphorylated hexapeptide (DYpMDMK) representing the alpha-PDGFR binding site for p85. Limited proteolysis studies showed a significant calcium-dependent increase in protection of p85NSH2 but not GAPNSH2 from degradation by subtilisin. Our results further indicate that holmium, a trivalent lanthanide ion, which has been previously shown to substitute for calcium, could also protect the p85NSH2 domain from proteolysis even at 10-fold lower concentrations. In vitro binding studies using purified preparations of activated alpha-PDGFR show that calcium did not affect the binding of GAPNSH2 domains to activated alpha-PDGFR.(ABSTRACT TRUNCATED AT 250 WORDS)

How frequent are correlated changes in families of protein sequences?

A loss-of-function point mutation in a protein is often rescued by an additional mutation that compensates for the original physical change. According to one hypothesis, such compensation would be most effective in maintaining a structural motif if the two mutated residues were spatial neighbors. If this hypothesis were correct, one would expect that many such compensatory mutations have occurred during evolution and that present-day protein families show some degree of correlation in the occurrence of amino acid residues at positions whose side chains are in contact. Here, a statistical theory is presented which allows evaluation of correlations in a family of aligned protein sequences by assigning a scalar metric (such as charge or side-chain volume) to each type of amino acid and calculating correlation coefficients of these quantities at different positions. For the family of myoglobins it is found that there is a high correlation between fluctuations in neighboring charges. The correlation is close to what would be expected for total conservation of local charge. For the metric side-chain volume, on the other hand, no correlation could be found.

Predicting the conformation of proteins from sequences. Progress and future progress

A new paradigm for predicting the secondary and tertiary structure of functional proteins from sequence data has emerged from detailed models of how natural selection, conservation, and neutral drift, the three fundamental factors in molecular evolution, leave their mark upon protein sequences. Structural information is extracted from a set of aligned homologous sequences via an analysis of patterns of conservation and variation between proteins with quantitatively defined evolutionary relationships. Tertiary structural information is obtained prior to the assignment of secondary structure, where it plays an important role. Throughout, structural predictions are made with the active involvement of a biochemist whose expertise and insight is critical both for making the prediction and in analyzing its successful and unsuccessful parts. Secondary structure predictions are evaluated based on their ability to sustain an effort to model tertiary structure. Several predictions made using the new paradigm can now be compared with those made under the classical paradigm, including a neural network. The results obtained from the new paradigm are clearly superior to those obtained with the classical paradigm, at least within the protein families that were examined.

Origins of structural diversity within sequentially identical hexapeptides

Efforts to predict protein secondary structure have been hampered by the apparent structural plasticity of local amino acid sequences. Kabsch and Sander (1984, Proc. Natl. Acad. Sci. USA 81, 1075-1078) articulated this problem by demonstrating that identical pentapeptide sequences can adopt distinct structures in different proteins. With the increased size of the protein structure database and the availability of new methods to characterize structural environments, we revisit this observation of structural plasticity. Within a set of proteins with less than 50% sequence identity, 59 pairs of identical hexapeptide sequences were identified. These local structures were compared and their surrounding structural environments examined. Within a protein structural class (alpha/alpha, beta/beta, alpha/beta, alpha + beta), the structural similarity of sequentially identical hexapeptides usually is preserved. This study finds eight pairs of identical hexapeptide sequences that adopt beta-strand structure in one protein and alpha-helical structure in the other. In none of the eight cases do the members of these sequences pairs come from proteins within the same folding class. These results have implications for class dependent secondary structure prediction algorithms.

The PH domain: a common piece in the structural patchwork of signalling proteins

The 'pleckstrin homology' domain is an approximately 100-residue protein module that has recently been added to the domain catalogue of signalling proteins. For this review we have made an extensive database search using a profile search method, and found a number of additional proteins that may contain PH domains. The PH domain is present in many kinases, isoforms of phospholipase C, GTPases, GTPase-activating proteins and nucleotide-exchange factors, including such proteins as Vav, Dbl and Bcr, and there are two PH domains in a guanine-nucleotide releasing factor of Ras. Many PH-domain-containing proteins interact with GTP-binding proteins. We have also identified a PH domain in beta-adrenergic receptor kinase exactly in the region that has already been shown to be involved in binding to the beta and gamma subunits of a heterotrimeric G protein. This suggests that PH domains may be involved in interactions with GTP-binding proteins.

Improved prediction of protein secondary structure by use of sequence profiles and neural networks

The explosive accumulation of protein sequences in the wake of large-scale sequencing projects is in stark contrast to the much slower experimental determination of protein structures. Improved methods of structure prediction from the gene sequence alone are therefore needed. Here, we report a substantial increase in both the accuracy and quality of secondary-structure predictions, using a neural-network algorithm. The main improvements come from the use of multiple sequence alignments (better overall accuracy), from "balanced training" (better prediction of beta-strands), and from "structure context training" (better prediction of helix and strand lengths). This method, cross-validated on seven different test sets purged of sequence similarity to learning sets, achieves a three-state prediction accuracy of 69.7%, significantly better than previous methods. In addition, the predicted structures have a more realistic distribution of helix and strand segments. The predictions may be suitable for use in practice as a first estimate of the structural type of newly sequenced proteins.

