Protein tyrosine phosphatases as wardens of STAT signaling

Abstract

Signaling by signal transducers and activators of transcription (STATs) is controlled at many levels of the signaling cascade. Protein tyrosine phosphatases (PTPs) regulate STAT activation at several layers, including direct pSTAT dephosphorylation in both cytoplasm and nucleus. Despite the importance of this regulation mode, many aspects are still incompletely understood, e.g., the identity of PTPs acting on certain members of the STAT family. After a brief introduction into the STAT and PTP families, we discuss here the current knowledge on PTP mediated regulation of STAT activity, focusing on the interaction of individual STATs with specific PTPs. Finally, we highlight open questions and propose important tasks of future research.

General Introduction

The JAK-STAT pathway is a fundamental intracellular signaling cascade that transduces multiple signals of crucial importance for development and homeostasis in animals. While its most elaborate versions are operative in mammals, elementary JAK-STAT systems also exist in Drosophila melanogaster and even in the slime mold Dictyostelium discoideum,¹ pointing to an early origin in evolutionary terms.

In higher animals, the JAK-STAT pathway is central for the transmission of signals from numerous cytokine and growth factor receptors to the cell nucleus. STAT activation by phosphorylation through Janus kinases and other cytoplasmic tyrosine kinases mediates cell proliferation, differentiation, cell migration and apoptosis. These processes are essential determinants for the development of blood and immune cells as well as of many other tissues throughout the organism (reviewed in refs. 2 and 3). Not surprisingly, aberrant STAT signaling is associated with immune disorders and neoplasias of both hematopoietic and solid tissues.

The canonical JAK-STAT signaling pathway employs only few components and relies on relatively simple mechanisms. Basically, STAT proteins are first recruited to activated receptors via interaction between the STATs’ SH2 domains and tyrosine phophorylated docking sites. Next, STATs become phosphorylated themselves at a critical, conserved tyrosine residue, leading to their release from the receptor complex and dimerization of STATs through reciprocal contact between SH2 domains and phosphotyrosine moieties and their local environment. As a consequence, these so-called “parallel” STAT dimers become capable of crossing the nuclear membrane and reach chromatin, where they bind specific cognate DNA elements and participate in complex gene regulation processes (for a recent review see ref. 4). Obviously, the JAK-STAT pathway represents an immediate means of linking extracellular signals to a transcriptional response and, hence, requires tight control. While mechanisms underlying STAT tyrosine phosphorylation have been extensively characterized, less work has been devoted to the control of STAT-mediated signaling by dephosphorylation. These reactions, however, are essential to ensure proper amplitudes and kinetics of STAT activation.⁵ Dephosphorylation adds a layer of complexity to the control of JAK-STAT pathways in that it can specifically direct physiological signals by addressing different phosphotyrosine bearing motifs of STATs or the activating receptors and kinases. Moreover, it involves different protein tyrosine phosphatases (PTPs) in different contexts. In line with the importance of PTP-mediated regulation, failure of correct PTP function is associated with characteristic states of disease.⁶

In this review, we will give an overview of dephosphorylation-based mechanisms regulating JAK-STAT signaling. After a brief discussion of upstream events such as the control of receptor and JAK phosphorylation, we focus on reactions which directly affect the STAT proteins, limiting the discussion to phosphotyrosine dephosphorylation. Reversible serine phosphorylation of STATs also contributes to gene regulation,⁷ but there is as yet only little information about underlying mechanisms.

We will first provide a brief introduction to STAT and PTP families. We will then discuss PTP-mediated regulation at different levels, with main emphasis on regulation of the individual STAT family members and, in the concluding section, highlight open questions and important areas of future research.

The STATs and Their Activation–Inactivation Cycle

The mammalian family of STAT factors consists of seven members (STAT1, STAT2, STAT3, STAT4, STAT5A and B, and STAT6). Activation of STATs (Fig. 1) is primarily driven by ligand-stimulated cytokine receptors, and, though described for fewer examples, by receptor tyrosine kinases (RTKs). The intracellular domains of cytokine receptors, which can come as homo- or heterodimers in their activated state, are associated with Janus kinases (JAKs). JAKs represent a family of cytoplasmic tyrosine kinases which comprises four mammalian members (JAK1, JAK2, JAK3, and TYK2), and each cytokine receptor subunit recruits with preference one of the JAKs. Upon ligand-induced receptor crosslinking, JAKs become activated by juxtaposition and mutual tyrosine phosphorylation, which leads to JAK-mediated phosphorylation of receptor-borne tyrosine residues. As a result, phosphotyrosine (pTyr)-based docking sites for SH2 domains are formed, a prerequisite for the recruitment of specific STAT proteins. STATs subsequently become tyrosine phosphorylated by the persistent JAK activity within the signaling complex. Via pTyr–SH2 domain interaction, STATs are engaged in homo- or heterodimerization and consequently become released from the receptor and undergo conformational changes. As a consequence, nuclear localization signals become accessible, enabling STAT dimers to translocate into the nucleus via the Ran nuclear import pathway, to bind to palindromic cognate sequence elements in the DNA and to participate in transcriptional regulation (for a recent review see ref. 4).

Figure 1. JAK-STAT signaling is regulated by protein tyrosine phosphatases at several levels. Activation of cell surface receptors leads to activation of associated Janus- or TYK2-kinases (designated JAK), and in turn tyrosine phosphorylation of the receptors, and recruitment and tyrosine phosphorylation of STAT molecules. Different protein tyrosine phosphatases, such as the transmembrane PTPs CD45 and PTPε, as well as the cytoplasmic PTPs PTP1B, TCPTP, and SHP-1 can attenuate JAK activation by dephosphorylation. SHP-1 (and the related PTP SHP-2) can associate through their SH2 domains with phosphorylated cytokine receptors, and can modulate JAK activation, and also further signaling pathways initiated by ligand stimulation (not shown). Tyrosine phosphorylated STATs (denoted pSTAT) form dimers, which are capable of translocation to the nucleus, by reciprocal interaction of the (C-terminally located) SH2-domains with the phosphotyrosine motifs. Dephosphorylation of pSTATs by PTPs occurs in the cytoplasm and nucleus, leading to control of amplitude and duration of STAT signaling. Dephosphorylated STAT molecules become available for a new cycle of activation.

