Evolution of the Multifunctional Protein Tyrosine Phosphatase Family
http://www.100md.com
分子生物学进展 2004年第4期
Computational Molecular Biology, Max-Planck-Institute for Molecular Genetics, Berlin, Germany
E-mail: joerg.schultz@biozentrum.uni-wuerzburg.de.
Abstract
The protein tyrosine phosphatase (PTP) family plays a central role in signal transduction pathways by controlling the phosphorylation state of serine, threonine, and tyrosine residues. PTPs can be divided into dual specificity phosphatases and the classical PTPs, which can comprise of one or two phosphatase domains. We studied amino acid substitutions at functional sites in the phosphatase domain and identified putative noncatalytic phosphatase domains in all subclasses of the PTP family. The presence of inactive phosphatase domains in all subclasses indicates that they were invented multiple times in evolution. Depending on the domain composition, loss of catalytic activity can result in different consequences for the function of the protein. Inactive single-domain phosphatases can still specifically bind substrate and protect it from dephosphorylation by other phosphatases. The inactive domains of tandem phosphatases can be further subdivided. The first class is more conserved, still able to bind phosphorylated tyrosine residues and might recruit multiphosphorylated substrates for the adjacent active domain. The second has accumulated several variable amino acid substitutions in the catalytic center, indicating a complete loss of tyrosine-binding capabilities. To study the impact of substitutions in the catalytic center to the evolution of the whole domain, we examined the evolutionary rates for each individual site and compared them between the classes. This analysis revealed a release of evolutionary constraint for multiple sites surrounding the catalytic center only in the second class, emphasizing its difference in function compared with the first class. Furthermore, we found a region of higher conservation common to both domain classes, suggesting a new regulatory center. We discuss the influence of evolutionary forces on the development of the phosphatase domain, which has led to additional functions, such as the specific protection of phosphorylated tyrosine residues, substrate recruitment, and regulation of the catalytic activity of adjacent domains.
Key Words: protein tyrosine phosphatase ? antiphosphatase ? signaling enzymes ? functional divergence ? evolutionary site rates
Introduction
Protein tyrosine phosphatases (PTPs) regulate physiological processes common to all metazoa, including growth, differentiation, metabolism, the cell cycle, and cytoskeletal function. Together with tyrosine kinases, they control the phosphorylation state of tyrosine and serine/threonine residues of signaling proteins in highly specific reactions. The level of protein phosphorylation is highly dynamic and any disturbance can lead to severe malfunction of the eukaryotic cell. Increased level of protein phosphorylation results in abnormal proliferation and many cancer types show a mutation or deletion of a PTP gene (Siminovitch et al. 1999). In contrast to protein tyrosine kinases, which have a growth promoting potential, PTPs can act as tumor suppressors and inhibit cell growth (Dahia 2000; Wu et al. 2003). PTPs have also been implicated in B-lymphocyte and T-lymphocyte activation and insulin signaling, so that PTPs represent attractive drug targets for a wide variety of diseases, such as cancer, inflammation, diabetes, and obesity (Justement 2001; van Huijsduijnen, Bombrun, and Swinnen 2002; Asante-Appiah and Kennedy 2003).
Functionally, two types of PTPs, conserved in sequence and structure can be distinguished: The classical PTPs, which are specific for tyrosine residues, and the dual-specificity phosphatases (DSPs), which can additionally dephosphorylate serine and threonine residues.
Obviously, the catalytic residues of an enzyme are key to its molecular function. Therefore, it came as a surprise when a member of the DSP family, Sfb1, with a replacement of the catalytically essential cysteine was described (Cui et al. 1998). As expected, this protein had lost its enzymatic activity, raising the question about its molecular function. Experimentally, it could be shown that Sbf1 has maintained its ability to stably bind phosphorylated substrate, protecting the substrate from other phosphatases at this specific site. Because of its antagonistic mechanism, this phosphatase has been termed "antiphosphatase" (De Vivo et al. 1998; Hunter 1998). Sbf1 function differs also on the cellular level. In contrast to active phosphatases exposing growth inhibitory behavior, it shows transforming abilities (Cui et al. 1998). A similar substitution of the catalytic cysteine was found in another subgroup of the DSPs, called STYX (Wishart et al. 1995). Within the classical PTPs, substitutions of functional residues have been described in the receptor protein tyrosine phosphatase (RPTP) family. Most RPTPs contain two phosphatase domains, of which the second phosphatase domain is inactive or remains with very low activity because of substitutions at functional sites. Although the detailed function of this catalytically inactive domain is not yet totally understood, it is of absolute importance for the function of the receptors. Partial or entire deletion of the second domain completely abolishes or severely reduces activity of the first domain, so that a role in regulating the catalytic activity or substrate specificity of the first domain has been hypothesized (Streuli et al. 1990; Johnson et al. 1992). Furthermore, specific interaction of domain II with domain I leading to active site blocking of domain I (Bilwes et al. 1996; Majeti et al. 1998; Blanchetot and den Hertog 2000) has been shown experimentally. Inactive phosphatase domains have also been studied experimentally in the single-phosphatase domain RPTPs (Kambayashi et al. 1995; Cui et al. 1996). Here, the PTP typical signature with the catalytic cysteine is present, but two other catalytically important residues are substituted, leading to loss of phosphatase activity.
On the basis of these more anecdotal reports, we analyze here on a genomic scale how frequent substitutions of functional residues in the PTP family are and whether they were invented multiple times in evolution. Furthermore, we use site-specific evolutionary rates to unravel the evolutionary implications of these substitutions for the whole domain, leading to the prediction of distinct functional classes and the delineation of an additional functional site.
Methods
Data Sets
Protein sequences for the whole genomes of Homo sapiens (version 9.30), Mus musculus (version 9.3), Fugu rubripes (version 10.2), Anopheles gambiae (version 9.1) and Caenorhabditis elegans (version 12.95) were retrieved from the ENSEMBL Web site (www.ensembl.org). The Drosophila melanogaster predicted protein sequences (version 3) were obtained from BDGP (www.fruitfly.org), and the Ciona intestinalis predicted protein sequences (version 1.0) was obtained from JGI (www.jgi.doe.gov). These data sets were searched with Hidden Markov Models (HMMs) (HMMER: http://hmmer.wustl.edu/) specific for the classical PTPs and DSPs, respectively, which were built from the family multiple sequence alignment retrieved from SMART (http://smart.embl-heidelberg.de/). For subclassification, SMART-combined E-value thresholds were used (Schultz et al. 1998). This schema allows setting of two E-value thresholds for each subfamily HMM of a larger, homologous family. The first is given by the E-value of the best-scoring family member not belonging to the actual subfamily, the second by the worst hit before the first nonfamily member. Searching with all subfamily specific HMMs assigns a sequence to a subfamliy if its E-value is lower than the according subfamily threshold. Sequences that cannot be assigned to any subfamily but with an E-value lower than the family cutoff in at least one search are assigned to the family without any subfamily classification. These sequences were marked as "undefined specificity" and not used in the further analysis. We cleaned the obtained data set by filtering the sequences for alternative splice variants and for fragments. Only the best-scoring hit per gene was used for further analysis, and sequences that did not cover the complete HMM, tolerating an uncovered interval of 10 amino acids at the beginning and end of the profile, were excluded as they might represent fragments.
Scan for Inactive Phosphatases
For each PTP subclass, a multiple sequence alignment of all found members was created according to the family alignment in SMART, using HMMalign (HMMER). The alignments were manually curated to remove unnecessary gaps and to connect interrupted secondary structure elements, partially with the help of secondary structure information, using a representative structure from PDB (1LAR for the classical PTPs, 1VHR for the DSPs). Knowing the positions of functional residues from literature, we were able to scan the sequences for substitutions at these sites and to extract the substituted amino acid from the alignment. If these sites were occupied by gaps, absence of the functional residue was not considered in the further analysis.
