Cell. 2004 Jun 11;117(6):699-711. Protein tyrosine phosphatases in the human genome.Alonso A1, Sasin J, Bottini N, Friedberg I, Friedberg I, Osterman A, Godzik A, Hunter T, Dixon J, Mustelin T.
- Program of Signal Transduction, The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA.
Abstract
Tyrosine phosphorylation is catalyzed by protein tyrosine
kinases (TKs), which are represented by 90 genes in the human genome.
Here, we present the set of 107 genes in the human genome that encode members of the four protein tyrosine phosphatase (PTP) families.
The four families of PTPases, their substrates, structure, function, regulation, and the role of these enzymes in human disease will be discussed.
Here, we present the set of 107 genes in the human genome that encode members of the four protein tyrosine phosphatase (PTP) families.
The four families of PTPases, their substrates, structure, function, regulation, and the role of these enzymes in human disease will be discussed.
- PMID:
- 15186772
- DOI:
- 10.1016/j.cell.2004.05.018
Main Text
Introduction
Tyrosine
phosphorylation is a fundamental mechanism for numerous important
aspects of eukaryote physiology, as well as human health and disease (Hunter 1987 , Mustelin et al. 2002a,Mustelin et al. 2002b).
Compared to protein phosphorylation in general, phosphorylation on
tyrosine is extensively utilized only in multicellular eukaryotes.
Tyrosine phosphorylation is used for communication between and within
cells, the shape and motility of cells, decisions to proliferate versus
differentiate, cellular processes like regulation of gene transcription,
mRNA processing, and transport of molecules in or out of cells.
Tyrosine phosphorylation also plays an important role in the
coordination of these processes among neighboring cells in
embryogenesis, organ development, tissue homeostasis, and the immune
system. Abnormalities in tyrosine phosphorylation play a role in the
pathogenesis of numerous inherited or acquired human diseases from
cancer to immune deficiencies.
Although
it is generally agreed that tyrosine phosphorylation is regulated by
the equal and balanced action of protein tyrosine kinases (PTKs) and
protein tyrosine phosphatases (PTPs), proportionately much more research
has focused on PTKs. This is partly for historical reasons: the first
PTP was purified (Charbonneau et al., 1989) and cloned (Guan et al., 1990) ten years after the first PTK (Czernilofsky et al., 1980).
Recent findings have now led to the emerging recognition that PTPs play
specific and active, even dominant, roles in setting the levels of
tyrosine phosphorylation in cells and in the regulation of many
physiological processes (Fischer et al. 1991,Walton and Dixon 1993,Tonks and Neel 1996,Mustelinetal. 2002a,Mustelin et al. 2002b,Mustelin and Taskén 2003).
An
important step toward a better appreciation of the PTPs is the
clarification of the set of genes in the human genome that encode PTPs.
This knowledge will allow for a more global approach to key questions in
the PTP field and the regulation of tyrosine phosphorylation-dependent
cellular processes in health and disease.
The 107 Human PTP Genes
Starting
with the amino acid sequence of all published PTPs, we used conserved
catalytic motifs and whole PTP domains to search the publicly available
databases, as well as the SEED database, for genes encoding PTPs or
PTP-like genes. Two of our novel PTPs were also found in the Celera and
Incyte databases. The genomic locus, exon-intron structure, expression
data, and mouse orthologs were clarified for each PTP.
Previously
unpublished PTPs were subjected to a more thorough investigation to
verify that they represented bona fide expressed genes with orthologs in
the mouse and other organisms.
Many of these genes have also recently
been experimentally verified in our laboratories. The results of this
work are a list (Table 1)
of all genes in the human genome that encode PTP family members, here
defined as all known protein tyrosine phosphatases plus all proteins
that contain a domain homologous to the catalytic domain of these known
PTPs.
The list contains 107 genes, 105 of which have a mouse ortholog
(one of the missing genes has a rat ortholog). In addition, we estimate
that there are at least as many PTP pseudogenes and there are a few bona
fide genes (e.g., paladin, KIAA1274) that encode proteins that contain
incomplete or distant PTP-like domains.
Of the 107 human genes, 106 are
covered by expressed sequence tags (ESTs). The only gene lacking EST
coverage is DUSP27 (DUPD1), which earlier was
predicted to encode a phosphatase with an N-terminal cyclophilin-like
domain.
We have been able to amplify a partial cDNA of this gene from
testis and muscle mRNA, demonstrating that this gene, too, is expressed.
DUSP27 has four exons and is located at 10q22.3 adjacent to DUSP13 at 10q22.2, which in fact consists of two phosphatase genes, which we designate as DUSP13A and DUSP13B (encoding BEDP and TMDP, respectively), which are similar to the catalytic domain of DUSP27.
In the mouse genome, the corresponding location on chromosome 14 also
contains three PTP genes in tandem, although the sequence of DUSP27 is incomplete. In the chimpanzee genome, there is also a DUSP27
gene, which is 99% identical to the human within the four exons, but
differs in the length of some introns. Thus, it appears that an
ancestral gene was triplicated (or duplicated twice) before mice and
primates diverged during the Cretaceous, 144–165 million years ago.