The canonical JAK-STAT signaling pathway outlined above is refined by additional reactions affecting or involving STATs. For instance, additional posttranslational modifications such as serine phosphorylation,⁷ acetylation,⁸ and sumoylation⁹ modulate rate and specificity of target gene regulation or protein stability (Table 1). Notably, STATs can also undergo dimerization and oligomerization processes via contacts between their N-terminal protein–protein interaction domains. By this mechanism, tyrosine phosphorylated STAT tetramers can form, but also dimers independently of phosphorylation. Unphosphorylated STAT1 (U-STAT1) is found in the cell nucleus, and while the mechanisms underlying nuclear-cytoplasmic shuttling of U-STAT1 are only vaguely understood, U-STAT1 was shown to exert profound influence on the pattern of regulated transcripts.¹⁰

Table 1. Mammalian STAT family members and reported mechanisms of regulatory modification.

Name	Modification, specific biochemical features, and functionsa,b
STAT1	Acetylation/deacetylation affects IFN-induced tyrosine phosphorylation and is involved in specific gene regulation; serine phosphorylation at S727 in response to IL-6-type cytokines and S708 in response to IFNs, methylation of STAT1 on R31; SUMOylation of STAT1 on K703 by PIAS1 inhibits its activation, favors nuclear dephophorylation and modulates DNA binding activity; C-terminally truncated STAT1 β isoform is presumably generated by alternative splicing
STAT2	Acetylation/deacetylation; serine phosphorylation at S287 regulates type I interferon-induced cellular responses; ubiquitinylation mediated by viral gene products; alternative splicing resulting in a truncated β isoform
STAT3	Acetylation/deacetylation on K685 by acetyltransferase p300 is involved in the control of growth and glucose metabolism; Serine phosphorylation at S727; STAT3 undergoes phosphorylation by pyruvate kinase M2, promoting cell proliferation in response to alterations in glucose metabolism; Dimethylation/demethylation on K140 in depending on IL-6 signaling; Formation of C-terminally truncated STAT3 β and γ isoforms, presumably by alternative splicing and proteolytic processing, respectively
STAT4	Alternative splicing resulting in a C-terminally truncated STAT4 β isoform
STAT5A	Serine phosphorylation is critical for malignant transformation; tetramerization drives transcription of specific gene sets and participates in displacement of STAT3 from cognate STAT binding sites in DNA, O-glycosylation modulates interaction with transcriptional co-activators; C-terminally truncated β and γ isoforms, presumably arising from alternative splicing and proteolitic processing, respectively,
STAT5B	Acetylation functionally antagonizes SUMOylation; alternative splicing results in β isoform, limited proteolysis in γ isoform; tetramerization
STAT6	Acetylation/deacetylation; serine phosphorylation at S407 upon virus infection, involving STING (ERIS) protein and STAT6 recruitment to the ER; potentially dominant negative β and γ isoforms

^aModifications in addition to tyrosine phosphorylation. ^bInformation is based on literature cited in the text and on the following articles and reviews: refs. 99–103

Equally important as the described steps of activation are mechanisms that limit and attenuate JAK-STAT signaling. They are operative at different stages along the time line of signal transmission from the activated receptor to target gene expression. Phosphorylation of receptors and Janus kinases, in effect seconds after receptor crosslinking, are being counteracted by PTPs, e.g., the SH2 domain containing tyrosine phophatases SHP-1 (PTPN6) and SHP-2 (PTPN11, Table 2) which are recruited to pTyr bearing SH2 recognition motifs in cytoplasmic receptor domains.¹¹ SOCS (suppressors of cytokine signaling) and PIAS proteins (protein inhibitors of activated STAT) are components of negative feedback loops, acting as inhibitors of JAKs or STATs or directing them to protein degradation pathways.⁴ The expression of SOCS proteins is strongly induced by STAT-mediated gene regulation. They can directly bind to JAKs or to the cytoplasmic domains of cytokine receptors, where they inhibit JAK binding and activity or the recruitment of STATs. SOCS proteins are also involved in the degradation of JAKs and STATs through the ubiquitin-proteasome pathway.¹² Another protein connecting STATs to ubiquination and subsequent degradation is SLIM, which acts as a ubiquitin E3 ligase, particularly targeting STAT1 and STAT4. SLIM, in addition, was also shown to inhibit STAT tyrosine phosphorylation.¹³ PIAS proteins inhibit STATs by specific interactions. PIAS proteins are constitutively expressed, their suppression of STAT function relies on blockade of the STATs’ DNA binding sites, or by functioning as ligases to subject STATs to SUMOylation and to funnel them into the SUMO-mediated degradation pathway.^4,9

Table 2. PTP family members with relevance for regulation of JAK-STAT signaling.

