Constructing ontology-driven protein family databases

MOTIVATION: Protein family databases provide a central focus for scientific communities as well as providing useful resources to aide research. However, such resources require constant curation and often become outdated and discontinued. We have developed an ontology-driven system for capturing and managing protein family data that addresses the problems of maintenance and sustainability. RESULTS: Using protein phosphatases and ABC transporters as model protein families, we constructed two protein family database resources around a central DAML+OIL ontology. Each resource contains specialist information about each protein family, providing specialized domain-specific resources based on the same template structure. The formal structure, combined with the extraction of biological data using GO terms, allows for automated update strategies. Despite the functional differences between the two protein families, the ontology model was equally applicable to both, demonstrating the generic nature of the system. AVAILABILITY: The protein phosphatase resource, PhosphaBase, is freely available on the internet (http://www.bioinf.man.ac.uk/phosphabase). The DAML+OIL ontology for the protein phosphatases and the ABC transporters is available on request from the authors. CONTACT: kwolstencroft@cs.man.ac.uk.     

Evolutionary radiation pattern of novel protein phosphatases revealed by analysis of protein data from the completely sequenced genomes of humans, green algae, and higher plants

In addition to the major serine/threonine-specific phosphoprotein phosphatase, Mg(2+)-dependent phosphoprotein phosphatase, and protein tyrosine phosphatase families, there are novel protein phosphatases, including enzymes with aspartic acid-based catalysis and subfamilies of protein tyrosine phosphatases, whose evolutionary history and representation in plants is poorly characterized. We have searched the protein data sets encoded by the well-finished nuclear genomes of the higher plants Arabidopsis (Arabidopsis thaliana) and Oryza sativa, and the latest draft data sets from the tree Populus trichocarpa and the green algae Chlamydomonas reinhardtii and Ostreococcus tauri, for homologs to several classes of novel protein phosphatases. The Arabidopsis proteins, in combination with previously published data, provide a complete inventory of known types of protein phosphatases in this organism. Phylogenetic analysis of these proteins reveals a pattern of evolution where a diverse set of protein phosphatases was present early in the history of eukaryotes, and the division of plant and animal evolution resulted in two distinct sets of protein phosphatases. The green algae occupy an intermediate position, and show similarity to both plants and animals, depending on the protein. Of specific interest are the lack of cell division cycle (CDC) phosphatases CDC25 and CDC14, and the seeming adaptation of CDC14 as a protein interaction domain in higher plants. In addition, there is a dramatic increase in proteins containing RNA polymerase C-terminal domain phosphatase-like catalytic domains in the higher plants. Expression analysis of Arabidopsis phosphatase genes differentially amplified in plants (specifically the C-terminal domain phosphatase-like phosphatases) shows patterns of tissue-specific expression with a statistically significant number of correlated genes encoding putative signal transduction proteins.     

Genome-wide survey of prokaryotic O-protein phosphatases

Complex and diverse signal transduction circuits are responsible for the efficient functioning of cellular network. Protein kinases and O-protein phosphatases are primarily responsible for propagating such stimuli within a eukaryotic cell. However, there is limited understanding of O-protein phosphatases in the prokaryotic genomes. The availability of complete genome sequence information for several prokaryotes permits a genome-wide survey of O-protein phosphatases. The distribution of the various protein phosphatase families has been observed to be mosaic, with the exception of the members of the phospho protein family P (PPP), which is consistent with previous studies. The PPP family is ubiquitous in the prokaryotic world and undergoes the highest sequence divergence within a genome amongst phosphatases studied. The co-occurrence of low molecular mass tyrosine phosphatase (LMWPc) and PPP domain in a single polypeptide suggests that the protein present in Archaeoglobus fulgidus might represent the progenitor for all protein phosphatases. The curation of data on prokaryotic protein phosphatases provides a convenient framework for the analysis of domain architectures and for characterising structural and functional properties of this important family of signalling proteins.     