Table 1The Set of PTP Genes in the Human Genome
Includes
all known PTPs in the published literature, verified PTPs or
dual-specificity phosphatases in public data bases, and ORFs found by
iterative BLAST searches with all the other sequences to find related
genes and to assign alternatively spliced forms to the correct genes. To
exclude synonyms, splice variants, and incorrectly assembled sequences,
a data base with the amino sequences was created. To exclude
pseudogenes, EST and other expression data bases were consulted, and the
exon-intron structures and chromosomal locations of every gene were
determined.
a the human DUSP13 gene contains two DSP genes: DUSP13A (BEDP; 3 exons), two noncoding exons, followed by DUSP13B (TMDP; three exons). In the mouse, the two genes are designated as separate genes.
b formerly DUPD1,
but does not contain an N-terminal cyclophilin domain, as originally
annotated. The mouse ortholog also lacks cyclophilin domain. Thus, the
name DUPD1 is a misnomer.
c we have verified its expression by PCR amplification of a partial cDNA.
d there is a rat homolog, suggesting that a mouse homolog probably does exist.
While the 38 classical PTPs have all been published (reviewed in
Andersen et al. [2004]), as many as nine of the 107 human PTP genes have not been previously reported (Supplemental Table S1 at http://www.cell.com/cgi/content/full/117/6/699/DC1),
although their genomic loci, exon/intron structure, and, in most cases,
their homology to PTPs were recognized in databases. These genes
include six small phosphatases, DUSP15, DUSP27 (DUPD1), and four genes for which we propose the genomic designations DUSP23, DUSP24, DUSP25, and DUSP26. The protein encoded by DUSP15 is a catalytically active PTP we term VHY (VH1-related member Y;Alonso et al., 2004), closely related to VHX encoded by DUSP22. The 150 amino acid residue VHZ (VH1-related member Z) encoded by DUSP25 is the smallest of all catalytically active PTPs (
Alonso et al., 2004) and is remarkably well conserved through evolution and even has a 147 amino acid homolog in Thermococcus kodakaraensis,
which is 30.3% identical and 49.7% similar over 145 residues. We have
verified the existence of a 16 kDa VHZ protein in human cells by
immunoblotting. The novel PTPs also include two myotubularin-related
proteins, for which we propose the genomic designation MTMR14 and MTMR15. The former is the human ortholog of the Sarcophaga peregrina egg-derived tyrosine phosphatase (Yamaguchi et al., 1999) and the Drosophila melanogaster gene jumpy (Seroude et al., 2002),
the mutation of which causes a muscle defect characterized by excessive
vibrations of the flight muscles (J.D. and M.J. Wishart, unpublished
observation). MTMR15 encodes a catalytically inactive member of the myotubularin group.
Why So Many PTPs?
The
number of genes in the human genome that encode members of the PTP
families is higher than anticipated and exceeds the number of genes
encoding PTKs (Manning et al., 2002).
However, a direct comparison of the two numbers, 107 and 90,
respectively, is not completely fair: of the 107 PTP genes,
11 are
catalytically inactive,
2 dephosphorylate mRNA, and
13 dephosphorylate
inositol phospholipids.
Thus, 81 PTPs are active protein phosphatases
with the ability to dephosphorylate phosphotyrosine.
Of the 90 PTKs tyrosine kinases), 85
are through to be catalytically active. Thus, the numbers of active PTPs
and PTKs are very similar and one can therefore assume that they have
comparable substrate specificities.
Both types of enzymes also display
comparable tissue distribution patterns, from ubiquitous to cell type
restricted, so that individual cells express 30%–60% of the entire
complement of PTPs and PTKs.
Neuronal and hematopoietic cells may
express even higher portions of all PTPs. That individual PTPs have
nonredundant functions is also well illustrated by the unique phenotypes
of many of the reported gene deletions in mice. For example, deletion
of PTPN22 (PTPN8 in the mouse) causes excessive expansion of memory T lymphocytes, prolonged secondary immune responses, and autoimmunity (Hasegawa et al., 2004). A polymorphism in the human PTPN22 gene also correlates with autoimmune diabetes in humans (Bottini et al., 2004).
Other human diseases caused by mutations in PTP genes (see discussion
below) also support the notion that individual PTPs have unique and
important functions.
Classification of the 107 PTPs into Four Families
Based
on the amino acid sequences of their catalytic domains, the PTPs can be
grouped into four separate families, each with a range of substrate
specificities (Figure 1).
Class I cysteine-based PTPs constitute by far the largest family and contain the 38 well-known “classical” PTPs (Andersen et al., 2004),
which are strictly tyrosine specific and all have mouse orthologs, and
the 61 VH1-like, “dual-specific” protein phosphatases (DSPs), which is
the most diverse group in terms of substrate specificity (Figure 1).
All Class I enzymes have evolved from a common ancestor, based upon
their similar structural folds for classical PTPs, DSPs, and other
VH1-like enzymes.
The single class II cysteine-based PTP in humans is a
tyrosine-specific low Mr enzyme, the origin of which appears to be more
ancient than class I PTPs: representatives of this family are found in a
all major phyla, including plants, numerous prokaryotes, and archea.
Class II PTPs are structurally related to bacterial arsenate reductases.
Class III cysteine-based PTPs are tyrosine(Y)/threonine(T)-specific
phosphatases that most likely evolved from a bacterial rhodanese-like
enzyme. They are only represented by the three p80Cdc25 cell
cycle regulators. Interestingly, a group of class I PTPs contain a
catalytically inactive rhodanese-like domain, referred to as the CDC25
homology (CH2) domain.
Despite similarities in catalytic mechanism (Guan and Dixon, 1991) and active site structure (Barford et al. 1994,Yuvaniyama et al. 1996, Su et al. 1994,Fauman et al. 1998),
class I, II, and III cysteine-based PTPs evolved independently.