Systematic name (also gene name)	Common synonyms	Subfamilya	Protein, alternate gene productsb,c	Regulation, specific biochemical features, and functionsc
PTPN1	PTP1B	Non-transmembrane PTP family (NT) 1	435 aa	Anchored through C-terminal peptide to cytoplasmic face of endoplasmic reticulum, released by C-terminal truncation through Ca2+/calpeptin-dependent cleavage; regulated by reversible oxidation and phosphorylation; PTP for JAK2 and TYK2, STAT6, and several RTKs (IR, EGFR, PDGFR); can activate SRC-family kinases; knockout mice are resistant to high-fat diet, and exhibit increased insulin and leptin sensitivity; promotes tumorigenesis in mouse models of ERB2/neu-driven breast cancer
PTPN2	TCPTP	NT1	415 aa (48 kDa, TCPTP48); 387 (45 kDa, TCPTP45)	TCPTP48 associated to endoplasmic reticulum; TCPTP45 has an NLS and is nuclear or cytoplasmic, shuttling in response to cell stimulation, in mice, only TCPTP45 is expressed; PTP for several STATs, JAK1, and JAK3, and several RTKs (EGFR, PDGFRβ); knockout mice die between 3–5 wk of age from anemia and systemic inflammation; inactivating mutations found in ALL patients
PTPN6	SHP-1, SH-PTP1, HCP, PTP1C	NT2	Hematopoietic (595 aa), epithelial (597 aa), long form (SHP-1L, 624 aa)	2 SH2 domains, negative regulation by SH2-domain – catalytic-domain interaction, C-terminus with regulatory function (lipid binding, pTyr residues) contains NLS; cytoplasmic to nuclear translocation observed; negative regulator of cytokine, growth factors and immunoreceptor signaling (e.g., Epo, CSF-1, BCR); substrates receptors, adaptor proteins, JAKs; naturally occurring mouse strains with full (motheaten) or partial (viable motheaten) inactivation suffer chronic inflammation and autoimmunity and die early; expression inactivated by promoter methylation in some hematopoietic malignancies
PTPN11	SHP-2, SH-PTP2, PTP1D	NT2	593 aa	2 SH2 domains, negative regulation by SH2-domain – catalytic-domain interaction, C-terminus with regulatory function (pTyr residues); largely cytoplasmic; knockout embryonically lethal; positive regulator of Ras signaling downstream of several RTKs and cytokines (e.g., PFGFRβ, IL-6) by dephosphorylation of inhibitory molecules; activating mutations in Noonan Syndrome and JMML
PTPN13	PTPL1, FAP1, PTP1E, PTP-BAS, PNP1; PTP-BL (mouse)	NT7 / 8	4 variants: 2490, 2485, 2466 and 2294 residues	Large molecule containing a KIND (putative kinase non-catalytic C-lobe domain), a FERM (Band 4.1, Ezrin, Radixin, Moesin) and five PDZ (PSD95, Dlg, ZO-1) domains, in addition to the PTP domain; localization cytoplasmic, cytoskeletal; can also shuttle to nucleus; multiple protein-protein interactions, e.g., with FAS; regulator of apoptosis; knockout mice exhibit impaired regenerative neurite outgrowth
DUSP3	VHR, Vaccinia H1-related phosphatase	Atypical dual-specificity phosphatases	185 aa	Small protein containing PTP/dual-specificity PTP domain, no reported regulatory domains; localization cytoplasmic, nuclear, possibly nucleolar; substrates include JNK and possibly ERK, and ERBB family receptors; required for cell-cycle progression; a detailed characterization of knockout mice yet unavailable
PTPRC	CD45, LCA	Receptor-like PTP family (R) 1/6	1304 aa	Transmembrane molecule with two intracellular PTP domains, membrane-proximal domain carries most of PTP activity; extracellular domain (ECD) contains cysteine-rich and three fibronectin type III-like domains, O- and N-glycosylated; alternative mRNA splicing gives rise to protein variants with different ECDs; high expression in nucleated hematopoietic cells; presumably negatively regulated by dimerization; essential for T- and B-cell receptor function, activates SRC-family kinases (e.g., LCK, FYN) by dephosphorylating their C-terminal inhibitory phosphosite; also reported as PTP for JAK1; knockout mice exhibit severe combined immunodeficiency; inactivating mutations in a subset of T-ALL patients
PTPRD	RPTPδ	R2A	1912 aa	Transmembrane molecule with two intracellular PTP domains, ECD (containing three Ig-like and eight fibronectin type III-like domains) may interact with proteoglycan-type ligands and PTP may undergo proteolytic processing (by analogy with other class R2A PTPs); neuronal expression, knockout mice exhibit retarded growth, impaired learning and memory, and early mortality; role in motorneuron axon targeting; mutated in solid cancers (e.g., glioma/glioblastoma)
PTPRE	PTPε	R4	700 aa	Transmembrane molecule with two intracellular PTP domains, and short, glycosylated ECD; alternative mRNA transcript and translation variants as well as partial proteolysis give rise to cytoplasmic protein variants with PTP activity; among characterized substrates are voltage-gated potassium channels, SRC family kinases, and JAK2; knockout mice exhibit hypomyelination, defective osteoclast and macrophage functions, and resistance to high-fat diet induced obesity.
PTPRN	ICA512, IA-2	R8	979 aa	Transmembrane molecule with one inactive PTP domain, presumably functioning as phosphotyrosine-binding domain; ECD with similarity to SEA (sea urchin sperm protein, enterokinase, agrin) domains of mucins; initially identified as autoantigen (“Islet Cell Antigen 512,” Islet Antigen-2) in type I Diabetes mellitus in pancreatic β-cells; localized to secretory granules, role in regulation of insulin secretion and presumably other neuroendocrine secretory processes
PTPRT	RPTPρ	R2B	1441 aa	Transmembrane molecule with two intracellular PTP domains, ECD (containing extracellular a meprin-A5 antigen-PTP (MAM) domain, an Ig-like domain and four fibronectin type III-like repeats) can undergo specific homophilic interactions mediating cell-cell adhesion; the PTP may be proteolytically processed (by analogy with other class R2B PTPs); among substrates are paxillin, and possibly proteins recruited to cell-cell adhesion complexes; frequently mutated in colon and other solid cancers; knockout mice are hypersensitive to chemical carcinogenesis of the colon

^a Nomenclature according to ref. 17; ^bData refer to human proteins; ^cInformation is based on publications cited in the text and on the following review articles and databases: refs. 6, 14, 17, 104–107; http://ptp.cshl.edu/; http://www.ptp-central.org/; http://www.genecards.org/; http://www.sanger.ac.uk/mouseportal/; http://www.uniprot.org/; information on PTP orthologs in other organisms can be found in recent review articles,^19-21 and the database http://www.ptp-central.org/

While all the mentioned mechanisms are involved in the control of JAK-STAT signaling, this review focuses on the contribution of PTPs to regulation of STAT activity.

Protein Tyrosine Phosphatases as Modulators of JAK-STAT Signaling

The structures and functions of PTPs have been extensively covered in a number of excellent recent reviews.^6,14-16 We will therefore summarize here only in brief some of their important features.