Evolutionary Rate Analysis
After definition of the subclasses (D2A, D2B, membrane proximal, and cytosolic), we compared the evolutionary rates of all sites between the subclasses. Therefore, we selected the following genes from human and mouse for further analysis: Homo sapiens: ENSP00000175756, ENSP00000246887, ENSP00000248594, ENSP00000256635, ENSP00000262539, ENSP00000263708, and ENSP00000311857; Mus musculus: ENSMUSP00000022508, ENSMUSP00000025420, ENSMUSP00000027633, ENSMUSP00000029053, ENSMUSP00000029433, ENSMUSP00000030556, and ENSMUSP00000048119. The evolutionary site rates were estimated with Tree-Puzzle version 5.1 (Strimmer and von Haeseler 1996) using the quartet puzzling algorithm under the substitution model of Jones-Taylor-Thornton with an eight site-rate category discretized gamma model (Yang 1994). This model sets the average rate of a site to 1 and assigns each site to one of eight rate classes. As the rate of one class can differ between analyses of different subfamilies, we considered a site as differentially evolving if its rate was below 0.8 in the conserved subfamily and above 1.6 in the fast-evolving one. These values were chosen because within all analyses, they covered classes 1 to 4 for the conserved domain and classes 7 and 8 in the fast-evolving domain (total rates range approximately from 0 to 3 within all analyses). We performed the analyses separately for mouse and human and accepted only sites that were classified as differentially evolving in at least three of the four possible comparisons.
Results and Discussion
Amino Acid Substitutions at Functional Sites
As a first step to analyze the evolution of tyrosine phosphatases, we searched all to date sequenced metazoan genomes for phosphatase domains using specific HMMs for the classical PTPs and DSPs created from the SMART family alignments. Table 1 shows the presence of phosphatase domains in different genomes after filtering for fragments and alternative splice variants. Phosphatase domains that could not be clearly assigned to either subclass are listed as "undefined specificity" phosphatases. The number of phosphatase domains found in human, mouse, and pufferfish, which have a comparable proteome size, are similar for the total number as well as within the subtypes. Drosophila and Anopheles show the same ratio of phosphatases to the proteome as seen in vertebrates. Contrasting this, the genomes of C. elegans and Ciona intestinalis contain a substantial increase of phosphatases. Although the Ciona proteome is comparable in size to the Drosophila proteome, Ciona has more than twice as many phosphatases and almost as many as vertebrates, caused by an expansion in the tandem domain RPTPs. An even larger expansion can be observed in C. elegans, whose genome contains twice as many classical PTPs as human or mouse. Here, the multiplications fall into the class of the single domain PTPs.
Table 1 Phosphatase Domains in Metazoa.
One aim of the analysis was to investigate the extent of substitutions in functional sites within the phosphatase domain family. We focused on sites directly involved in catalysis or substrate binding, as it can be expected that substitutions in these sites will lead to a loss of catalytic activity. Within the PTP subfamily, these sites are (here and in the following, all positions regarding a classical PTP refer to pdb structure 1LAR): C-216, for attacking the phosphorous atom nucleophilically to form a phosphoenzyme intermediate and D-184 and Q-260 to position and polarize the active water molecule that dephosphorylates the phosphoenzyme in a second nucleophilic reaction, the substrate binding R-222 and the aromatic tyrosine or phenylalanine at position 49 that forms a stack with the phenyl ring of the phosphotyrosine (Stuckey et al. 1994; Fauman et al. 1996; Puius et al. 1997; Pannifer et al. 1998). Catalytic functional sites in the DSP family are (positions regarding a DSP refer to pdb structure 1VHR) C-123 and R-129 analog to the PTPs and D-91, which acts as general acid in the dephosphorylation step and as general base in the hydrolysis of the phosphoenzyme, the latter reaction is supported by S-130 (Yuvaniyama et al. 1996; Puius et al. 1997). Using a multiple sequence alignment of all obtained PTP and DSP sequences, respectively, we were able to extract substitutions at these positions.
The analysis revealed that all subclasses of the DSPs and classical PTPs, including the membrane-proximal, membrane-distal, and cytosolic domains, carry substitutions in functional positions.
Table 2 shows the distribution of phosphatase domains among subclasses for each investigated species and the number of domains with substitutions at functional sites within each subclass. The number of putative inactive domains varies strongly among the different species; however, it accounts for a significant part of all subclasses, even in the single-domain phosphatase subfamily. One might argue that the catalytically inactive proteins are nonfunctional relics in the genome, but their conservation between species and also within subfamilies, strongly indicates their functional importance. In many of the observed cases, the protein contains either a single nonactive phosphatase domain or if it contains two, both are inactive. Depending on the type of substitution, these proteins are good candidates as potential antiphosphatases.
Table 2 Number of PTP Domains and of Putative Inactive Domains Among the Different PTP Subclasses.
An extremely high portion of amino acid substitutions is found in the membrane-distal domain of RPTPs, in which all sequences feature two or more substitutions. Based on the type of substitutions, we split this group into two subfamilies. This split is supported by a phylogenetic analysis, which revealed a monophyletic origin of the subfamilies. One subclass (D2A) shows a very high degree of conservation in these substitutions, with the catalytic aspartic acid (position 184) replaced by glutamic acid and the substrate binding tyrosine (position 49) either replaced by valine or leucine. The fact that these substitutions in functional residues maintain the biochemical properties of the original amino acids suggests that this domain might be able to carry out the dephosphorylation reaction. Indeed, experiments with HPTP, a receptor type protein tyrosine phosphatase carrying a DE and YV substitution at the functional sites in the membrane-distal domain shows a small rest activity, even if D2 is expressed by itself (Wang and Pallen 1991). Still, mutation of these sites to the amino acids found in active phosphatases lead to a full recovery of catalytic activity (Lim et al. 1998). These experimental results on the one hand corroborate the functional importance of the identified sites, but on the other hand, they hint that there is still selective pressure on the catalytic center of D2A.
In the second subclass (D2B) substitutions in functional sites are more frequent and more heterogeneous than in subclass D2A. The high number of substitutions and the high variety in amino acids makes it seem unlikely that these domains maintained their catalytic activity. Indeed, Streuli et al. (1990) experimentally showed that the D2B domain of CD45 completely lost its phosphatase activity. The fact that the inactive phosphatase domain has not been lost during evolution and that orthologs are found in all species investigated in our study excludes the mutation to a nonfunctional "pseudodomain." On the contrary, because the PTP structure is still conserved, a specialization on other functions is likely to have occurred during evolution.
In summary, substitutions in functional sites reside in all subclasses of the phosphatase family. This raises the question of whether the inactive phosphatase was invented once or multiple times. If invented once, its widespread presence would indicate an origin before the split into the subclasses. Assuming this monophyletic origin would imply that inactive phophatases of different subclasses are more related to each other than to other members of the subclass, which is not the case. Furthermore, it was shown that the major subclasses evolved monophyletically (Andersen et al. 2001). Therefore, we conclude that the invention of inactive phophatases happened multiple times independently during evolution. The percentage of nonfunctional phosphatases in metazoan genomes is surprisingly high. The regulation of signaling pathways by protection of phosphorylated serine/threonine or tyrosine residues might, therefore, turn out to be an important mechanism to modulate signaling pathways. It has to be further investigated whether the phenomenon of nonfunctional enzymes is restricted to the phosphatase family or whether other signaling enzymes show a similar behavior.