Nevertheless, a structural comparison indicates that they may have
originated from an ancestral fold by rearrangment of elements.
In
contrast, the fourth family of PTPs use a different catalytic mechanism
with a key aspartic acid(D) and dependence on a cation. This family
contains the Eya proteins, which were recently discovered to be
tyrosine(Y)-, or dual serine (SS)- and tyrosine(Y)-specific protein phosphatases (
Tootle et al. 2003, Rayapureddi et al. 2003, Li et al. 2003).
Determination of the substrate specificities of the large family of
hydrolases to which the Eya PTPs belong will require extensive analysis.
Here, only the four Eya proteins are included because they were shown
experimentally to be PTPs.
Class I Cysteine-Based PTPs
The
99 class I cysteine-based family of PTPs can be further classified into
several subfamilies based on domain architecture and the degree of
homology between catalytic domains. The 38 strictly tyrosine-specific
“classical PTPs” can be divided into transmembrane, receptor-like
enzymes (RPTPs), and the intracellular, nonreceptor PTPs (NRPTPs). They
are represented in the human genome by 21 and 17 genes, respectively (Andersen et al., 2004).
As is the case for the protein kinases, the number of PTP catalytic
domains encoded by the genome is greater than the number of PTP genes
because many of the RPTPs have tandem catalytic domains.
The 61 VH1-like
enzymes are much more diverse and can be divided into several
subgroups, which share much less sequence identity with each other than
the RPTPs do with the NRPTPs. Eleven of the 61 VH1-like PTPs ( MKPs)encoded by
the human genome are specific for the mitogen-activated protein (MAP)
kinases Erk, Jnk, and p38 (
Alonso et al. 2003a,Keyse 1998, Saxena and Mustelin 2000).
These MAP kinase phosphatases (MKPs) are characterized by dual
phosphothreonine and phosphotyrosine specificity and the presence of a
CH2 region and other MAP kinase targeting motifs (Alonso et al. 2003b,Bordo and Bork 2002). Another subgroup of DSPs, which we have referred to as the “atypical” DSPs (Alonso et al., 2003b
),
includes a number of poorly characterized enzymes that lack specific
MAP kinase targeting motifs and tend to be much smaller enzymes, less
than 250 amino acid residues. The first DSP to be cloned, the VH1
protein from Vaccinia virus (Guan et al., 1991), is related to this group, as are the human VHR (Ishibashi et al., 1992) and a number of genes given the genomic designations DUSP11, DUS13, DUSP14, DUSP15, DUSP18, DUSP19, DUSP21, DUSP22, and others.
While VH1 has been reported to dephosphorylate both MAP kinases and Stat1 (Najarro et al., 2001), and VHR can dephosphorylate Erk and Jnk in 293T cells (Todd et al., 1999) and T cells (Alonso et al. 2001,Alonso et al. 2003b), it appears that many of these small atypical DSP have functions unrelated to MAP kinases. A true outlier is PIR (DUSP11), which dephosphorylates mRNA (Deshpande et al., 1999).
The
three slingshots (SSH1, SSH2, and SSH3) and the three PRLs (PRL-1,
PRL-2, and PRL-3) are very poorly understood, while the CDC14 group,
which includes KAP, is involved in dephosphorylation of the Cdk
activation loop phospho-Thr and inactivation of cyclin-dependent kinases
and in exit from mitosis (Visintin et al., 1998).
Finally, the two last subgroups of DSPs, the PTENs (5 genes) and
myotubularins (16 genes), have evolved to specifically dephosphorylate
the D3-phosphate of inositol phospholipids (Wishart and Dixon, 2002).
Enzymes with this specificity are present also in yeast. While PTEN
dephosphorylates phosphatidylinositol-3,4,5-trisphosphate at the plasma
membrane, the myotubularins primarily dephosphorylate
phosphatidylinositol-3-phosphate on internal cell membranes (Wishart and Dixon, 2002). Both groups also contain catalytically inactive members.
The Class II Cysteine-Based PTPs
This family is represented in the human genome by a single gene, referred to as ACP1,
which encodes the 18 kDa low Mr phosphatase (LMPTP). Related class II
enzymes are widely distributed in living organisms with most bacterial
genomes encoding at least one member of this family, which are
remarkably well conserved through evolution. For example, the human
LMPTP is 31% identical to the corresponding protein in Saccharomyces cerevisiae and 39% identical to the YfkJ protein in Bacillus subtilis.
Despite the paucity of tyrosine phosphorylation in prokaryotes, the
bacterial class II PTPs are bona fide tyrosine-specific phosphatases and
in many cases dephosphorylate tyrosine “autokinases” involved in
regulation of capsule polysaccharide synthesis. This may represent the
ancestral form of tyrosine phosphorylation as a mechanism by which cells
sense their extracellular environment, from which receptor PTKs and the
counterbalancing PTPs may have evolved. Although the human LMPTP can
dephosphorylate a number of tyrosine kinases and their substrates, its
physiological function is still unclear. The preservation of a class II
PTP through evolution to humans and the correlation of allelic variants
of LMPTP with many common human diseases (Bottini et al., 2002),
such as rheumatoid arthritis, asthma, diabetes, cardiomyopathy, and
Alzheimer's disease, indicate that LMPTP likely is involved in the
regulation of one or several fundamental processes in cell physiology.