The initial classification of human (and related mouse) PTPs was based on structural similarity and defined a family encoded by approximately 100 genes.^17,18 They share a conserved fold and several conserved motifs, notably the motif [I/V]HCXXGxxR[S/T] in the catalytic center, with the cysteine acting as a nucleophil in the first step of the pTyr hydrolysis. These molecules can be further subclassified into classical PTPs (38 genes), which exclusively dephosphorylate protein-pTyr, dual-specificity enzymes (DUSPs, 61 genes), which have a more diverse range of substrates, including pTyr and phosphoserine/threonine residues in proteins, as well as phosphorylated lipids and carbohydrates. Recent studies have refined the initial classification and disclosed the identity of the tyrosine phosphatomes also in many other organisms, e.g., including Drosophila melanogaster, or Danio rero.^19-21 Interestingly, both among the classical PTPs as well as among the DUSPs there are molecules with no or very little enzymatic activity. Presumably, in these cases the PTP fold serves a function in protein–protein interactions, such as binding to pTyr motifs. Classical PTPs are either non-transmembrane molecules (NT PTPs), which have one conserved PTP domain of about 280 amino acids and regulatory domains, or transmembrane, receptor-like molecules (RPTPs), of which some have two PTP domains. Regulation of classical PTPs occurs at several levels, including mRNA expression, translational control, phosphorylation, and other posttranslational modifications.²² Of note in the context of this review are two regulatory mechanisms. One is reversible oxidation of the catalytic cysteine, which leads to (reversible) inactivation and appears to be exploited as signaling event.²³ The other is limited proteolysis. In the case of transmembrane PTPs, this process can lead to the release of the intracellular, enzymatically active domain which can act in the cytoplasm or even in the nucleus, as has been e.g., shown for PTPRK.²⁴ More distantly related PTP family members, which still make use of the same, cysteine-based enzymatic principle, comprise low-molecular weight phosphatase (ACP1, one gene), and the Cdc25 subfamily (three genes).¹⁸ Finally, several genes code for enzymes with distantly related structure and alternative enzymatic mechanisms which can also dephosphorylate pTyr in proteins. This heterogeneous group comprises four members of the Eyes absent family (Eya) whose catalysis is based on aspartate residues,¹⁸ and the UBASH3/STS/TULA family enzymes (two genes), which hydrolyze pTyr residues by a reaction employing a catalytic histidine.²⁵ Up to know, only cysteine-based PTPs have been identified to play a role in regulating JAK-STAT signaling (Tables 2 and 3).

Table 3. STATs and regulating PTPs.

STAT	PTPs involved in regulation	Type of evidencea and comments
STAT1	PTPN2 (TCPTP45)	Elevated phosphorylation and delayed dephosphorylation of STAT1 and enhanced activation of target genes in TCPTP deficient cells; immunochemical depletion experiments; “PTP substrate trapping”
	PTPN11 (SHP-2)	Elevated phosphorylation and delayed dephosphorylation of STAT1 in SHP-2 deficient cells
STAT2	Not reported	STAT1-interacting PTPs may affect STAT2 phosphorylation in heterodimers
STAT3	PTPN2 (TCPTP45)	Complex formation of endogenous TCPTP with STAT3 in a cell line; “PTP substrate trapping”; elevated phosphorylation and delayed dephosphorylation of STAT3 in TCPTP deficient cells
	PTPN11 (SHP-2)	Reduced phosphorylation of STAT3 in cells with constitutively active SHP-2; genetic evidence for SHP-2 targeting STAT3 in suppression of liver carcinogenesis
	PTPRD	Elevated STAT3 phosphorylation in PTPRD-depleted cells (siRNA) and diminished pSTAT3 by PTPRD overexpression; STAT3 substrate of PTPRD in vitro; proposed role of pSTAT3 dephosphorylation in tumor suppressor function of PTPRD in glioma/glioblastoma
	PTPRT	Elevated STAT3 phosphorylation in PTPRT-depleted cells (siRNA); “PTP substrate trapping”; reduced pSTAT3 signaling by PTPRT overexpression; proposed role of pSTAT3 dephosphorylation in tumor suppressor function of PTPRT in colon cancer
	PTPN6 (SHP-1)	Reduced STAT3 activity upon SHP-1 overexpression or re-expression in cancer cells; no evidence for direct interaction
STAT4	PTPN13 (PTP-BL)	Physical association, e.g., by yeast two-hybrid assay; STAT4 dephosphorylation by overexpression and in vitro; elevated phosphorylation of STAT4 in PTP-BL deficient cells
STAT5A/B	DUSP-3 (VHR)	Reduced STAT5 phosphorylation by overexpression; elevated pSTAT5 levels in VHR-deficient cells; binds in phosphorylated state to STAT5-SH2 domain
	PTPN11 (SHP-2)	STAT5 dephosphorylation in vitro; “PTP substrate trapping”; prolonged dephosphorylation of cytoplasmic (but not nuclear) STAT5 in SHP-2 deficient cells; in contradiction, positive function of SHP-2 for STAT5 signaling found in different biological contexts
	PTPN6 (SHP-1)	Overexpression and “PTP substrate trapping” experiments; may act on STAT5 in trimeric complexes with PLCβ3
	PTPRN (ICA-512, IA-2)	Fragment of PTPRN harboring the catalytically inactive PTP-domain protects STAT5 from dephosphorylation, thereby enhancing activity
STAT6	PTPN6 (SHP-1)	Delayed dephosphorylation of pSTAT6 and enhanced biological responses through STAT6 in SHP-1 deficient cells
	PTPN2 (TCPTP45)	Enhanced and prolonged phosphorylation of nuclear STAT6 in TCPTP45-deficient cells; “PTP substrate trapping”
	PTPN1 (PTP1B)	Enhanced and prolonged phosphorylation, and delayed dephosphorylation of cytoplasmic STAT6 in PTP1B-deficient cells; “PTP substrate trapping”
	PTPN13 (PTP-BL)	Dephosphorylation of pSTAT6 by PTP-BL in vitro

^a For references, see text.