Analysis of Altered Evolutionary Rates Between Phosphatase Subclasses
The strikingly high number of amino acid substitutions in functional residues in the membrane-distal phosphatase domain of receptor PTPs, which is associated with loss of activity, raises the question how on the one hand these substitutions evolved and how on the other hand they influenced the evolution of the whole domain. As a change in function should be mirrored in a change of evolutionary constraints at the involved sites, we compared the site-specific evolutionary rates between the active (cytosolic and membrane-proximal) and the inactive (D2A and D2B) PTP subclasses, a method that has been used to identify functionally important sites (Gu and Vander Velden 2002; Blouin, Boucher, and Roger 2003). Because of the lack of a representative quantity of PTP sequences in the single subclasses of most species, evolutionary rates could only be estimated reliably for human and mouse. Indeed, we found a substantial number of sites with changing evolutionary constraints (table 3). To further understand how these work together, we mapped these sites onto the structure of the PTP domain (pdb 1LAR) (fig. 1). This revealed that these sites cluster within two regions. One group is located around the catalytic center, and the other group is located on the backside of the protein. In the following section we discuss these regions separately.
Table 3 Sites with Altered Evolutionary Rates Between the PTP Subclasses.
FIG. 1. Sites with altered evolutionary rates mapped onto the tertiary structure. Functional sites and sites with altered evolutionary rates were mapped onto the pdb structure 1LAR. Catalytic and substrate-binding residues are colored blue, residues at sites with altered evolutionary rates yellow, and residues, which belong to both categories are colored green. (A) and (B) show a view on the catalytic center of the domain, (B) and (C) show the backside of the domain. (A) Sites fast evolving in D2A but conserved in the active domains. (B) Sites fast evolving in D2B but conserved in the active domains. (C) Sites more conserved in D2A than in the active domains. (D) Sites more conserved in domain D2B than in the active domains
Fast-Evolving Sites Around the Catalytic Center
The comparison of evolutionary site rates of the membrane-distal domains D2B versus the active membrane-proximal domains and the cytosolic domains revealed 15 sites that are fast evolving in domain D2B but conserved in the other domain subclasses (table 3). Most of these are located around or within the catalytic center (fig. 1). For example, one of the sites (260) is in the corresponding active domain occupied by the catalytic aspartate, and two other sites (49 and 185) are involved in substrate binding. Because domain D2B has accumulated various substitutions in its functional sites and has lost catalytic activity, it is expected that the selective constraint on the catalytic center was relaxed, which consequently allowed the surrounding sites to evolve at a faster rate. The fact that these sites are not only fast evolving in the inactive domain D2B but also conserved in the active domains in addition to their appropriate location on the surface of the domain, suggests their role in substrate binding in the active domains.
The evolutionary events in D2B after domain duplication might have been triggered by a single mutation of a functional residue, which led to a noncatalytic domain. Subsequently, the evolutionary constraint of the catalytic center was relaxed, leading to the accumulation of mutations in the surrounding.
In contrast to domain D2B, there are only five sites that are fast evolving in D2A but conserved in the active domains. One of them (position 185) is located next to the functional site 184, occupied by aspartate in the active domain but replaced by glutamate in domain D2A. This mutation might have freed the immediate surrounding from selective constraint and allowed a faster evolutionary rate at site 185. The functional residues in the catalytic center of D2A are affected by only two amino acid substitutions. The catalytic aspartate is replaced by glutamate acid and the substrate binding tyrosine either by valine or leucine. These substitutions maintain the biochemical properties, and although the catalytic activity of this domain is barely detectable, it is still able to stably bind its substrate (Bliska et al. 1992). The analysis demonstrates that the catalytic center is still under selective pressure because residues that are fast evolving in D2B and predicted to function in substrate binding are conserved in D2A. This is confirmed experimentally by regaining a catalytically fully active domain if the two substituted functional residues are converted to their original amino acids (Lim et al. 1998). We conclude that the catalytic center of D2A, in contrast to D2B, plays a pivotal role in the function of tandem domain phosphatases, leading to the question of what this function is. As the domain has lost its catalytic activity but still can bind to phosphotyrosine, one could assume two complementary molecular functions. First, the domain could function as "antiphosphatase" as described for DSPs (Cui et al. 1998). Second, it might work as an adaptor domain for phosphotyrosine substrates, similar to SH2 and PTB domains, revealing an additional function of the PTP domain family.
Slow Evolving Sites on the Backside of the Domain
The complete loss of evolutionary pressure on the catalytic center of the D2B family leads to the question of what the function of this domain is and whether there is a similar role of the D2A domains. If a new function was acquired, this should be reflected in novel conserved sites. Therefore, we searched for sites that evolve at a higher rate in the active domains while they are conserved in D2A or D2B. The comparison found 11 sites conserved in D2B and 12 sites conserved in D2A, of which five sites are found in both analyses (table 3). Almost all sites are located on the surface of the "backside" of the domain (fig. 1). This could indicate, that a new functional center has evolved in this region. The solvent exposure as well as the nature of the conserved amino acids might hint that this region is involved in protein-protein interactions. Indeed, interactions of the membrane-distal and membrane-proximal domains have been described recently. The direct interaction of the membrane-distal domain with the membrane-proximal domain stabilizes the enzyme and enhances catalytic activity (Felberg and Johnson 2000). This effect can be abrogated by deletion of the two carboxy-terminal -helices of the membrane-distal domain (244 to 278) (Johnson et al. 1992). These two helices host two residues (245 and 256), which are significantly more conserved in D2A and D2B than in the active domains, and one residue (250) that is more conserved if D2B is compared against the active domains. Another highly conserved site (240) found in both comparisons is preceding the two helices. These sites might play an important role in the interaction between the phosphatase domains. The additional sites with altered selective constraint might contribute to the stable binding but are not sufficient for stable interaction without presence of the two carboxy-terminal helices. Experimental mutation of these sites might give further insight into the molecular mechanism of regulation of RPTPs.
Together with the variation of the catalytic site, our results indicate, that domain D2A and D2B have distinct influences on the activity of membrane-proximal domains. We suggest that both domains can control activity of the first domain by interaction of residues from the "backside" of the membrane-distal domain and residues from the membrane-proximal domain. In addition, D2A can regulate substrate specificity of the membrane-proximal domain and remain associated with the substrate protein, which is accomplished by the inactive catalytic center of D2A.
The results of our analyses allow delineating a possible scenario for the evolution of the membrane-distal domain of RPTPs. The overlap of conserved residues on the "backside" of both D2A and D2B indicates that their common ancestor already had evolved this novel functional site. Whether the membrane-distal domain of the first RPTP was still active remains unclear, but the complete absence of a domain without substitutions of functional residues within the catalytic center hints that it indeed was inactive. This ancestral RPTP gave rise to one lineage with a conserved catalytic center that is still able to bind substrate (D2A) and one lineage that accumulated substitutions around the catalytic center, completely loosing the substrate binding function (D2B).
Conclusions
Our analysis of the PTP family illustrates how a closely related domain family can evolve multiple molecular functions. On the catalytic site, the family varies in the specificity of substrates, allowing the dephosphorylation of a wide range of phosphoproteins as well as phosphoinositides (Maehama and Dixon 1998). Loss of catalytic activity opens the opportunity to evolve novel functions at the catalytic site. In single-domain phosphatases, this event led to the evolution of proteins antagonizing phosphatase function, the antiphosphatases. The inactive domain in tandem domain phosphatases functions in substrate recognition and binding to multiple phosphorylated proteins. Here, it might work as a competitor for other phosphotyrosine-binding domains such as SH2 and PTB. In addition to these changes within the catalytic site, loss of catalytic activity also enabled the evolution of a novel functional site within the domain and specialization on regulatory functions. In summary, our analysis shows how evolution can create novel functionality based on an existing, well-adapted enzyme, illustrating the versatility of the PTP family.