The Class III Cysteine-Based PTPs
These
rhodanese-derived PTPs comprise three cell cycle regulators, CDC25A,
CDC25B, and CDC25C, in humans. Their function is to dephosphorylate Cdks
at their inhibitory dually phosphorylated N-terminal Thr-Tyr motifs, a
reaction that is required for activation of these kinases to drive
progression of cells through the cell cycle (
Honda et al., 1993).
CDC25s are themselves regulated by phosphorylation. It is curious that a
unique type of PTP evolved to serve regulation of the cell cycle,
instead of class I or class II enzymes, which presumably already existed
at the time. Interestingly, the budding yeast cell cycle can function
in the absence of Cdc28 Tyr-15 phosphorylation, and so this layer of
regulation may have been a later addition. In other words, CDC25 appears
to have entered cell cycle regulation hand in hand with tyrosine
phosphorylation of the Cdks. Alternatively, it is possible that an
ancestral CDC25 already acted as a rhodanese-type enzyme on Cdks and was
transformed into a phosphatase when Cdk tyrosine/threonine
phosphorylation evolved. This possibility would better explain why
CDC25, rather than an existing PTP, was utilized.
Asp-Based PTPs
We
have listed only the four Eya genes as members of the Asp-based PTPs
because they have recently been shown to have Tyr/Ser phosphatase
activity (Tootle et al. 2003,Rayapureddi et al. 2003,Li et al. 2003). It is clear that this is a much larger family of enzymes, which play important roles in development (Tootle et al. 2003,Rayapureddi et al. 2003,Li et al. 2003,Satow et al. 2002), sodium stress in yeast (Siniossoglou et al., 2000), and nuclear morphology (Siniossoglou et al., 1998).
RNA polymerase II C-terminal domain phosphatase is also a member of
this family. However, a clear and structurally based definition of this
family of enzymes will be needed before the human gene complement of
Asp-based phosphatases can be more accurately determined.
Modular Structure of PTPs
One of the most striking features of the PTP families (Figure 2)
is that most enzymes consist of combinations of modular domains. At
least 79 of the 107 PTPs contain at least one additional motif or domain
(Table 2)
outside of their catalytic PTP domain. Many of the domains are
protein-protein interaction or phospholipid binding modules. In this
respect, PTPs resemble the PTKs (
Manning et al., 2002),
but they differ markedly from the serine/threonine-specific protein
phosphatases. In PTKs, protein-protein interaction domains serve two
distinct purposes: regulation and targeting to substrates and/or
subcellular compartments. This appears to be the case also for PTPs,
although many of the domains in PTPs are still poorly understood. The
set of domains found in PTPs are listed in Table 2
and they include domains that bind specific domains or motifs in other
proteins (unphosphorylated or phosphorylated), cellular membranes, the
cytoskeleton, or specific phospholipids.
It is interesting to note that the set of protein domains found in PTPs is only partly overlapping with those found in PTKs (Manning et al., 2002).
While many PTKs have SH3 and SH2 domains, often in combinations, such
as SH3-SH2 or SH2-SH2, only two human PTPs (SHP1 and SHP2) have SH2
domains organized in a tandem fashion like in the Syk family PTKs. Also,
the catalytically inactive PTEN-related tensin and C1-TEN have an SH2
domain adjacent to a PTB domain. In contrast, there are no PTPs with SH3
domains. Conversely, there are no known kinases with CRAL/TRIO (Sec14p
homology), rhodanese, FYVE, or mRNA capping domains. Nevertheless, it is
interesting to note that different domains may serve similar functions
in PTPs compared to PTKs. For example, while PTKs mostly use PH domains
to interact with phosphoinositides, PTPs use CRAL/TRIO (
Huynh et al., 2003)
and FYVE domains for the same purpose. Only the myotubularins have PH
domains, but it is unclear if they function to interact selectively with
phospholipid since myotubularins mostly do not localize to the plasma
membrane. In fact, the PH domain of myotubularins was referred to as a
GRAM domain until the crystal structure of MTMR2 (
Begley et al., 2003) showed that this region folds as a PH domain.
PTPs
also use FERM domains to direct them to the cytoskeleton/plasma
membrane interface in a phosphoinositide-dependent manner, analogous to
the use of SH3, PH, and C2 domains by kinases. FERM domains may also be
able to bind phosphotyrosine, and PTPs may use catalytically inactive
PTP (“STYX”) domains in a SH2-like manner to interact with tyrosine
phosphorylated proteins (Wishart and Dixon, 1998).
Together, these differences in domain structure between PTPs and PTKs
may reflect the need to regulate these two classes of enzymes in
temporally and spatially distinct and often reciprocal manners.
Hints of Function from the Multidomain Architecture
RPTPs
In
many PTPs, the nature of their extracatalytic domains gives some
indications as to subcellular location or function. For example, all
transmembrane classical PTPs have a membrane-spanning α helix (by
definition) and are located in cellular membranes, mostly the plasma
membrane, where they interact with the extracellular milieu in a
receptor-like fashion. All but RPTPα and RPTPϵ have extended
extracellular portions with immunoglobulin, fibronectin, MAM, and
carbonic anhydrase domains. Another example is laforin, a PTP that is
mutated in an inherited form of progressive myoclonus epilepsy (Lafora's
disease;Minassian et al., 1998),
characterized histologically by accumulation of glycogen-containing
granules in the cytoplasm of cells. Recently, laforin was shown to
contain a glycogen binding domain, implying a direct role for laforin in
glycogen metabolism (
The 17 nonreceptor classical PTPs are particularly rich in
protein-protein interaction domains and many of them have several such
domains. Within this group, SHP1 and SHP2 provide a good example of how
protein-protein interaction domains can cooperate with a PTP domain to
achieve both intramolecular regulation and targeting to substrates. In
the absence of ligands for the tandem SH2 domain of these PTPs, the more
N-terminal SH2 domain folds onto the catalytic domain to block
substrate access by the insertion of a loop on the backside of the SH2
domain into the catalytic pocket of the PTP (Hof et al., 1998).