Regulation of STAT Activation by PTPs—General Aspects

PTPs can regulate JAK-STAT signaling at several levels (Fig. 1). The activity of JAK/TYK kinases is under control of PTPs, which can attenuate signaling by dephosphorylating pTyr residues in the JAK/TYK activation loop. Furthermore, phosphorylation sites at the cytoplasmic domains of the cytokine receptors can be dephosphorylated, including STAT binding sites but also sites linked to the activation of other signaling molecules. Depending on the identity of these signaling proteins and on the affected regulatory networks, these dephosphorylations can have both, negative or positive effects on cytokine-induced signaling. Transmembrane PTPs as well as cytoplasmic PTPs have been described which can control these phosphorylations (Fig. 1; Table 2). Finally, dephosphorylation of phosphorylated STATs by PTPs occurs both in cytoplasm as well as in the nucleus (Fig. 1), as we will describe in detail later. PTP-mediated regulation of STAT signaling is integral to the JAK-STAT pathway, occurs in vertebrates and invertebrates²⁶ and has been found to be already an important signaling mechanism in Dictyostelium discoidum.^1,27 In a recent study, Araki et al.¹ could show that the activated Dictyostelium STATc is subject to dephosphorylation by the protein tyrosine phosphatase PTP3. This enzyme, which directly interacts with STATc, is itself inactivated by a stimulus-dependent mechanism based on serine-threonine phosphorylation. STATc activity in this interesting system, hence, is enhanced by downregulation of PTP activity. In Drosophila melanogaster, a genome-wide siRNA screen identified the PTP Ptp61F (the D.m. ortholog of human PTP1B (PTPN1) and TCPTP (PTPN2), Table 2) as important regulator of Hopscotch (Hop, D.m. JAK ortholog)–STAT92E (D.m. STAT ortholog) signaling.²⁸ Ptp61F can suppress melanocytic tumor formation driven by a constitutively active Hop variant,²⁸ and absence of Ptp61F was found associated with a fecundity defect, which is partially mediated by elevated STAT92E activity.²⁹ In mammals, expansion of receptor, JAK, STAT, and PTP families has greatly diversified the repertoire for JAK-STAT signaling and its regulation. In the following, a few examples shall illustrate how PTPs regulate the membrane-proximal events of JAK-STAT signaling, before we focus on direct PTP–pSTAT interactions.

Erythropoietin (EPO) controls red blood cell development via homodimerization and activation of its high affinity receptor (EPO-R) on the surface of erythroid progenitor cells. Intracellular signal transduction supports erythroid cell survival, proliferation, and differentiation. Among the EPO-R-borne tyrosine residues phosphorylated by activated JAK2, pY429 provides a binding site for the SH2 domain of SHP-1, and EPO-R mutants lacking Y429 lose SHP-1 binding. SHP-1, which is expressed predominantly in hematopoietic cells was identified as negative regulator of EPO-R signaling, since recruitment of SHP-1 to the activated EPO-R leads to inactivation of JAK2.³⁰ The interaction of SHP-1 with JAK2 and concomitant dephosphorylation of the kinase takes place by direct association³¹ and is critical for physiological down-modulation of proliferative signals emanating from the activated EPO-R. Interestingly, mutations within the EPO-R affecting Y429 are observed in primary familial polycythemia, a disorder characterized by congenital enhanced erythrocytosis.³²

As an additional PTP with regulatory functions in hematopoiesis, the transmembrane phosphatase CD45 (PTPRC, Table 2) has been identified. Comparable to SHP-1, CD45 directly contacts and thereby dephosphorylates JAKs. CD45 exerts dampening effects on cellular functions triggered by several cytokine receptors, among them EPO-R-mediated hematopoiesis.³³

An interesting novel mode of PTP-mediated JAK-STAT regulation was recently reported for signaling by IL-6 type cytokines. This mechanism relies on the transmembrane glycoprotein signal regulatory protein/SHP2-substrate (SIRP1α/SHPS-1) whose function was originally connected to cell adhesion-induced signaling. It was shown that the cytoplasmic domain of SIRP1α contacts SHP2 in response to IL-6 receptor stimulation, and from experiments with SIRP1α/SHPS-1-deficient cells it was deduced that this interaction results in decreased SHP-2 phosphorylation and ERK1/2 activation. On the other hand, tyrosine phosphorylation of STAT3 and its transcriptional activity appear to be enhanced under these conditions. It has been speculated that this PTP-dependent mechanism is involved in balancing MAP kinase and STAT3 pathways which simultaneously emanate from activated receptors of the gp130 family.³⁴

Leptin is an adipocyte-derived cytokine which acts on receptors in the hypothalamus and thereby affects energy homeostasis. A number of studies in mice with inactivated PTP encoding genes have shed light on the regulation of leptin signaling by PTPs, recently reviewed in references 35 and 36. The early observation that PTP1B knockout mice (Table 2) exhibit reduced weight gain under high-fat diet,³⁷ was later found to be partially related to the capacity of PTP1B for effective dephosphorylation of JAK2,³⁸ which functions downstream of activated leptin receptors. PTP1B deficiency therefore augments leptin signaling, leading to decreased food intake and enhanced energy expenditure. TCPTP, a related PTP (Table 2) has also been identified as a negative regulator of leptin signaling.³⁹ TCPTP is an efficient PTP for JAK1 and 3,⁴⁰ but cannot efficiently dephosphorylate JAK2. In the leptin pathway, it acts by regulating STAT3, the downstream substrate of JAK2.³⁹ Another PTP which negatively regulates leptin signaling by JAK2 dephosphorylation is the transmembrane molecule PTPε (PTPRE, Table 2).⁴¹ Absence of PTPε improves glucose metabolism under conditions of high-fat diet and attenuates weight gain specifically in female mice. Interestingly, JAK2 can phosphorylate and thereby activate PTPε, indicating the possibility of feedback regulation.⁴¹ Obviously, different PTPs can regulate this pathway in a partially overlapping, partially complementary manner, a scenario which may be of general relevance for PTP-mediated regulation of JAK-STAT signaling and applies also to the direct regulation of pSTATs, discussed below (for an overview see Table 3).