Do metabolic HAD phosphatases moonlight as protein phosphatases?

Mammalian haloacid dehalogenase (HAD)-type phosphatases have evolved to dephosphorylate a wide range of small metabolites, but can also target macromolecules such as serine/threonine, tyrosine-, and histidine-phosphorylated proteins. To accomplish these tasks, HAD phosphatases are equipped with cap domains that control access to the active site and provide substrate specificity determinants. A number of capped HAD phosphatases impact protein phosphorylation, although structural data are consistent with small metabolite substrates rather than protein substrates. This review discusses the structures, functions and disease implications of the three closely related, capped HAD phosphatases pyridoxal phosphatase (PDXP or chronophin), phosphoglycolate phosphatase (PGP, also termed AUM or glycerol phosphatase) and phospholysine phosphohistidine inorganic pyrophosphate phosphatase (LHPP or HDHD2B). Evidence in support of small metabolite and protein phosphatase activity is discussed in the context of the diversity of their biological functions.     

Assay of phosphotyrosyl protein phosphatase using synthetic peptide 1142-1153 of the insulin receptor

Synthetic peptide 1142-1153 of the insulin receptor was phosphorylated on tyrosine by the insulin receptor and found to be a potent substrate for dephosphorylation by rat liver particulate and soluble phosphotyrosyl protein phosphatases. Apparent Km values were approximately 5 microM. Vm values (nmol phosphate removed/min per mg protein) were 0.62 (particulate) and 0.2 (soluble). This corresponds to 80% of total activity being membrane-associated, indicating that membrane-bound phosphatases are important receptor phosphatases. The phosphatase activities were distinct from acid and alkaline phosphatase. In conclusion peptide 1142-1153 provides a useful tool for the further study and characterization of phosphotyrosyl protein phosphatases.     

The flip side of phospho‐signalling: Regulation of protein dephosphorylation and the protein phosphatase 2Cs

Protein phosphorylation is a key signalling mechanism and has myriad effects on protein function. Phosphorylation by protein kinases can be reversed by protein phosphatases, thus allowing dynamic control of protein phosphorylation. Although this may suggest a straightforward kinase–phosphatase relationship, plant genomes contain five times more kinases than phosphatases. Here, we examine phospho‐signalling from a protein phosphatase centred perspective and ask how relatively few phosphatases regulate many phosphorylation sites. The most abundant class of plant phosphatases, the protein phosphatase 2Cs (PP2Cs), is surrounded by a web of regulation including inhibitor and activator proteins as well as posttranslational modifications that regulate phosphatase activity, control phosphatase stability, or determine the subcellular locations where the phosphatase is present and active. These mechanisms are best established for the Clade A PP2Cs, which are key components of stress and abscisic acid signalling. We also describe other PP2C clades and illustrate how these phosphatases are highly regulated and involved in a wide range of physiological functions. Together, these examples of multiple layers of phosphatase regulation help explain the unbalanced kinase–phosphatase ratio. Continued use of phosphoproteomics to examine phosphatase targets and phosphatase–kinase relationships will be important for deeper understanding of phosphoproteome regulation.     

In-gel protein phosphatase assay using fluorogenic substrates

We developed a method for the detection of phosphatase activity using fluorogenic substrates after polyacrylamide gel electrophoresis. When phosphatases such as Ca²⁺/calmodulin-dependent protein kinase phosphatase (CaMKP), protein phosphatase 2C (PP2C), protein phosphatase 5 (PP5), and alkaline phosphatase were resolved by polyacrylamide gel electrophoresis in the absence of SDS and the gel was incubated with a fluorogenic substrate such as 4-methylumbelliferyl phosphate (MUP), all of these phosphatase activities could be detected in situ. Although 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP) as well as MUP could be used as a fluorogenic substrate for an in-gel assay, MUP exhibited lower background fluorescence. Using this procedure, several fluorescent bands that correspond to endogenous phosphatases were observed after electrophoresis of various crude samples. The in-gel phosphatase assay could also be used to detect protein phosphatases resolved by SDS-polyacrylamide gel electrophoresis. In this case, however, the denaturation/renaturation process of resolved proteins was necessary for the detection of phosphatase activity. This procedure could be used for detection of renaturable protein phosphatases such as CaMKP and some other phosphatases expressed in cell extracts. The present fluorescent in-gel phosphatase assay is very useful, since no radioactive compounds or no special apparatus are required.     