Acknowledgements
We would like to thank Heiko Schmidt for help with the Tree-Puzzle analysis.
Literature Cited
Andersen, J. N., O. H. Mortensen, G. H. Peters, P. G. Drake, L. F. Iversen, O. H. Olsen, P. G. Jansen, H. S. Andersen, N. K. Tonks, and N. P. Moller. 2001. Structural and evolutionary relationships among protein tyrosine phosphatase domains. Mol. Cell. Biol. 21:7117-7136.
Asante-Appiah, E., and B. P. Kennedy. 2003. Protein tyrosine phosphatases: the quest for negative regulators of insulin action. Am. J. Physiol. Endocrinol. Metab. 284:E663-670.
Bilwes, A. M., J. den Hertog, T. Hunter, and J. P. Noel. 1996. Structural basis for inhibition of receptor protein-tyrosine phosphatase-alpha by dimerization. Nature 382:555-559.
Blanchetot, C., and J. den Hertog. 2000. Multiple interactions between receptor protein-tyrosine phosphatase (RPTP) alpha and membrane-distal protein-tyrosine phosphatase domains of various RPTPs. J. Biol. Chem. 275:12446-12452.
Bliska, J. B., J. C. Clemens, J. E. Dixon, and S. Falkow. 1992. The Yersinia tyrosine phosphatase: specificity of a bacterial virulence determinant for phosphoproteins in the J774A.1 macrophage. J. Exp. Med. 176:1625-1630.
Blouin, C., Y. Boucher, and A. J. Roger. 2003. Inferring functional constraints and divergence in protein families using 3D mapping of phylogenetic information. Nucleic Acids Res. 31:790-797.
Cui, L., W. P. Yu, H. J. DeAizpurua, R. S. Schmidli, and C. J. Pallen. 1996. Cloning and characterization of islet cell antigen-related protein-tyrosine phosphatase (PTP), a novel receptor-like PTP and autoantigen in insulin-dependent diabetes. J. Biol. Chem. 271:24817-24823.
Cui, X., I. De Vivo, R. Slany, A. Miyamoto, R. Firestein, and M. L. Cleary. 1998. Association of SET domain and myotubularin-related proteins modulates growth control. Nat. Genet. 18:331-337.
Dahia, P. L. 2000. PTEN, a unique tumor suppressor gene. Endocr. Relat. Cancer 7:115-129.
Dehal, P., Y. Satou, and R. K. Campbell, et al. (86 co-authors). 2002. The draft genome of Ciona intestinalis: insights into chordate and vertebrate origins. Science 298:2157-2167.
De Vivo, I., X. Cui, J. Domen, and M. L. Cleary. 1998. Growth stimulation of primary B cell precursors by the antiphosphatase Sbf1. Proc. Natl. Acad. Sci. USA 95:9471-9476.
Fauman, E. B., C. Yuvaniyama, H. L. Schubert, J. A. Stuckey, and M. A. Saper. 1996. The X-ray crystal structures of Yersinia tyrosine phosphatase with bound tungstate and nitrate: mechanistic implications. J. Biol. Chem. 271:18780-18788.
Felberg, J., and P. Johnson. 2000. Stable interdomain interaction within the cytoplasmic domain of CD45 increases enzyme stability. Biochem. Biophys. Res. Commun. 271:292-298.
Gu, X., and K. Vander Velden. 2002. DIVERGE: phylogeny-based analysis for functional-structural divergence of a protein family. Bioinformatics 18:500-501.
Hunter, T. 1998. Anti-phosphatases take the stage. Nat. Genet. 18:303-305.
Johnson, P., H. L. Ostergaard, C. Wasden, and I. S. Trowbridge. 1992. Mutational analysis of CD45: a leukocyte-specific protein tyrosine phosphatase. J. Biol. Chem. 267:8035-8041.
Justement, L. B. 2001. The role of the protein tyrosine phosphatase CD45 in regulation of B lymphocyte activation. Int. Rev. Immunol. 20:713-738.
Kambayashi, Y., K. Takahashi, S. Bardhan, and T. Inagami. 1995. Cloning and expression of protein tyrosine phosphatase-like protein derived from a rat pheochromocytoma cell line. Biochem. J. 306:331-335.
Lim, K. L., P. R. Kolatkar, K. P. Ng, C. H. Ng, and C. J. Pallen. 1998. Interconversion of the kinetic identities of the tandem catalytic domains of receptor-like protein-tyrosine phosphatase PTPalpha by two point mutations is synergistic and substrate-dependent. J. Biol. Chem. 273:28986-28993.
Maehama, T., and J. E. Dixon. 1998. The tumor suppressor, PTEN/MMAC1, dephosphorylates the lipid second messenger, phosphatidylinositol 3,4,5-trisphosphate. J. Biol. Chem. 273:13375-13376.
Majeti, R., A. M. Bilwes, J. P. Noel, T. Hunter, and A. Weiss. 1998. Dimerization-induced inhibition of receptor protein tyrosine phosphatase function through an inhibitory wedge. Science 279:88-91.
Pannifer, A. D., A. J. Flint, N. K. Tonks, and D. Barford. 1998. Visualization of the cysteinyl-phosphate intermediate of a protein-tyrosine phosphatase by x-ray crystallography. J. Biol. Chem. 273:10454-10462.
Puius, Y. A., Y. Zhao, M. Sullivan, D. S. Lawrence, S. C. Almo, and Z. Y. Zhang. 1997. Identification of a second aryl phosphate-binding site in protein-tyrosine phosphatase 1B: a paradigm for inhibitor design. Proc. Natl. Acad. Sci. USA 94:13420-13425.
Schultz, J., F. Milpetz, P. Bork, and C. P. Ponting. 1998. SMART, a simple modular architecture research tool: identification of signaling domains. Proc. Natl. Acad. Sci. USA 95:5857-5864.
Siminovitch, K. A., A. M. Lamhonwah, A. K. Somani, R. Cardiff, and G. B. Mills. 1999. Involvement of the SHP-1 tyrosine phosphatase in regulating B lymphocyte antigen receptor signaling, proliferation and transformation. Curr. Top. Microbiol. Immunol. 246:291-297.
Streuli, M., N. X. Krueger, T. Thai, M. Tang, and H. Saito. 1990. Distinct functional roles of the two intracellular phosphatase like domains of the receptor-linked protein tyrosine phosphatases LCA and LAR. EMBO J. 9:2399-2407.
Strimmer, K., and A. von Haeseler. 1996. Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13:964-969.
Stuckey, J. A., H. L. Schubert, E. B. Fauman, Z. Y. Zhang, J. E. Dixon, and M. A. Saper. 1994. Crystal structure of Yersinia protein tyrosine phosphatase at 2.5 A and the complex with tungstate. Nature 370:571-575.
van Huijsduijnen, R. H., A. Bombrun, and D. Swinnen. 2002. Selecting protein tyrosine phosphatases as drug targets. Drug Discov. Today 7:1013-1019.
Wang, Y., and C. J. Pallen. 1991. The receptor-like protein tyrosine phosphatase HPTP alpha has two active catalytic domains with distinct substrate specificities. EMBO J. 10:3231-3237.
Wishart, M. J., J. M. Denu, J. A. Williams, and J. E. Dixon. 1995. A single mutation converts a novel phosphotyrosine binding domain into a dual-specificity phosphatase. J. Biol. Chem. 270:26782-26785.
Wu, C., M. Sun, L. Liu, and G. W. Zhou. 2003. The function of the protein tyrosine phosphatase SHP-1 in cancer. Gene 306:1-12.
Yang, Z. 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39:306-314.