Direct Regulation of Individual STAT Family Members by PTPs

STAT1 and 2

For STAT1, the biochemistry of the activation-inactivation cycle has so far been best investigated, and some of the key features in this process are presumably of general importance and may extend to further members of the STAT family. Importantly, early experiments with STAT1 have shown dephosphorylation as the principal mechanism of inactivation. In these experiments, a generic assay was applied, which was later used in many other studies: After stimulation of STAT1 activation, the “forward” reaction of STAT1 phosphorylation was quenched with staurosporine, a general, potent, and rapidly acting protein-kinase inhibitor, and the decay of pSTAT over time was subsequently monitored.⁴² The importance of dephosphorylation of pSTATs by PTPs is also supported by the observation that STAT activation can be achieved in the absence of ligands for upstream receptors, merely by induction of reactive oxygen species (ROS) and most likely through the inactivation of critical PTP(s) by reversible oxidation.⁴³

How can PTPs access the pTyr residues buried in a pSTAT–pSTAT dimer by intimate contact with SH2-domains (Figs. 1 and 2)? This issue has also been addressed first for STAT1. Based on structural studies of STAT1 and mutational analyses, a model was proposed in which the pSTAT1–pSTAT1 dimer undergoes reorientation from the parallel structure stabilized by SH2–pTyr interactions into an antiparallel structure (Fig. 2). This conformational change is triggered and stabilized by interactions between N-terminal STAT domains. In the antiparallel dimer, pTyr residues are exposed and accessible for PTP-mediated dephosphorylation (Fig. 2). Stabilization of the SH2–pTyr-mediated dimer structure or prevention of the antiparallel dimer formation by different types of mutations causes resistance to PTP-mediated dephosphorylation and “gain-of-function” of the corresponding mutant proteins.^44-47 Interestingly, similar gain-of-function mutations of STAT1 were recently found in patients with chronic mucocutaneous candidiasis disease (CMCD).⁴⁸ The discovery of these mutants highlights the physiological relevance of efficient nuclear pSTAT1 dephosphorylation. The mutations are localized in the N-terminal coiled-coil domain and are likely to impair formation of the antiparallel STAT1 conformation (which would enable PTP interactions). They thereby cause persistent phosphorylation of nuclear STAT1 and drive the disease in an autosomal-dominant manner.

Figure 2. Regulation of STAT signaling by T-cell protein tyrosine phosphatase (TCPTP). pSTAT dimers can be transported into the nucleus, bind to DNA and regulate gene transcription. In the pSTAT dimers formed by reciprocal interaction of the SH2-domains with the phosphotyrosines (so-called “parallel” dimers), the phosphotyrosines are buried. STAT molecules can also form “antiparallel” dimers through their N-terminal domains, which exist in equilibrium with the parallel dimers. In the antiparallel state, protein tyrosine phosphatases (PTPs) can more easily access the pSTAT phosphotyrosine residues. A major STAT PTP is TCPTP45, which can shuttle between cytoplasm and nucleus and dephosphorylate both, nuclear and cytoplasmic pSTAT1, 3, and 6, and possibly further STATs.

TCPTP (PTPN2), which can localize to nucleus and cytoplasm (TCPTP45, Table 2), has been identified as a major PTP for nuclear pSTAT1.⁴⁹ STAT1 was found hyperphosphorylated and -activated in TCPTP-deficient cells. The dephosphorylation of pSTAT1 in TCPTP-deficient mouse fibroblasts was strongly delayed, hand in hand with an elevated transcription of STAT1 target genes. Furthermore, nuclear extracts of TCPTP deficient mouse fibroblasts exhibited no pSTAT1 phosphatase activity in vitro. Finally, pSTAT1–TCPTP complexes could be recovered in immunoprecipitates using a so-called “substrate trapping” mutant of TCPTP,^50,51 which is inactive but still binds to cognate substrates.

Interestingly, several modifications of STAT1 were found to enhance TCPTP mediated dephosphorylation. These comprise STAT1 methylation at Arg-31,⁵² STAT1 acetylation at Lys 410 and Lys413,⁵³ and the association of STAT1 with nuclear β-arrestin 1.⁵⁴ In all these studies, an enhanced physical association of TCPTP with STAT1 was observed. It remains to be clarified if STAT1 methylation, acetylation, or complex formation with β-arrestin 1 create novel TCPTP interaction motifs, or rather facilitate TCPTP binding by causing conformational changes of STAT1, e.g., promoting the antiparallel dimer formation (Fig. 1) and thereby leading to exposure of the pTyr sites to TCPTP for dephosphorylation.

The currently only other PTP implicated in STAT1 dephosphorylation is SHP-2 (PTPN11, Table 2). STAT1 phosphorylation was reportedly elevated and prolonged in SHP-2 deficient fibroblasts, STAT1 dephosphorylation was delayed, and recombinant GST-SHP-2 was shown to dephosphorylate pSTAT1 in vitro.⁵⁵

To the best of our knowledge, no study has yet specifically addressed the identity of PTPs dephosphorylating STAT2. Since STAT2 can form heterodimes with STAT1, one could speculate that—mediated by interaction with STAT1—TCPTP potentially also affects the phosphorylation state of STAT2 in such dimers.

STAT3

TCPTP has also been established as a negative regulator of pSTAT3. pSTAT3 was efficiently dephosphorylated by TCPTP overexpression, and by recombinant TCPTP in vitro, and a TCPTP trapping mutant bound to pSTAT3.⁵⁶ Moreover, pSTAT3 (but not pSTAT5 or pSTAT6) dephosphorylation was remarkably delayed in TCPTP-deficient mouse fibroblasts.⁴⁹ Direct dephosphorylation of pSTAT3 by TCPTP has recently also been proposed as a mechanism of TCPTP tumor suppressor activity in mammary carcinoma.⁵⁷

SHP-2, at least its activated versions found in Noonan syndrome (NS) and juvenile myelomonocytic leukemia (JMML),⁵⁸ may also be able to dephosphorylate pSTAT3. pSTAT3 levels were lower in NS patients, were reduced in cells expressing activated SHP2 mutants and enhanced in cells treated with a SHP-2 inhibitor, and recombinant activated SHP-2 could dephosphoryate pSTAT3 in vitro.⁵⁹ The capacity of SHP-2 to dephosphorylate STAT3 may play a role in its function as a tumor suppressor in hepatocellular carcinogenesis.⁶⁰