Alpha4 protein as a common regulator of type 2A-related serine/threonine protein phosphatases

The catalytic activity of the C subunit of serine/threonine phosphatase 2A is regulated by the association with A (PR65) and B subunits. It has been reported that the alpha4 protein, a yeast homolog of the Tap42 protein, binds the C subunit of serine/threonine phosphatase 2A and protein phosphatase 2A-related protein phosphatases such as protein phosphatase 4 and protein phosphatase 6. In the present study, we showed that alpha4 binds these three phosphatases and the association of alpha4 reduces the activities of these phosphatases in vitro. In contrast, PR65 binds to the C subunit of serine/threonine phosphatase 2A but not to protein phosphatase 4 and protein phosphatase 6. These results suggest that the alpha4 protein is a common regulator of the C subunit of serine/threonine phosphatase 2A and protein phosphatase 2A-related protein phosphatases.     

Substrate specifity in Amoeba proteus

Three different phosphatases ("slow", "middle" and "fast") were found in Amoeba proteus (strain B) after PAGE and a subsequent gel staining in 1-naphthyl phosphate containing incubation mixture (pH 9.0). Substrate specificity of these phosphatases was determined in supernatants of homogenates using inhibitors of phosphatase activity. All phosphatases showed a broad substrate specificity. Of 10 tested compounds, p-nitrophenyl phosphate was a preferable substrate for all 3 phosphatases. All phosphatases were able to hydrolyse bis-p-nitrophenyl phosphate and, hence, displayed phosphodiesterase activity. All phosphatases hydrolysed O-phospho-L-tyrosine to a greater or lesser degree. Only little differences in substrate specificity of phosphatases were noticed: 1) "fast" and "middle" phosphatases hydrolysed naphthyl phosphates and O-phospho-L-tyrosine less efficiently than did "slow" phosphatase; 2) "fast" and "middle" phosphatases hydrolysed 2- naphthyl phosphate to a lesser degree than 1-naphthyl phosphate 3) "fast" and "middle" phosphatases hydrolysed O-phospho-L-serine and O-phospho-L-threonine with lower intensity as compared with "slow" phosphatase; 4) as distinct from "middle" and "slow" phosphatases, the "fast" phosphatase hydrolysed glucose-6-phosphate very poorly. The revealed broad substrate specificity of "slow" phosphatase together with data of inhibitory analysis and results of experiments with reactivation of this phosphatase by Zn2+-ions after its inactivation by EDTA strongly suggest that only the "slow" phosphatase is a true alkaline phosphatase (EC 3.1.3.1). The alkaline phosphatase of A. proteus is secreted into culture medium where its activity is low. The enzyme displays both phosphomono- and phosphodiesterase activities, in addition to supposed protein phosphatase activity. It still remains unknown, to which particular phosphatase class the amoeban "middle" and "fast" phosphatases (pH 9.0) may be assigned.     

Crystal structure of mammalian purple acid phosphatase.