When the SH2 domains of this inhibited form of SHP1 or SHP2 encounter a
tyrosine phosphorylated ligand, the closed conformation opens and the
enzyme is activated some 100-fold (Pei et al. 1994,Pluskey et al. 1995).
Under physiological conditions, SH2 domain ligand binding also
juxtaposes the PTP domain to its substrates, which contain, or associate
with, the phosphorylated SH2 domain ligand.
The DSP
subfamily contains 11 members with a CH2 domain, a region derived from
the bacterial rhodanese enzyme and adapted to a MAP kinase docking motif
(Alonso et al. 2003b,
Bordo and Bork 2002). In CDC25, on the other hand, the rhodanese has evolved into a catalytic PTP domain (Bordo and Bork, 2002).
Other noncatalytic domains and motifs found in DSPs include FYVE,
glycogen binding, mRNA capping (guanyl methyltransferase), and consensus
recognition motifs for N-terminal myristylation or C-terminal
prenylation. PTEN contains a C2 domain tightly packed against the PTP
domain, while members of the myotubularin family contain PH, FYVE,
coiled-coil, and PDZ binding motifs (Wishart and Dixon, 2002).
A number of additional protein-protein interaction domains are found in
PTPs in other organisms, for example a WW domain in a C. elegans PTP (Sudol, 1996), but are not present in the human PTPs either because they have been lost or because they evolved in a separate lineage.
Large Multidomain PTPs versus Multisubunit Ser/Thr Phosphatases
The
multidomain structure of most PTPs is in sharp contrast to the
serine/threonine protein phosphatases, which generally consist of small
catalytic subunits that bind regulatory or targeting subunits encoded by
separate genes to form a large number of distinct holoenzymes with
different biological functions. This dichotomy may explain the
differences in gene numbers: 107 PTPs to counteract 90 PTKs versus many
fewer catalytic Ser/Thr phosphatase subunits to counter 428 protein
kinases and the extensive phosphorylation of more than a third of all
cellular proteins. Presumably, the combinatorial subunit principle of
serine/threonine protein phosphatases can generate much more diversity
and flexibility, but may lack some of the strict specificity and tight
regulation possible with single-chain multidomain PTPs. In this context,
it may also be significant that the atypical DSP subgroup of PTPs
contains many small enzymes devoid of other domains or motifs. It is
possible that these enzymes participate in multisubunit complexes where
other subunits provide regulation and substrate targeting. However, no
examples of this have yet been found.
Catalytically Inactive PTP Domains
Among
the class I cysteine-based PTPs, there are at least 14 catalytically
inactive PTP domains, in which critical catalytic residues have been
altered. These are the second (D2) domains of CD45, RPTPγ, and RPTPζ,
and the only PTP domains of the secretory vesicle-located receptor-like
PTP IA-2 (Solimena et al., 1996), a VHR-like protein termed STYX (Wishart et al., 1995), the MKP-like MK-STYX (Wishart and Dixon, 1998), the PTEN-related tensin and C1-ten, and the myotubularin-related proteins MTMR5, MTMR9, MTMR11, MTMR12, MTMR13 (Wishart and Dixon, 2002),
and MTMR15. Despite lack of catalytic activity, many of these proteins
still play important roles in cells, in many cases by partnering with
active PTPs. For example, CD45 needs its inactive D2 domain for
dephosphorylation by D1 of its physiological substrates (Kashio et al., 1998).
Similarly, the inactive MTMRs seem to act as important regulatory
subunits for active members of the group, as shown by the human disease
Charcot-Marie-Tooth type 4B, which is caused by mutation of the
catalytically active MTMR2 (Bolino et al., 2000) or the catalytically inactive MTMR13 (
Azzedine et al., 2003),
in both cases with identical pathology and symptoms. It turns out that
these two proteins form a heterodimer, in which the activity of MTMR2
depends on MTMR13. There are similar examples where an inactive kinase
domain is needed to activate a catalytically competent kinase domain;
e.g., LKB1 is activated by STRAD (Baas et al., 2003).
In addition to these inherently inactive PTP domains, alternative
splicing of PTP transcripts (which occurs in at least half of all PTPs)
can create isoforms that have catalytically inactive PTP domains (Tailor et al., 1999
), or lack them altogether (Bult et al., 1997).
Physiologically important alternative splicing within the catalytic
domains is known to create two isoforms of the class II enzyme LMPTP
with different specific activity, substrate specificity, and regulation (Mitchell Bryson et al., 1995).
In this enzyme, the ratio of the two isoforms is genetically determined
and subject to allelic variation, which correlates with the
predisposition to several common human diseases (Bottini et al., 2002).
Posttranslational Modifications and Reversible Oxidation of PTPs
The
majority of PTPs are posttranslationally modified. Glycosylation
appears to be restricted to the transmembrane PTPs, which contain
abundant N- and O-linked carbohydrates in their extracellular portions.