Interestingly, the tumor suppressor activity of two further PTPs has recently also been linked to pSTAT3 inactivation: PTPRD (Table 2) was found frequently inactivated by multiple mechanisms in glioblastoma and in other human cancers such as lung carcinoma. Overexpression of wild-type but not of mutant PTPRD (as found in the tumors) diminished STAT3 phosphorylation and activation of transcription, siRNA-mediated knockdown caused elevation of pSTAT3 in immortalized human astrocytes. Recombinant PTPRD efficiently dephosphorylated pSTAT3.⁶¹ How PTPRD can access pSTAT3 was not addressed in this study. PTPRT is another transmembrane PTP (Table 2), which is frequently mutated in human cancers, and at least some of the mutations cause enzymatic inactivation. pSTAT3 was identified as an important substrate of PTPRT starting out from a phosphoproteomic approach. PTPRT attenuated IL-6 mediated STAT3 activation in cell lines, PTPRT siRNA elevated pSTAT3 levels, and pSTAT3 bound to a PTPRT substrate trapping mutant.⁶² The authors propose that pSTAT3 dephosphorylation either occurs by full-length PTPRT in the cytoplasm, or that a proteolytic product of PTPRT may execute dephosphorylation in the nucleus. PTPRK and PTPRM, close relatives of PTPRT, have indeed been shown to undergo proteolytic processing, leading to formation of a catalytically active fragment which can travel to the nucleus and modulate gene transcription.²⁴ A similar situation may exist for the transmembrane PTP PTPRO (also designated GLEPP1). PTPRO deficient mice exhibit enhanced tumor formation in a model of chemical carcinogen-induced hepatocellular carcinoma. As underlying signaling mechanism, the authors identified enhanced JAK2-STAT3 signaling in PTPRO-deficient cells.⁶³

Finally, there are many reports showing that SHP-1 can negatively regulate STAT3 signaling, and elevation of SHP-1 expression or activity by different compounds can inhibit tumor cell growth through this mechanism (e.g., see ref. 64). However, to our knowledge there are no reports to date showing that SHP-1 can directly interact with pSTAT3, or that SHP-1 deficiency would cause diminished dephosphorylation of pSTAT3. Thus, the effects of SHP-1 activation may rather be mediated by dephosphorylation of upstream components of the STAT3 activating pathway.

Constitutive activation of STAT3 by introduction of a free cysteine residue into the C-terminal domain, which causes spontaneous formation of dimers and activation of STAT3 in the absence of tyrosine phosphorylation has been described many years ago.⁶⁵ Recently, somatic gain-of function mutations of STAT3 were described in human inflammatory hepatocellular adenomas,⁶⁶ large granular lymphocytic leukemia,⁶⁷ and hyper-IgE syndrome.⁶⁸ These STAT3 mutants also dimerize spontaneously and are highly phosphorylated in the absence of a cytokine stimulus. Dephosphorylation kinetics were not yet assessed for the STAT3 variants, it is tempting, however, to speculate that these mutant proteins are constitutively active because the stabilized parallel dimers are protected from dephosphorylation.

STAT4

A yeast-two-hybrid search for interactors of STAT4 uncovered an up to date unique role of the large non-receptor PTP-BL (PTPN13, Table 2) for STAT4 dephosphorylation. PTP-BL bound to STAT4 and dephosphorylated pSTAT4 when overexpressed in a Th1-like cell line, but did not affect phosphorylation of the upstream activator Tyk2. Importantly, STAT4 phosphorylation was increased, and dephosphorylation was clearly impaired in PTP-BL-deficient CD4⁺ T cells, correlating with enhanced biological responses to cytokine stimulation. PTP-BL mediated dephosphorylation of STAT6 in vitro was also demonstrated, suggesting that this PTP can target also pSTAT6 and possibly further pSTATs.⁶⁹

STAT5A and B

Resistance to dephosphorylation was convincingly shown to form the basis for constitutive activation of several STAT5 variants, which were generated by random mutagenesis.^70,71 Interestingly, an equivalent of one of these previously described mutations (STAT5B N642H) has recently been found in large granular lymphocytic leukemia.⁷² The latter variant as well as the earlier described constitutively active STAT5 mutants can potently transform hematopoietic cells.^70,71,73 Given these findings and the extensive investigation of biological functions of STAT5A and B in many different contexts, the knowledge about the mechanisms of their dephosphorylation is surprisingly limited.

Again, TCPTP may play a role, a conclusion largely derived from overexpression experiments in prolactin stimulated cells.⁷⁴ However, studies supporting a role of TCPTP for STAT5 dephosphorylation using TCPTP loss-of function models such as TCPTP-deficient mice are lacking up to now. In fact, TCPTP deficiency had no impact on STAT5 phosphorylation in two different cell systems.^49,57

SHP-2 has also been proposed as a pSTAT5 PTP. STAT5 dephosphorylation in IL2-stimulated and subsequently IL2-deprived cells occurred mainly in the cytosol. A trapping mutant of SHP-2, but not of SHP-1 formed complexes with STAT5 and SHP-2 dephosphorylated pSTAT5 in vitro.⁷⁵ Moreover, STAT5A exhibited prolonged phosphorylation in cytoplasmic, but not in nuclear extracts of SHP-2-deficient fibroblasts.⁷⁶ In contrast to these biochemical findings, SHP-2 seems to rather play a positive role in different biological processes which depend on STAT5 signaling. For example, STAT5 activity was reduced and lactogenic differentiation was inhibited in mammary glands of SHP-2-deficient mice.⁷⁷ Another example is the positive role of SHP-2 for STAT5-dependent differentiation of human CD34⁺ cells.⁷⁸ Effects of SHP-2 deficiency in other pathways, notably for Ras activation, may have obscured its possible role in limiting STAT5 phosphorylation in these studies. Clearly, the biological relevance of STAT5 dephosphorylation by SHP-2 requires further investigation.