Yuvaniyama, J., J. M. Denu, J. E. Dixon, and M. A. Saper. 1996. Crystal structure of the dual specificity protein phosphatase VHR. Science 272:1328-1331.(Birgit Pils and J?rg Schu)
E-mail: joerg.schultz@biozentrum.uni-wuerzburg.de.
Abstract
The protein tyrosine phosphatase (PTP) family plays a central role in signal transduction pathways by controlling the phosphorylation state of serine, threonine, and tyrosine residues. PTPs can be divided into dual specificity phosphatases and the classical PTPs, which can comprise of one or two phosphatase domains. We studied amino acid substitutions at functional sites in the phosphatase domain and identified putative noncatalytic phosphatase domains in all subclasses of the PTP family. The presence of inactive phosphatase domains in all subclasses indicates that they were invented multiple times in evolution. Depending on the domain composition, loss of catalytic activity can result in different consequences for the function of the protein. Inactive single-domain phosphatases can still specifically bind substrate and protect it from dephosphorylation by other phosphatases. The inactive domains of tandem phosphatases can be further subdivided. The first class is more conserved, still able to bind phosphorylated tyrosine residues and might recruit multiphosphorylated substrates for the adjacent active domain. The second has accumulated several variable amino acid substitutions in the catalytic center, indicating a complete loss of tyrosine-binding capabilities. To study the impact of substitutions in the catalytic center to the evolution of the whole domain, we examined the evolutionary rates for each individual site and compared them between the classes. This analysis revealed a release of evolutionary constraint for multiple sites surrounding the catalytic center only in the second class, emphasizing its difference in function compared with the first class. Furthermore, we found a region of higher conservation common to both domain classes, suggesting a new regulatory center. We discuss the influence of evolutionary forces on the development of the phosphatase domain, which has led to additional functions, such as the specific protection of phosphorylated tyrosine residues, substrate recruitment, and regulation of the catalytic activity of adjacent domains.
Key Words: protein tyrosine phosphatase ? antiphosphatase ? signaling enzymes ? functional divergence ? evolutionary site rates
Introduction
Protein tyrosine phosphatases (PTPs) regulate physiological processes common to all metazoa, including growth, differentiation, metabolism, the cell cycle, and cytoskeletal function. Together with tyrosine kinases, they control the phosphorylation state of tyrosine and serine/threonine residues of signaling proteins in highly specific reactions. The level of protein phosphorylation is highly dynamic and any disturbance can lead to severe malfunction of the eukaryotic cell. Increased level of protein phosphorylation results in abnormal proliferation and many cancer types show a mutation or deletion of a PTP gene (Siminovitch et al. 1999). In contrast to protein tyrosine kinases, which have a growth promoting potential, PTPs can act as tumor suppressors and inhibit cell growth (Dahia 2000; Wu et al. 2003). PTPs have also been implicated in B-lymphocyte and T-lymphocyte activation and insulin signaling, so that PTPs represent attractive drug targets for a wide variety of diseases, such as cancer, inflammation, diabetes, and obesity (Justement 2001; van Huijsduijnen, Bombrun, and Swinnen 2002; Asante-Appiah and Kennedy 2003).
Functionally, two types of PTPs, conserved in sequence and structure can be distinguished: The classical PTPs, which are specific for tyrosine residues, and the dual-specificity phosphatases (DSPs), which can additionally dephosphorylate serine and threonine residues.
Obviously, the catalytic residues of an enzyme are key to its molecular function. Therefore, it came as a surprise when a member of the DSP family, Sfb1, with a replacement of the catalytically essential cysteine was described (Cui et al. 1998). As expected, this protein had lost its enzymatic activity, raising the question about its molecular function. Experimentally, it could be shown that Sbf1 has maintained its ability to stably bind phosphorylated substrate, protecting the substrate from other phosphatases at this specific site. Because of its antagonistic mechanism, this phosphatase has been termed "antiphosphatase" (De Vivo et al. 1998; Hunter 1998). Sbf1 function differs also on the cellular level. In contrast to active phosphatases exposing growth inhibitory behavior, it shows transforming abilities (Cui et al. 1998). A similar substitution of the catalytic cysteine was found in another subgroup of the DSPs, called STYX (Wishart et al. 1995). Within the classical PTPs, substitutions of functional residues have been described in the receptor protein tyrosine phosphatase (RPTP) family. Most RPTPs contain two phosphatase domains, of which the second phosphatase domain is inactive or remains with very low activity because of substitutions at functional sites. Although the detailed function of this catalytically inactive domain is not yet totally understood, it is of absolute importance for the function of the receptors. Partial or entire deletion of the second domain completely abolishes or severely reduces activity of the first domain, so that a role in regulating the catalytic activity or substrate specificity of the first domain has been hypothesized (Streuli et al. 1990; Johnson et al. 1992). Furthermore, specific interaction of domain II with domain I leading to active site blocking of domain I (Bilwes et al. 1996; Majeti et al. 1998; Blanchetot and den Hertog 2000) has been shown experimentally. Inactive phosphatase domains have also been studied experimentally in the single-phosphatase domain RPTPs (Kambayashi et al. 1995; Cui et al. 1996). Here, the PTP typical signature with the catalytic cysteine is present, but two other catalytically important residues are substituted, leading to loss of phosphatase activity.
On the basis of these more anecdotal reports, we analyze here on a genomic scale how frequent substitutions of functional residues in the PTP family are and whether they were invented multiple times in evolution. Furthermore, we use site-specific evolutionary rates to unravel the evolutionary implications of these substitutions for the whole domain, leading to the prediction of distinct functional classes and the delineation of an additional functional site.
Methods
Data Sets
Protein sequences for the whole genomes of Homo sapiens (version 9.30), Mus musculus (version 9.3), Fugu rubripes (version 10.2), Anopheles gambiae (version 9.1) and Caenorhabditis elegans (version 12.95) were retrieved from the ENSEMBL Web site (www.ensembl.org). The Drosophila melanogaster predicted protein sequences (version 3) were obtained from BDGP (www.fruitfly.org), and the Ciona intestinalis predicted protein sequences (version 1.0) was obtained from JGI (www.jgi.doe.gov). These data sets were searched with Hidden Markov Models (HMMs) (HMMER: http://hmmer.wustl.edu/) specific for the classical PTPs and DSPs, respectively, which were built from the family multiple sequence alignment retrieved from SMART (http://smart.embl-heidelberg.de/). For subclassification, SMART-combined E-value thresholds were used (Schultz et al. 1998). This schema allows setting of two E-value thresholds for each subfamily HMM of a larger, homologous family. The first is given by the E-value of the best-scoring family member not belonging to the actual subfamily, the second by the worst hit before the first nonfamily member. Searching with all subfamily specific HMMs assigns a sequence to a subfamliy if its E-value is lower than the according subfamily threshold. Sequences that cannot be assigned to any subfamily but with an E-value lower than the family cutoff in at least one search are assigned to the family without any subfamily classification. These sequences were marked as "undefined specificity" and not used in the further analysis. We cleaned the obtained data set by filtering the sequences for alternative splice variants and for fragments. Only the best-scoring hit per gene was used for further analysis, and sequences that did not cover the complete HMM, tolerating an uncovered interval of 10 amino acids at the beginning and end of the profile, were excluded as they might represent fragments.
Scan for Inactive Phosphatases
For each PTP subclass, a multiple sequence alignment of all found members was created according to the family alignment in SMART, using HMMalign (HMMER). The alignments were manually curated to remove unnecessary gaps and to connect interrupted secondary structure elements, partially with the help of secondary structure information, using a representative structure from PDB (1LAR for the classical PTPs, 1VHR for the DSPs). Knowing the positions of functional residues from literature, we were able to scan the sequences for substitutions at these sites and to extract the substituted amino acid from the alignment. If these sites were occupied by gaps, absence of the functional residue was not considered in the further analysis.