Predicting the conformation of proteins. Man versus machine

Two types of approaches for predicting the conformation of proteins from sequence data have lately received attention: 'black box' tools that generate fully automated predictions of secondary structure from a set of homologous protein sequences, and methods involving the expertise of a human biochemist who is assisted, but not replaced, by computer tools. A friendly controversy has emerged as to which approach offers a brighter future. In fact, both are necessary. Nevertheless, a snapshot of the controversy at this instant offers much insight into the structure prediction problem itself.

Progress in protein structure prediction?

Prediction of protein secondary structure is an old problem and progress has been slow. Recently, spectacular success has been claimed in the blind prediction of the catalytic subunit of the cAMP-dependent protein kinase. When predictions in this and other test cases are assessed critically, some claims of prediction success turn out to be exaggerated, but a kernel of real progress remains: protein structure prediction can be improved substantially when a family of related sequences is available. Enough so that molecular biologists equipped with a new amino acid sequence and a multiple sequence alignment in hand may be tempted to test the new prediction methods.

Recognition of a high-affinity phosphotyrosyl peptide by the Src homology-2 domain of p56lck

The Src homology-2 (SH2) domains are modules of about 100 amino-acid residues that are found in many intracellular signal-transduction proteins. They bind phosphotyrosine-containing sequences with high affinity and specificity, recognizing phosphotyrosine in the context of the immediately adjacent polypeptide sequence. The protein p56lck (Lck) is a Src-like, lymphocyte-specific tyrosine kinase. A phosphopeptide library screen has recently been used to deduce an 'optimal' binding sequence for the Lck SH2 domain. There is selectivity for the residues Glu, Glu and Ile in the three positions C-terminal to the phosphotyrosine. An 11-residue phosphopeptide derived from the hamster polyoma middle-T antigen, EPQpYEEIPIYL, binds with an approximately 1 nM dissociation constant to the Lck SH2 (ref. 17), an affinity equivalent to that of the tightest known SH2-phosphopeptide complex. We report here the high-resolution crystallographic analysis of the Lck SH2 domain in complex with this phosphopeptide. Recent crystallographically derived structures of the Src SH2 domain in complex with low-affinity peptides, which do not contain the EEI consensus, and NMR-derived structures of unliganded Abl (ref. 19) and p85 (ref. 20) SH2 domains have revealed the conserved fold of the SH2 domain and the properties of a phosphotyrosine binding pocket. Our high-affinity complex shows the presence of a second pocket for the residue (pY + 3) three positions C-terminal to the phosphotyrosine (pY). The peptide is anchored by insertion of the pY and pY + 3 side chains into their pockets and by a network of hydrogen bonds to the peptide main chain. In the low-affinity phosphopeptide/Src complexes, the pY + 3 residues do not insert into the homologous binding pocket and the peptide main chain remains displaced from the surface of the domain.

Crystal structure of a Src-homology 3 (SH3) domain

The Src-homologous SH3 domain is a small domain present in a large number of proteins that are involved in signal transduction, such as the Src protein tyrosine kinase, or in membrane-cytoskeleton interactions, but the function of SH3 is still unknown (reviewed in refs 1-3). Here we report the three-dimensional structure at 1.8 A resolution of the SH3 domain of the cytoskeletal protein spectrin expressed in Escherichia coli. The domain is a compact beta-barrel made of five antiparallel beta-strands. The amino acids that are conserved in the SH3 sequences are located close to each other on one side of the molecule. This surface is rich in aromatic and carboxylic amino acids, and is distal to the region of the molecule where the N and C termini reside and where SH3 inserts into the alpha-spectrin chain. We suggest that a protein ligand binds to this conserved surface of SH3.

Crystal structure of the phosphotyrosine recognition domain SH2 of v-src complexed with tyrosine-phosphorylated peptides

Three-dimensional structures of complexes of the SH2 domain of the v-src oncogene product with two phosphotyrosyl peptides have been determined by X-ray crystallography at resolutions of 1.5 and 2.0 A, respectively. A central antiparallel beta-sheet in the structure is flanked by two alpha-helices, with peptide binding mediated by the sheet, intervening loops and one of the helices. The specific recognition of phosphotyrosine involves amino-aromatic interactions between lysine and arginine side chains and the ring system in addition to hydrogen-bonding interactions with the phosphate.

Structure of an SH2 domain of the p85 alpha subunit of phosphatidylinositol-3-OH kinase

Receptor protein-tyrosine kinases, through phosphorylation of specific tyrosine residues, generate high-affinity binding sites which direct assembly of multienzyme signalling complexes. Many of these signalling proteins, including phospholipase C gamma, GTPase-activating protein and phosphatidylinositol-3-OH kinase, contain src-homology 2 (SH2) domains, which bind with high affinity and specificity to tyrosine-phosphorylated sequences. The critical role played by SH2 domains in signalling has been highlighted by recent studies showing that mutation of specific phosphorylation sites on the platelet-derived growth factor receptor impair its association with phosphatidylinositol-3-OH kinase, preventing growth factor-induced mitogenesis. Here we report the solution structure of an isolated SH2 domain from the 85K regulatory subunit of phosphatidylinositol-3-OH kinase, determined using multidimensional nuclear magnetic resonance spectroscopy. The structure is characterized by a central region of beta-sheet flanked by two alpha-helices, with a highly flexible loop close to functionally important residues previously identified by site-directed mutagenesis.