Two groups of DSPs are modified by fatty acids: VHY (DUSP15) and VHX (DUSP22) are N-terminally myristylated in a manner reminiscent of Src family PTKs (
Alonso et al., 2004),
while the PRLs are C-terminally farnesylated similar to Ras GTPases. By
far the most common modification is the phosphorylation of PTPs on
serine, threonine, and tyrosine. There are many examples where Ser/Thr
phosphorylation has been shown to regulate activity. The phosphorylation
of PTPs on tyrosine is particularly intriguing as it implies a
physical, if not functional, interaction with PTKs (Mustelin and Hunter, 2002).
At least 15 PTPs have been reported to be tyrosine phosphorylated, but
the physiological significance remains unclear in most cases. Tyrosine
phosphorylation also introduces the possibility of
autodephosphorylation. In fact, many PTPs become noticeably more
phosphorylated on tyrosine when expressed as catalytically inactive
mutants. In most cases, however, the phosphorylation sites are clearly
inaccessible for intramolecular dephosphorylation, suggesting that
dephosphorylation must occur in trans. Thus, regulatory
networks of PTPs and other protein phosphatases, perhaps in the form of
“phosphatase cascades,” may exist in cells.
It
has been known for many years that the catalytic cysteine of class I
and II PTPs is highly susceptible to oxidation in vitro, and it has been
speculated that this could play a role in intact cells exposed to
oxidizing agents. An exciting new development was the discovery that
oxidation of PTP1B does not result in a stable sulfenic acid derivative
of the catalytic site cysteine, but rapidly transforms into a sulfenyl
amide ring involving the adjacent serine residue (Salmeen et al. 2003,
van Montfort et al. 2003).
This form is resistant to further oxidation, which would be
irreversible, and is readily reduced back to the free cysteinyl,
catalytically active form. Together with recent insights into the
production of reactive oxygen species and nitrogen oxides in cells
during growth factor stimulation in cancer, inflammation, and
neurodegenerative diseases, it now seems likely that reversible redox
regulation of PTPs occurs in intact cells. Since PTPs often play
dominant roles in setting the levels of tyrosine phosphorylation in
cells, this regulation may be physiologically very important.
PTPs as Drug Targets
Compared
to the PTKs, many inhibitors (TKI) of which already are in clinical trials,
the PTPs are newcomers in the field of drug development. The effect of
disruption of the PTP1B gene in mice, which indicated that this PTP acts as a negative regulator of insulin signaling (Elchebly et al., 1999),
ignited the interest of the pharmaceutical industry. Recent discoveries
that many other PTPs also play critical roles in a variety of human
disease (Table 3)
have sparked a growing interest in PTPs as drug targets. In addition to
metabolic, neurological, muscle, and autoimmune diseases, at least 30
PTPs have been implicated in cancer (listed inAndersen et al. [2004]).
We expect that with improved understanding of the molecular mechanisms
by which PTPs affect the pathophysiology of these diseases, combined
with the dominant role that PTPs often play in the regulation of
tyrosine phosphorylation-dependent processes, PTP inhibitors will become
clinically relevant therapeutics in the future. A rational design of
small-molecule inhibitors against PTPs using in silico docking,
NMR-based methods, high-throughput crystallization, and specific
chemistry will most likely be involved. Following initial concerns about
specificity and problems associated with the hydrophilicity of
phosphomimetics, promising successes have been achieved by
structure-based drug design, particularly those that utilize unique
features of the surface topology surrounding the catalytic pocket of
each PTP. Figure 3