BACKGROUND: Mammalian purple acid phosphatases are highly conserved binuclear metal-containing enzymes produced by osteoclasts, the cells that resorb bone. The enzyme is a target for drug design because there is strong evidence that it is involved in bone resorption. RESULTS: The 1.55 A resolution structure of pig purple acid phosphatase has been solved by multiple isomorphous replacement. The enzyme comprises two sandwiched beta sheets flanked by alpha-helical segments. The molecule shows internal symmetry, with the metal ions bound at the interface between the two halves. CONCLUSIONS: Despite less than 15% sequence identity, the protein fold resembles that of the catalytic domain of plant purple acid phosphatase and some serine/threonine protein phosphatases. The active-site regions of the mammalian and plant purple acid phosphatases differ significantly, however. The internal symmetry suggests that the binuclear centre evolved as a result of the combination of mononuclear ancestors. The structure of the mammalian enzyme provides a basis for antiosteoporotic drug design.     

Protein Ser/Thr phosphatases with kelch-like repeat domains

This report describes the presence in plants of protein Ser/Thr phosphatases of the PPP family, homologous to PfPPalpha phosphatase from Plasmodium falciparum. Like PfPPalpha, they possess large N-terminal domains and catalytic domains that are more closely related to the protein phosphatase 1 group. The N-terminal domains of PfPPalpha and its plant homologues contain tandem kelch-like repeats, not previously identified in any protein phosphatases, suggesting that the N-terminal domains may form beta-propeller structures mediating protein-protein interactions. We therefore suggest that this novel phosphatase group be designated as PPKLs for protein phosphatases with kelch-like repeat domains. Four PPKL isoforms are encoded in the Arabidopsis thaliana genome, of which at least three are expressed. PPKLs appear to be ubiquitous in Viridiplantae. The existence of a protein phosphatase group shared by Viridiplantae and Apicomplexa, but not other eukaryotes, is in line with the theory of the origin of Apicomplexa by endosymbiosis of nonphotosynthetic eukaryotes with red algae.     

Studies of small molecule interactions with protein phosphatases using biosensor technology

Reversible protein phosphorylation of serine, threonine, and tyrosine residues by protein kinases and phosphatases is important for the regulation of cellular signal transduction and controls many cellular functions. Disturbances in this regulation have been implicated in a growing number of diseases, making kinases and phosphatases useful targets for therapeutic intervention. The suitability of surface plasmon resonance (SPR) technology has been widely demonstrated in many drug discovery applications. A novel and straightforward methodology is presented for analyzing small molecule binding to two serine/threonine phosphatases, PP1 and PP2B (calcineurin), and to the prototypic tyrosine phosphatase, PTP1B. Emphasis was placed on investigating the immobilization conditions of the phosphatases by using reducing conditions, inhibitors and metal ions. A comparison of inhibitor binding, either to phosphatase (PP2B) alone or in complex with the regulatory protein subunit calmodulin, revealed different kinetics. The methodology was also used to test inhibitor specificity toward different phosphatases. Inhibition of regulatory protein PP-inhibitor-2 binding to PP1 by a small molecule inhibitor was demonstrated. This type of information, together with data on compound binding that is independent of enzyme activity and in which affinities are resolved into kinetic rate constants, may be of great significance for the development of highly specific and high-affinity phosphatase inhibitors.     

Histone H1 phosphorylated by protein kinase C is a selective substrate for the assay of protein phosphatase 2A in the presence of phosphatase 1

A protein phosphatase assay, selective for protein phosphatase 2A, has been developed. Bovine histone H1 phosphorylated by protein kinase C and [gamma-32P]ATP, designated H1(C), was tested as the substrate for various preparations of protein phosphatases 1 and 2A. The phosphatase 2A preparations were 10-60-times more active with H1(C) as the substrate when compared to phosphorylase a. The phosphatase 1 enzymes showed very little dephosphorylation of the H1(C) substrate, the activity being less than 5% of the phosphorylase phosphatase activity. This preference and selectivity was demonstrated for purified phosphatase preparations in addition to fresh tissue extracts. The assay provides a rapid, simple assay for the routine analysis of phosphatase 2A in the presence of phosphatase 1, without the use of heat-stable inhibitor proteins.     