Evolutionary Rate Analysis
After definition of the subclasses (D2A, D2B, membrane proximal, and cytosolic), we compared the evolutionary rates of all sites between the subclasses. Therefore, we selected the following genes from human and mouse for further analysis: Homo sapiens: ENSP00000175756, ENSP00000246887, ENSP00000248594, ENSP00000256635, ENSP00000262539, ENSP00000263708, and ENSP00000311857; Mus musculus: ENSMUSP00000022508, ENSMUSP00000025420, ENSMUSP00000027633, ENSMUSP00000029053, ENSMUSP00000029433, ENSMUSP00000030556, and ENSMUSP00000048119. The evolutionary site rates were estimated with Tree-Puzzle version 5.1 (Strimmer and von Haeseler 1996) using the quartet puzzling algorithm under the substitution model of Jones-Taylor-Thornton with an eight site-rate category discretized gamma model (Yang 1994). This model sets the average rate of a site to 1 and assigns each site to one of eight rate classes. As the rate of one class can differ between analyses of different subfamilies, we considered a site as differentially evolving if its rate was below 0.8 in the conserved subfamily and above 1.6 in the fast-evolving one. These values were chosen because within all analyses, they covered classes 1 to 4 for the conserved domain and classes 7 and 8 in the fast-evolving domain (total rates range approximately from 0 to 3 within all analyses). We performed the analyses separately for mouse and human and accepted only sites that were classified as differentially evolving in at least three of the four possible comparisons.
Results and Discussion
Amino Acid Substitutions at Functional Sites
As a first step to analyze the evolution of tyrosine phosphatases, we searched all to date sequenced metazoan genomes for phosphatase domains using specific HMMs for the classical PTPs and DSPs created from the SMART family alignments. Table 1 shows the presence of phosphatase domains in different genomes after filtering for fragments and alternative splice variants. Phosphatase domains that could not be clearly assigned to either subclass are listed as "undefined specificity" phosphatases. The number of phosphatase domains found in human, mouse, and pufferfish, which have a comparable proteome size, are similar for the total number as well as within the subtypes. Drosophila and Anopheles show the same ratio of phosphatases to the proteome as seen in vertebrates. Contrasting this, the genomes of C. elegans and Ciona intestinalis contain a substantial increase of phosphatases. Although the Ciona proteome is comparable in size to the Drosophila proteome, Ciona has more than twice as many phosphatases and almost as many as vertebrates, caused by an expansion in the tandem domain RPTPs. An even larger expansion can be observed in C. elegans, whose genome contains twice as many classical PTPs as human or mouse. Here, the multiplications fall into the class of the single domain PTPs.
Table 1 Phosphatase Domains in Metazoa.
One aim of the analysis was to investigate the extent of substitutions in functional sites within the phosphatase domain family. We focused on sites directly involved in catalysis or substrate binding, as it can be expected that substitutions in these sites will lead to a loss of catalytic activity. Within the PTP subfamily, these sites are (here and in the following, all positions regarding a classical PTP refer to pdb structure 1LAR): C-216, for attacking the phosphorous atom nucleophilically to form a phosphoenzyme intermediate and D-184 and Q-260 to position and polarize the active water molecule that dephosphorylates the phosphoenzyme in a second nucleophilic reaction, the substrate binding R-222 and the aromatic tyrosine or phenylalanine at position 49 that forms a stack with the phenyl ring of the phosphotyrosine (Stuckey et al. 1994; Fauman et al. 1996; Puius et al. 1997; Pannifer et al. 1998). Catalytic functional sites in the DSP family are (positions regarding a DSP refer to pdb structure 1VHR) C-123 and R-129 analog to the PTPs and D-91, which acts as general acid in the dephosphorylation step and as general base in the hydrolysis of the phosphoenzyme, the latter reaction is supported by S-130 (Yuvaniyama et al. 1996; Puius et al. 1997). Using a multiple sequence alignment of all obtained PTP and DSP sequences, respectively, we were able to extract substitutions at these positions.
The analysis revealed that all subclasses of the DSPs and classical PTPs, including the membrane-proximal, membrane-distal, and cytosolic domains, carry substitutions in functional positions.
Table 2 shows the distribution of phosphatase domains among subclasses for each investigated species and the number of domains with substitutions at functional sites within each subclass. The number of putative inactive domains varies strongly among the different species; however, it accounts for a significant part of all subclasses, even in the single-domain phosphatase subfamily. One might argue that the catalytically inactive proteins are nonfunctional relics in the genome, but their conservation between species and also within subfamilies, strongly indicates their functional importance. In many of the observed cases, the protein contains either a single nonactive phosphatase domain or if it contains two, both are inactive. Depending on the type of substitution, these proteins are good candidates as potential antiphosphatases.
Table 2 Number of PTP Domains and of Putative Inactive Domains Among the Different PTP Subclasses.
An extremely high portion of amino acid substitutions is found in the membrane-distal domain of RPTPs, in which all sequences feature two or more substitutions. Based on the type of substitutions, we split this group into two subfamilies. This split is supported by a phylogenetic analysis, which revealed a monophyletic origin of the subfamilies. One subclass (D2A) shows a very high degree of conservation in these substitutions, with the catalytic aspartic acid (position 184) replaced by glutamic acid and the substrate binding tyrosine (position 49) either replaced by valine or leucine. The fact that these substitutions in functional residues maintain the biochemical properties of the original amino acids suggests that this domain might be able to carry out the dephosphorylation reaction. Indeed, experiments with HPTP, a receptor type protein tyrosine phosphatase carrying a DE and YV substitution at the functional sites in the membrane-distal domain shows a small rest activity, even if D2 is expressed by itself (Wang and Pallen 1991). Still, mutation of these sites to the amino acids found in active phosphatases lead to a full recovery of catalytic activity (Lim et al. 1998). These experimental results on the one hand corroborate the functional importance of the identified sites, but on the other hand, they hint that there is still selective pressure on the catalytic center of D2A.
In the second subclass (D2B) substitutions in functional sites are more frequent and more heterogeneous than in subclass D2A. The high number of substitutions and the high variety in amino acids makes it seem unlikely that these domains maintained their catalytic activity. Indeed, Streuli et al. (1990) experimentally showed that the D2B domain of CD45 completely lost its phosphatase activity. The fact that the inactive phosphatase domain has not been lost during evolution and that orthologs are found in all species investigated in our study excludes the mutation to a nonfunctional "pseudodomain." On the contrary, because the PTP structure is still conserved, a specialization on other functions is likely to have occurred during evolution.
In summary, substitutions in functional sites reside in all subclasses of the phosphatase family. This raises the question of whether the inactive phosphatase was invented once or multiple times. If invented once, its widespread presence would indicate an origin before the split into the subclasses. Assuming this monophyletic origin would imply that inactive phophatases of different subclasses are more related to each other than to other members of the subclass, which is not the case. Furthermore, it was shown that the major subclasses evolved monophyletically (Andersen et al. 2001). Therefore, we conclude that the invention of inactive phophatases happened multiple times independently during evolution. The percentage of nonfunctional phosphatases in metazoan genomes is surprisingly high. The regulation of signaling pathways by protection of phosphorylated serine/threonine or tyrosine residues might, therefore, turn out to be an important mechanism to modulate signaling pathways. It has to be further investigated whether the phenomenon of nonfunctional enzymes is restricted to the phosphatase family or whether other signaling enzymes show a similar behavior.