shows the surface topology and charge distribution of a representative
set of catalytic PTP domains to illustrate the striking diversity within
the class I PTPs. For example, for PTP1B it was found that a unique
second PTyr binding site (Puius et al. 1997, Salmeen et al. 2000) could be used to develop highly specific bidentate inhibitors that bind both sites (Zhang, 2002). This general approach can be used to design highly specific and effective inhibitors.
| PTPRC (CD45) | SCID (Kung et al., 2000; ), multiple sclerosis ( ) |
|---|
| PTPRN (IA-2) | Antigen for autoimmune diabetes ( | ||
| PTPRN2 (phogrin) | Antigen for autoimmune diabetes | ||
| PTPN1 (PTP1B) | Insulin resistance, obesity | ||
| PTPN6 (SHP1) | Sezary syndrome ( | ||
| PTPN9 (PTP-MEG2) | Autism (Smith et al., 2000) | ||
| PTPN11 (SHP2) | Noonan syndrome | ||
| PTPN22 (LYP) | SNP polymorphism in type I diabetes | ||
| PTEN (PTEN) | Bannayan-Zonana (Marsh et al., 1997), Cowden syndrome and Lhermitte-Duclos disease | ||
| MTM1 (myotubularin) | X-linked myotubular myopathy ( | ||
| MTMR2 (MTMR2) | Charcot-Marie-Tooth syndrome type 4B ( | ||
| MTMR13 (MTMR13) | Charcot-Marie-Tooth syndrome type 4B ( | ||
| EPM2A (laforin) | Progressive myoclonus epilepsy (Lafora's disease) ( | ||
| ACP1 (LMPTP) | Polymorphism correlates with many common diseases ( |
107 PTP genes
| A. Class I Cys-Based PTPs (99 Genes) | ||||
| A. 1. Classical PTPs (38 Genes) | ||||
| A. 1. 1. Transmembrane Classical PTPs (21 Genes) | ||||
| 1. PTPRA | RPTPα | 20p13 | YES | YES |
| 2. PTPRB | RPTPβ | 12q15-q21 | YES | YES |
| 3. PTPRC | CD45, LCA | 1q31-q32 | YES | YES |
| 4. PTPRD | RPTPδ | 9p23-p24.3 | YES | YES |
| 5. PTPRE | RPTPϵ | 10q26 | YES | YES |
| 6. PTPRF | LAR | 1p34 | YES | YES |
| 7. PTPRG | RPTPγ | 3p21-p14 | YES | YES |
| 8. PTPRH | SAP1 | 19q13.4 | YES | YES |
| 9. PTPRJ | DEP1, CD148, RPTPη | 11p11.2 | YES | YES |
| 10. PTPRK | RPTPκ | 6q22.2-23.1 | YES | YES |
| 11. PTPRM | RPTPμ | 18p11.2 | YES | YES |
| 12. PTPRN | IA-2, Islet cell antigen 512 | 2q35-q36.1 | YES | YES |
| 13. PTPRN2 | PTPRP, RPTPπ, IA-2β, phogrin, | 7q36 | YES | YES |
| 14. PTPRO | GLEPP1/PTP-U2/PTPROτ isoforms A/B/C | 12p13.3-p13.2 | YES | YES |
| 15. PTPRQ | PTPS31 | 12q21.31 | YES | YES |
| 16. PTPRR | PTP-SL, PCPTP,PTPBR7, PC12-PTP1 | 12q15 | YES | YES |
| 17. PTPRS | RPTPσ | 19p13.3 | YES | YES |
| 18. PTPRT | RPTPρ | 20q12-q13 | YES | YES |
| 19. PTPRU | PTPJ/PTP-U1/PTPRomicron isoforms 1/2/3 | 1p35.3-p35.1 | YES | YES |
| 20. PTPRV | OST-PTP | 1q32.1 | YES | YES |
| 21. PTPRZ | RPTPζ | 7q31.3 | YES | YES |
| A. 1.2. NRPTPs (17 Genes), inracellular | ||||
| 22. PTPN1 | PTP1B | 20q13.1-13.2 | YES | YES |
| 23. PTPN2 | TCPTP, MPTP, PTP-S | 18p11.3-11.2 | YES | YES |
| 24. PTPN3 | PTPH1 | 9q31 | YES | YES |
| 25. PTPN4 | PTP-MEG1, TEP | 2q14.2 | YES | YES |
| 26. PTPN5 | STEP | 11p15.1 | YES | YES |
| 27. PTPN6 | SHP1, PTP1C, SH-PTP1, HCP | 12p12-13 | YES | YES |
| 28. PTPN7 | HePTP, LCPTP | 1q32.1 | YES | YES |
| 29. PTPN9 | PTP-MEG2 | 15q23 | YES | YES |
| 30. PTPN11 | SHP2, SH-PTP2, Syp, PTP1D, PTP2C, SH-PTP3 | 12q24.1 | YES | YES |
| 31. PTPN12 | PTP-PEST, PTP-P19, PTPG1) | 7q11.23 | YES | YES |
| 32. PTPN13 | PTP-BAS, FAP-1, PTP1E, RIP, PTPL1, PTP-BL | 4q21.3 | YES | YES |
| 33. PTPN14 | PTP36, PEZ, PTPD2 | 1q32.2 | YES | YES |
| 34. PTPN18 | PTP-HSCF, PTP20, BDP | 2q21.2 | YES | YES |
| 35. PTPN20 | TypPTP | 10q11.22 | YES | YES |
| 36. PTPN21 | PTPD1, PTP2E, PTP-RL10 | 14q31.3 | YES | YES |
| 37. PTPN22 | LYP, PEP | 1p13.3-p13.1 | YES | YES |
| 38. PTPN23 | HD-PTP, HDPTP, PTP-TD14, KIAA1471, DKFZP564F0923 | 3p21.3 | YES | YES |
| A. 2. DSPs or VH1-like (61 Genes) A. 2. 1. MKPs (11 Genes) | ||||
| 39. DUSP1 | MKP-1, 3CH134, PTPN10, erp, CL100/ HVH1 | 5q34 | YES | YES |
| 40. DUSP2 | PAC-1 | 2q11 | YES | YES |
| 41. DUSP4 | MKP-2, hVH2/TYP1 | 8p12-p11 | YES | YES |