Specificity of protein phosphatases in the dephosphorylation of protein kinase C.

Protein kinase C can autophosphorylate in vitro and has also been shown to be phosphorylated in vivo. In order to investigate the factors that may determine the phosphorylation state of protein kinase C in vivo, we determined the ability of the ATP + Mg2+-dependent phosphatase and the polycation-stimulated (PCS) phosphatases to dephosphorylate protein kinase C in vitro. These studies show that all the oligomeric forms of the PCS phosphatases (PCSH1, PCSH2, PCSM and PCSL phosphatases) are effective in the dephosphorylation of protein kinase C, showing 34-82% of the activity displayed with phosphorylase a as substrate. In contrast both the catalytic subunit of the PCS phosphatase and that of the ATP+Mg2+-dependent phosphatase showed only weak activity with protein kinase C as substrate. All these phosphatases, however, were activated by protamine (Ka 14-16 micrograms/ml) through what appears to be a substrate-directed effect. The relative role of these phosphatases in the control of protein kinase C is discussed.     

Thrombin-induced effects are selectively inhibited following treatment of intact human platelets with okadaic acid.

The involvement of protein phosphatases in regulating platelet activation was studied. The major portion of the phosphorylase phosphatase activity found in platelet lysates appears to be of the type 1 variety. The identification of this enzyme was based on the finding that greater than 80% of protein phosphatase activity was inhibited by the heat-stable inhibitor protein inhibitor 2 and, while only 20% of the phosphorylase phosphatase activity in platelet extracts was inhibited by 2 nM okadaic acid, greater than 95% of the activity was inhibited in the presence of 1 microM okadaic acid. Increases in protein phosphorylations occurred and thrombin-induced release of serotonin was prevented as a result of artificially inhibiting the enzyme with okadaic acid in intact platelets. This implies either that the regulation of okadaic acid sensitive protein phosphatases is necessary for some agonist-induced effects or that okadaic acid sensitive phosphatases are required for maintaining platelets in a responsive state.     

Inhibitory effect of a marine-sponge toxin, okadaic acid, on protein phosphatases. Specificity and kinetics.

The inhibitory effect of a marine-sponge toxin, okadaic acid, was examined on type 1, type 2A, type 2B and type 2C protein phosphatases as well as on a polycation-modulated (PCM) phosphatase. Of the protein phosphatases examined, the catalytic subunit of type 2A phosphatase from rabbit skeletal muscle was most potently inhibited. For the phosphorylated myosin light-chain (PMLC) phosphatase activity of the enzyme, the concentration of okadaic acid required to obtain 50% inhibition (ID50) was about 1 nM. The PMLC phosphatase activities of type 1 and PCM phosphatase were also strongly inhibited (ID50 0.1-0.5 microM). The PMCL phosphatase activity of type 2B phosphatase (calcineurin) was inhibited to a lesser extent (ID50 4-5 microM). Similar results were obtained for the phosphorylase a phosphatase activity of type 1 and PCM phosphatases and for the p-nitrophenyl phosphate phosphatase activity of calcineurin. The following phosphatases were not affected by up to 10 microM-okadaic acid: type 2C phosphatase, phosphotyrosyl phosphatase, inositol 1,4,5-trisphosphate phosphatase, acid phosphatases and alkaline phosphatases. Thus okadaic acid had a relatively high specificity for type 2A, type 1 and PCM phosphatases. Kinetic studies showed that okadaic acid acts as a non-competitive or mixed inhibitor on the okadaic acid-sensitive enzymes.     