Analysis of Altered Evolutionary Rates Between Phosphatase Subclasses
The strikingly high number of amino acid substitutions in functional residues in the membrane-distal phosphatase domain of receptor PTPs, which is associated with loss of activity, raises the question how on the one hand these substitutions evolved and how on the other hand they influenced the evolution of the whole domain. As a change in function should be mirrored in a change of evolutionary constraints at the involved sites, we compared the site-specific evolutionary rates between the active (cytosolic and membrane-proximal) and the inactive (D2A and D2B) PTP subclasses, a method that has been used to identify functionally important sites (Gu and Vander Velden 2002; Blouin, Boucher, and Roger 2003). Because of the lack of a representative quantity of PTP sequences in the single subclasses of most species, evolutionary rates could only be estimated reliably for human and mouse. Indeed, we found a substantial number of sites with changing evolutionary constraints (table 3). To further understand how these work together, we mapped these sites onto the structure of the PTP domain (pdb 1LAR) (fig. 1). This revealed that these sites cluster within two regions. One group is located around the catalytic center, and the other group is located on the backside of the protein. In the following section we discuss these regions separately.
Table 3 Sites with Altered Evolutionary Rates Between the PTP Subclasses.
FIG. 1. Sites with altered evolutionary rates mapped onto the tertiary structure. Functional sites and sites with altered evolutionary rates were mapped onto the pdb structure 1LAR. Catalytic and substrate-binding residues are colored blue, residues at sites with altered evolutionary rates yellow, and residues, which belong to both categories are colored green. (A) and (B) show a view on the catalytic center of the domain, (B) and (C) show the backside of the domain. (A) Sites fast evolving in D2A but conserved in the active domains. (B) Sites fast evolving in D2B but conserved in the active domains. (C) Sites more conserved in D2A than in the active domains. (D) Sites more conserved in domain D2B than in the active domains
Fast-Evolving Sites Around the Catalytic Center
The comparison of evolutionary site rates of the membrane-distal domains D2B versus the active membrane-proximal domains and the cytosolic domains revealed 15 sites that are fast evolving in domain D2B but conserved in the other domain subclasses (table 3). Most of these are located around or within the catalytic center (fig. 1). For example, one of the sites (260) is in the corresponding active domain occupied by the catalytic aspartate, and two other sites (49 and 185) are involved in substrate binding. Because domain D2B has accumulated various substitutions in its functional sites and has lost catalytic activity, it is expected that the selective constraint on the catalytic center was relaxed, which consequently allowed the surrounding sites to evolve at a faster rate. The fact that these sites are not only fast evolving in the inactive domain D2B but also conserved in the active domains in addition to their appropriate location on the surface of the domain, suggests their role in substrate binding in the active domains.
The evolutionary events in D2B after domain duplication might have been triggered by a single mutation of a functional residue, which led to a noncatalytic domain. Subsequently, the evolutionary constraint of the catalytic center was relaxed, leading to the accumulation of mutations in the surrounding.
In contrast to domain D2B, there are only five sites that are fast evolving in D2A but conserved in the active domains. One of them (position 185) is located next to the functional site 184, occupied by aspartate in the active domain but replaced by glutamate in domain D2A. This mutation might have freed the immediate surrounding from selective constraint and allowed a faster evolutionary rate at site 185. The functional residues in the catalytic center of D2A are affected by only two amino acid substitutions. The catalytic aspartate is replaced by glutamate acid and the substrate binding tyrosine either by valine or leucine. These substitutions maintain the biochemical properties, and although the catalytic activity of this domain is barely detectable, it is still able to stably bind its substrate (Bliska et al. 1992). The analysis demonstrates that the catalytic center is still under selective pressure because residues that are fast evolving in D2B and predicted to function in substrate binding are conserved in D2A. This is confirmed experimentally by regaining a catalytically fully active domain if the two substituted functional residues are converted to their original amino acids (Lim et al. 1998). We conclude that the catalytic center of D2A, in contrast to D2B, plays a pivotal role in the function of tandem domain phosphatases, leading to the question of what this function is. As the domain has lost its catalytic activity but still can bind to phosphotyrosine, one could assume two complementary molecular functions. First, the domain could function as "antiphosphatase" as described for DSPs (Cui et al. 1998). Second, it might work as an adaptor domain for phosphotyrosine substrates, similar to SH2 and PTB domains, revealing an additional function of the PTP domain family.
Slow Evolving Sites on the Backside of the Domain
The complete loss of evolutionary pressure on the catalytic center of the D2B family leads to the question of what the function of this domain is and whether there is a similar role of the D2A domains. If a new function was acquired, this should be reflected in novel conserved sites. Therefore, we searched for sites that evolve at a higher rate in the active domains while they are conserved in D2A or D2B. The comparison found 11 sites conserved in D2B and 12 sites conserved in D2A, of which five sites are found in both analyses (table 3). Almost all sites are located on the surface of the "backside" of the domain (fig. 1). This could indicate, that a new functional center has evolved in this region. The solvent exposure as well as the nature of the conserved amino acids might hint that this region is involved in protein-protein interactions. Indeed, interactions of the membrane-distal and membrane-proximal domains have been described recently. The direct interaction of the membrane-distal domain with the membrane-proximal domain stabilizes the enzyme and enhances catalytic activity (Felberg and Johnson 2000). This effect can be abrogated by deletion of the two carboxy-terminal -helices of the membrane-distal domain (244 to 278) (Johnson et al. 1992). These two helices host two residues (245 and 256), which are significantly more conserved in D2A and D2B than in the active domains, and one residue (250) that is more conserved if D2B is compared against the active domains. Another highly conserved site (240) found in both comparisons is preceding the two helices. These sites might play an important role in the interaction between the phosphatase domains. The additional sites with altered selective constraint might contribute to the stable binding but are not sufficient for stable interaction without presence of the two carboxy-terminal helices. Experimental mutation of these sites might give further insight into the molecular mechanism of regulation of RPTPs.
Together with the variation of the catalytic site, our results indicate, that domain D2A and D2B have distinct influences on the activity of membrane-proximal domains. We suggest that both domains can control activity of the first domain by interaction of residues from the "backside" of the membrane-distal domain and residues from the membrane-proximal domain. In addition, D2A can regulate substrate specificity of the membrane-proximal domain and remain associated with the substrate protein, which is accomplished by the inactive catalytic center of D2A.
The results of our analyses allow delineating a possible scenario for the evolution of the membrane-distal domain of RPTPs. The overlap of conserved residues on the "backside" of both D2A and D2B indicates that their common ancestor already had evolved this novel functional site. Whether the membrane-distal domain of the first RPTP was still active remains unclear, but the complete absence of a domain without substitutions of functional residues within the catalytic center hints that it indeed was inactive. This ancestral RPTP gave rise to one lineage with a conserved catalytic center that is still able to bind substrate (D2A) and one lineage that accumulated substitutions around the catalytic center, completely loosing the substrate binding function (D2B).
Conclusions
Our analysis of the PTP family illustrates how a closely related domain family can evolve multiple molecular functions. On the catalytic site, the family varies in the specificity of substrates, allowing the dephosphorylation of a wide range of phosphoproteins as well as phosphoinositides (Maehama and Dixon 1998). Loss of catalytic activity opens the opportunity to evolve novel functions at the catalytic site. In single-domain phosphatases, this event led to the evolution of proteins antagonizing phosphatase function, the antiphosphatases. The inactive domain in tandem domain phosphatases functions in substrate recognition and binding to multiple phosphorylated proteins. Here, it might work as a competitor for other phosphotyrosine-binding domains such as SH2 and PTB. In addition to these changes within the catalytic site, loss of catalytic activity also enabled the evolution of a novel functional site within the domain and specialization on regulatory functions. In summary, our analysis shows how evolution can create novel functionality based on an existing, well-adapted enzyme, illustrating the versatility of the PTP family.
Acknowledgements
We would like to thank Heiko Schmidt for help with the Tree-Puzzle analysis.