| 42. DUSP5 | hVH3/B23 | 10q25 | YES | YES |
| 43. DUSP6 | PYST1, MKP-3/rVH6 | 12q22-q23 | YES | YES |
| 44. DUSP7 | PYST2, B59, MKP-X | 3p21 | YES | YES |
| 45. DUSP8 | hVH5, M3/6, HB5 | 11p15.5 | YES | YES |
| 46. DUSP9 | MKP-4, Pyst3 | Xq28 | YES | YES |
| 47. DUSP10 | MKP-5 | 1q41 | YES | YES |
| 48. DUSP16 | MKP-7, MKP-M | 12p13 | YES | YES |
| 49. MK-STYX | MK-STYX | 7q11.23 | YES | YES |
| A. 2. 2. Atypical DSPs (19 Genes) | ||||
| 50. DUSP3 | VHR, T-DSP11 | 17q21 | YES | YES |
| 51. DUSP11 | PIR1 | 2p13.1 | YES | YES |
| 52. DUSP12 | HYVH1, GKAP, LMW-DSP4 | 1q21-q22 | YES | YES |
| 53. DUSP13A | BEDP | 10q22.2 | YES | YES |
| 54. DUSP13B | TMDP, TS-DSP6 | 10q22.2 | YES | YES |
| 55. DUSP14 | MKP6, MKP-L | 17q12 | YES | YES |
| 56. DUSP15 | VHY, Q9H1R2 | 20q11.21 | YES | YES |
| 57. DUSP18 | DUSP20, LMW-DSP20 | 22q12.2 | YES | YES |
| 58. DUSP19 | DUSP17, SKRP1, LDP-2, TS-DSP1 | 2q32.1 | YES | YES |
| 59. DUSP21 | LMW-DSP21, BJ-HCC-26 tumor antigen | Xp11.4-p11.23 | YES | YES |
| 60. DUSP22 | VHX, MKPX, JSP1, LMW-DSP2, TS-DSP2, JKAP | 6p25.3 | YES | YES |
| 61. DUSP23 | MOSP, similar to RIKEN cDNA 2810004N20 | 11p11.2 | YES | YES |
| 62. DUSP24 | MGC1136 | 8p12 | YES | YES |
| 63. DUSP25 | VHZ, FLJ20442, LMW-DSP3 | 1q23.1 | YES | YES |
| 64. DUSP26 | VHP, “similar to RIKEN cDNA 0710001B24” | 2q37.3 | YES | YES |
| 65. DUSP27 | DUPD1, FMDSP, “similar to cyclophilin” | 10q22.3 | NO | YES |
| 66. EPM2A | Laforin | 6q24 | YES | YES |
| 67. RNGTT | mRNA capping enzyme | 6q16 | YES | YES |
| 68. STYX | STYX | 14 | YES | YES |
| A. 2. 3. Slingshots (3 Genes) | ||||
| 69. SSH1 | SSH1, slingshot 1 | 12q24.12 | YES | YES |
| 70. SSH2 | SSH2, slingshot 2 | 17q11.2 | YES | YES |
| 71. SSH3 | SSH3, slingshot 3 | 11q13.1 | YES | YES |
| A. 2. 4. PRLs (3 Genes) | ||||
| 72. PTP4A1 | PRL-1 | 6q12 | YES | YES |
| 73. PTP4A2 | PRL-2, OV-1 | 1p35 | YES | YES |
| 74. PTP4A3 | PRL-3 | 8q24.3 | YES | YES |
| A. 2. 5. CDC14s (4 Genes) | ||||
| 75. CDC14A | CDC14A | 1p21 | YES | YES |
| 76. CDC14B | CDC14B | 9q22.33 | YES | YES |
| 77. CDKN3 | KAP | 14q22 | YES | YES |
| 78. PTP9Q22 | PTP9Q22 | 9q22.32 | YES | YES |
| A. 2. 6. PTENs (5 Genes) | ||||
| 79. PTEN | PTEN, MMAC1, TEP1 | 10q23.3 | YES | YES |
| 80. TPIP | TPIPα, TPTE and PTEN homologous | 13q12.11 | YES | YES |
| 81. TPTE | PTEN-like, PTEN2 | 21p11 | YES | YES |
| 82. TNS | Tensin | 2q35-q36 | YES | YES |
| 83. TENC1 | C1-TEN, TENC1, KIAA1075 | 12q13.13 | YES | YES |
| A. 2. 7. Myotubularins (16 Genes) | ||||
| 84. MTM1 | myotubularin | Xq28 | YES | YES |
| 85. MTMR1 | MTMR1 | Xq28 | YES | YES |
| 86. MTMR2 | MTMR2 | 11q22 | YES | YES |
| 87. MTMR3 | MTMR3, FYVE-DSP1 | 22q12.2 | YES | YES |
| 88. MTMR4 | MTMR4, FYVE-DSP2 | 17q22-q23 | YES | YES |
| 89. MTMR5 | MTMR1, SBF1 | 22q13.33 | YES | YES |
| 90. MTMR6 | MTMR6 | 13q12 | YES | YES |
| 91. MTMR7 | MTMR7 | 8p22 | YES | YES |
| 92. MTMR8 | MTMR8 | Xq11.2 | YES | NO |
| 93. MTMR9 | MTMR9, LIP-STYX | 8p23-p22 | YES | YES |
| 94. MTMR10 | MTMR10 | 15q13.1 | YES | NO |
| 95. MTMR11 | MTMR11CRA α/β | 1q12.3 | YES | YES |
| 96. MTMR12 | MTMR12, 3-PAP | 5p13.3 | YES | YES |
| 97. MTMR13 | MTMR13, SBF2, CMT4B2 | 11p15.3 | YES | YES |
| 98. MTMR14 | FLJ22075, hJumpy, hEDTP | 3p26 | YES | YES |
| 99. MTMR15 | KIAA1018 | 15q13.1 | YES | YES |
| B. Class II Cys-Based PTPs (1 Gene) | ||||
| 100. ACP1 | LMPTP, low Mr PTP, LMWPTP, BHPTP | 2p25 | YES | YES |
| C. Class III Cys-Based PTPs (3 Genes) | ||||
| 101. CDC25A | CDC25A | 3p21 | YES | YES |
| 102. CDC25B | CDC25B | 20p13 | YES | YES |
| 103. CDC25C | CDC25C | 5q31 | YES | YES |
| D. Asp-Based PTPs (4 Genes) | ||||
| 104. EYA1 | Eya1 | 8q13.3 | YES | YES |
| 105. EYA1 | Eya2 | 20q13.1 | YES | YES |
| 106. EYA1 | Eya3 | 1p36 | YES | YES |
| 107. EYA1 | Eya4 | 6q23 | YES | YES |
| 107. Total | ||||
Inga kommentarer:
Skicka en kommentar