Modulation of rat liver hydroxymethylglutaryl-CoA reductase by protein phosphatases: purification of nonspecific hydroxymethylglutaryl-CoA reductase phosphatases

Four phosphoprotein phosphatases, with the ability to act upon hydroxymethylglutaryl (HMG)-CoA reductase, phosphorylase, and glycogen synthase have been purified from rat liver cytosol through a process that involves DEAE-cellulose, aminohexyl-Sepharose-4B, and Bio-Gel A 1.5 m chromatographies. Protein phosphatase II (Mr 180,000) was the major enzyme (68%) with a very broad substrate specificity, showing similar activity toward the three substrates. Phosphatases I1 (Mr 180,000) and I3 (Mr 250,000) accounted for only 12 and 15% of the total activity, respectively, and they were also able to dephosphorylate the three substrates. In contrast, phosphatase I2 (Mr 200,000) showed only phosphorylase phosphatase activity with insignificant dephosphorylating capacity toward HMG-CoA reductase and glycogen synthase. Upon ethanol treatment at room temperature, the Mr of all phosphatases changed; protein phosphatases I2, I3, and II were brought to an Mr of 35,000, while phosphatase I1 was reduced to an Mr of 69,000. Glycogen synthase phosphatase activity was decreased in all four phosphatases. There was also a decrease in phosphatase I1 activity toward HMG-CoA reductase and phosphorylase as substrates. The HMG-CoA reductase phosphatase and phosphorylase phosphatase activities of phosphatases I2, I3, and II were increased after ethanol treatment. Each protein phosphatase showed a different optimum pH, which changed depending on the substrate. The four phosphatases increased their activity in the presence of Mn2+ and Mg2+. In general, Mn2+ was a better activator than Mg2+, and phosphatase I1 showed a stronger dependency on these cations than any other phosphatase. Phosphorylase was a competitive substrate in the HMG-CoA reductase phosphatase and glycogen synthase phosphatase reactions of protein phosphatases I1, I3, and II. HMG-CoA reductase was also able to compete with phosphorylase and glycogen synthase for phosphatase activity. Glycogen synthase phosphatase activity presented less inhibition in the low-Mr forms. A comparison has been made with other protein phosphatases previously reported in the literature.     

Identification of the phosphatase deinhibitor protein phosphatases in rabbit skeletal muscle.

In rabbit skeletal muscle the polycation-stimulated (PCS) protein phosphatases [Merlevede (1985) Adv. Protein Phosphatases 1, 1-18] are the only phosphatases displaying significant activity toward the deinhibitor protein. Among them, the PCSH protein phosphatase represents more than 80% of the measurable deinhibitor phosphatase activity associated with the PCS phosphatases. The deinhibitor phosphatase activity co-purifies with the PCSH phosphatase to apparent homogeneity. In the last purification step two forms of PCSH phosphatase were separated (PCSH1, containing 62, 55 and 34 kDa subunits, and PCSH2, containing 62 and 35 kDa subunits), both showing the same deinhibitor/phosphorylase phosphatase activity ratio. The activity of the PCSH phosphatase toward the deinhibitor is not stimulated by polycations such as protamine, histone H1 or polylysine, unlike the stimulation observed with phosphorylase as the substrate. The phosphorylase phosphatase activity of PCSH phosphatase is inhibited by ATP, PPi and Pi, whereas the deinhibitor phosphatase activity of the enzyme is much less sensitive to these agents.     

Fluorogenic probes for imaging cellular phosphatase activity

The ability to visualize enzyme activity in a cell, tissue, or living organism can greatly enhance our understanding of the biological roles of that enzyme. While many aspects of cellular signaling are controlled by reversible protein phosphorylation, our understanding of the biological roles of the protein phosphatases involved is limited. Here, we provide an overview of progress toward the development of fluorescent probes that can be used to visualize the activity of protein phosphatases. Significant advances include the development of probes with visible and near-infrared (near-IR) excitation and emission profiles, which provides greater tissue and whole-animal imaging capabilities. In addition, the development of peptide-based probes has provided some selectivity for a phosphatase of interest. Key challenges involve the difficulty of achieving sufficient selectivity for an individual member of a phosphatase enzyme family and the necessity of fully validating the best probes before they can be adopted widely.