Literature Cited
Andersen, J. N., O. H. Mortensen, G. H. Peters, P. G. Drake, L. F. Iversen, O. H. Olsen, P. G. Jansen, H. S. Andersen, N. K. Tonks, and N. P. Moller. 2001. Structural and evolutionary relationships among protein tyrosine phosphatase domains. Mol. Cell. Biol. 21:7117-7136.
Asante-Appiah, E., and B. P. Kennedy. 2003. Protein tyrosine phosphatases: the quest for negative regulators of insulin action. Am. J. Physiol. Endocrinol. Metab. 284:E663-670.
Bilwes, A. M., J. den Hertog, T. Hunter, and J. P. Noel. 1996. Structural basis for inhibition of receptor protein-tyrosine phosphatase-alpha by dimerization. Nature 382:555-559.
Blanchetot, C., and J. den Hertog. 2000. Multiple interactions between receptor protein-tyrosine phosphatase (RPTP) alpha and membrane-distal protein-tyrosine phosphatase domains of various RPTPs. J. Biol. Chem. 275:12446-12452.
Bliska, J. B., J. C. Clemens, J. E. Dixon, and S. Falkow. 1992. The Yersinia tyrosine phosphatase: specificity of a bacterial virulence determinant for phosphoproteins in the J774A.1 macrophage. J. Exp. Med. 176:1625-1630.
Blouin, C., Y. Boucher, and A. J. Roger. 2003. Inferring functional constraints and divergence in protein families using 3D mapping of phylogenetic information. Nucleic Acids Res. 31:790-797.
Cui, L., W. P. Yu, H. J. DeAizpurua, R. S. Schmidli, and C. J. Pallen. 1996. Cloning and characterization of islet cell antigen-related protein-tyrosine phosphatase (PTP), a novel receptor-like PTP and autoantigen in insulin-dependent diabetes. J. Biol. Chem. 271:24817-24823.
Cui, X., I. De Vivo, R. Slany, A. Miyamoto, R. Firestein, and M. L. Cleary. 1998. Association of SET domain and myotubularin-related proteins modulates growth control. Nat. Genet. 18:331-337.
Dahia, P. L. 2000. PTEN, a unique tumor suppressor gene. Endocr. Relat. Cancer 7:115-129.
Dehal, P., Y. Satou, and R. K. Campbell, et al. (86 co-authors). 2002. The draft genome of Ciona intestinalis: insights into chordate and vertebrate origins. Science 298:2157-2167.
De Vivo, I., X. Cui, J. Domen, and M. L. Cleary. 1998. Growth stimulation of primary B cell precursors by the antiphosphatase Sbf1. Proc. Natl. Acad. Sci. USA 95:9471-9476.
Fauman, E. B., C. Yuvaniyama, H. L. Schubert, J. A. Stuckey, and M. A. Saper. 1996. The X-ray crystal structures of Yersinia tyrosine phosphatase with bound tungstate and nitrate: mechanistic implications. J. Biol. Chem. 271:18780-18788.
Felberg, J., and P. Johnson. 2000. Stable interdomain interaction within the cytoplasmic domain of CD45 increases enzyme stability. Biochem. Biophys. Res. Commun. 271:292-298.
Gu, X., and K. Vander Velden. 2002. DIVERGE: phylogeny-based analysis for functional-structural divergence of a protein family. Bioinformatics 18:500-501.
Hunter, T. 1998. Anti-phosphatases take the stage. Nat. Genet. 18:303-305.
Johnson, P., H. L. Ostergaard, C. Wasden, and I. S. Trowbridge. 1992. Mutational analysis of CD45: a leukocyte-specific protein tyrosine phosphatase. J. Biol. Chem. 267:8035-8041.
Justement, L. B. 2001. The role of the protein tyrosine phosphatase CD45 in regulation of B lymphocyte activation. Int. Rev. Immunol. 20:713-738.
Kambayashi, Y., K. Takahashi, S. Bardhan, and T. Inagami. 1995. Cloning and expression of protein tyrosine phosphatase-like protein derived from a rat pheochromocytoma cell line. Biochem. J. 306:331-335.
Lim, K. L., P. R. Kolatkar, K. P. Ng, C. H. Ng, and C. J. Pallen. 1998. Interconversion of the kinetic identities of the tandem catalytic domains of receptor-like protein-tyrosine phosphatase PTPalpha by two point mutations is synergistic and substrate-dependent. J. Biol. Chem. 273:28986-28993.
Maehama, T., and J. E. Dixon. 1998. The tumor suppressor, PTEN/MMAC1, dephosphorylates the lipid second messenger, phosphatidylinositol 3,4,5-trisphosphate. J. Biol. Chem. 273:13375-13376.
Majeti, R., A. M. Bilwes, J. P. Noel, T. Hunter, and A. Weiss. 1998. Dimerization-induced inhibition of receptor protein tyrosine phosphatase function through an inhibitory wedge. Science 279:88-91.
Pannifer, A. D., A. J. Flint, N. K. Tonks, and D. Barford. 1998. Visualization of the cysteinyl-phosphate intermediate of a protein-tyrosine phosphatase by x-ray crystallography. J. Biol. Chem. 273:10454-10462.
Puius, Y. A., Y. Zhao, M. Sullivan, D. S. Lawrence, S. C. Almo, and Z. Y. Zhang. 1997. Identification of a second aryl phosphate-binding site in protein-tyrosine phosphatase 1B: a paradigm for inhibitor design. Proc. Natl. Acad. Sci. USA 94:13420-13425.
Schultz, J., F. Milpetz, P. Bork, and C. P. Ponting. 1998. SMART, a simple modular architecture research tool: identification of signaling domains. Proc. Natl. Acad. Sci. USA 95:5857-5864.
Siminovitch, K. A., A. M. Lamhonwah, A. K. Somani, R. Cardiff, and G. B. Mills. 1999. Involvement of the SHP-1 tyrosine phosphatase in regulating B lymphocyte antigen receptor signaling, proliferation and transformation. Curr. Top. Microbiol. Immunol. 246:291-297.
Streuli, M., N. X. Krueger, T. Thai, M. Tang, and H. Saito. 1990. Distinct functional roles of the two intracellular phosphatase like domains of the receptor-linked protein tyrosine phosphatases LCA and LAR. EMBO J. 9:2399-2407.
Strimmer, K., and A. von Haeseler. 1996. Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13:964-969.
Stuckey, J. A., H. L. Schubert, E. B. Fauman, Z. Y. Zhang, J. E. Dixon, and M. A. Saper. 1994. Crystal structure of Yersinia protein tyrosine phosphatase at 2.5 A and the complex with tungstate. Nature 370:571-575.
van Huijsduijnen, R. H., A. Bombrun, and D. Swinnen. 2002. Selecting protein tyrosine phosphatases as drug targets. Drug Discov. Today 7:1013-1019.
Wang, Y., and C. J. Pallen. 1991. The receptor-like protein tyrosine phosphatase HPTP alpha has two active catalytic domains with distinct substrate specificities. EMBO J. 10:3231-3237.
Wishart, M. J., J. M. Denu, J. A. Williams, and J. E. Dixon. 1995. A single mutation converts a novel phosphotyrosine binding domain into a dual-specificity phosphatase. J. Biol. Chem. 270:26782-26785.
Wu, C., M. Sun, L. Liu, and G. W. Zhou. 2003. The function of the protein tyrosine phosphatase SHP-1 in cancer. Gene 306:1-12.
Yang, Z. 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39:306-314.
Yuvaniyama, J., J. M. Denu, J. E. Dixon, and M. A. Saper. 1996. Crystal structure of the dual specificity protein phosphatase VHR. Science 272:1328-1331.(Birgit Pils and J?rg Schu)