SHP-1 may be another STAT5 PTP. In the context of IL-3 stimulated STAT5 activation in Ba/F3 cells, wild-type SHP-1 attenuated, while a dominant negative SHP-1 variant (R459M), enhanced STAT5 activation. Notably, dephosphorylation of pSTAT5 was attenuated by expression of the C453S SHP-1 trapping mutant.⁷⁹ More recently, SHP-1-mediated dephosphorylation of STAT5 was reported to play a role in the tumor suppressor activity of PLCβ3. PLCβ3, STAT5, and SHP-1 were shown to form multimolecular complexes in hematopoietic cells. Inefficient control of STAT5 activity in PLCβ3-deficient mice was proposed as a mechanism for the observed development of myeloproliferative disease in these animals.⁸⁰

A very interesting scenario of regulating STAT5 dephosphorylation was observed in pancreatic β cells involving ICA512 (PTPRN, also known as IA-2, Table 2), a catalytically inactive member of the PTP family with a transmembrane architecture. A cleaved cytosolic fragment of ICA512 harboring the PTP domain promoted the STAT5-dependent transcription of secretory granule genes by protecting pSTAT5 from dephosphorylation in intact cells and (by recombinant TCPTP) in vitro. However, the identity of the endogenous active PTP(s) whose action on pSTAT5 is impaired by this mechanism, was not addressed in this study.⁸¹

Finally, the dual-specificity PTP VHR (DUSP-3, Table 2) has been implicated in pSTAT5 dephosphorylation. Starting out from screening of different PTPs with known or potential nuclear localization for STAT5 dephosphorylating capacity in HEK293 cells, VHR was found to be most potent. Importantly, pSTAT5 dephosphorylation was strongly attenuated in IFNβ- stimulated splenic B cells of VHR-deficient mice.⁸²

STAT6

SHP-1 is most likely an important PTP for the limitation of STAT6 signaling. Earlier studies in IL-4 stimulated hematopoietic precursor cells from motheaten mice (mice lacking functional SHP-1, see Table 2) had already indicated a regulatory role for this enzyme.⁸³ Very recently, convincing observations were made in T cells with a cell-specific inactivation of the SHP-1 encoding gene: While the extent of short-term induction of STAT6 phosphorylation by IL-4 was not affected by absence or presence of SHP-1, the dephosphorylation of pSTAT6 after removal of IL-4 was strongly delayed in SHP-1-deficient cells. This biochemical phenotype was associated with an increased frequency of memory T cells in the absence of SHP-1, whose generation depends on IL-4 signaling.⁸⁴

Further, TCPTP has been established as a nuclear PTP also for STAT6. pSTAT6 dephosphorylation was inversely correlated with TCPTP expression in activated B-cell like diffuse large B-cell lymphomas (DLBCLs). Immunodepletion of TCPTP removed pSTAT6 PTP activity from nuclear extracts, STAT6 activation by IL-4 was enhanced and prolonged in cells with knockdown or genetic deficiency of TCPTP expression. pSTAT6 dephosphorylation was delayed (albeit not abrogated) in TCPTP-deficient mouse fibroblasts, and pSTAT6 efficiently interacted with a TCPTP trapping mutant.⁸⁵ Using similar techniques, PTP1B was later identified as a PTP for cytosolic pSTAT6. IL-4-stimulated STAT6 phosphorylation was elevated in fibroblasts from PTP1B knockout mice and in cells with PTP1B depletion by siRNA. This effect concerned mainly the cytoplasmic but not the nuclear pool of pSTAT6, and a PTP1B trapping mutant associated with pSTAT6 in the cytoplasm. SiRNA depletion of PTP1B delayed dephosphorylation of pSTAT6 under conditions of upstream kinase inhibition. Interestingly, IL-4 could induce rapid TCPTP and PTP1B gene expression in DLBCL cells, suggesting the possibility of a negative feedback loop, and the concerted activity of PTP1B and TCPTP in regulation of cytoplasmic and nuclear pSTAT6, respectively.⁸⁶

Finally, another pSTAT6 PTP may be PTP-BL, which was convincingly shown to mediate pSTAT4 dephosphorylation (see above and ref. 69) and could effectively dephosphorylate also pSTAT6 in vitro.

Conclusions and Outlook

It has become obvious that the function of PTPs is of crucial importance to fine-tune the activities of STAT factors. However, our knowledge on specific structural and functional partnerships between individual STATs and PTPs and of their general significance is still fragmentary, since only some examples of PTP-dependent modulation of STAT activities have been dissected in full detail. It is likely that cell type- as well as STAT-specific functions of PTPs may constitute an additional layer for signal modulation. An example pointing to that direction is a recent report by Chen et al.,⁸⁷ who showed that the natural compound wedenolactone selectively activates IFNγ-signaling by inhibition of TCPTP, suggesting a potential dominance of this PTP in the STAT1 pathway. To obtain a full picture, it will be necessary to perform systematic screens for PTPs regulating the individual STAT family members, also in the context of different tissues and cell types.

Such studies will have to include biochemical approaches addressing novel protein-protein interaction analysis as well as approaches from the field of functional genomics. The biochemical branch of this future research may employ the analysis of binding partners of the different STATs in various cell types and tissues by the tools of state-of-the-art proteomics and mammalian two hybrid methods that can sensitively detect productive protein-protein contacts within signaling pathways, e.g., the MAPPIT technique.^88,89 On the genomic side, siRNA screens have the potential to yield novel information on the influence of specific PTPs on STATs in given cell types.⁹⁰ Functional readouts for the activity of individual STATs based on reporter gene modulation are available that are suitable for high throughput screening (e.g., ref. 91).

Another potential level of regulation associated with the balance of STAT tyrosine phosphorylation and dephosphorylation lies in the fact that the phosphorylated STATs can be retained in the cytoplasm through interaction with binding partners such as PI-3 kinase⁹² or SRC family kinases.⁹³ In these two cases, the mechanism affects STAT5 and is believed to exert influence on as yet undefined non-transcriptional cytoplasmic functions with relevance for cancer. Similarly, STAT3 has been shown to participate in various processes outside the cell nucleus in a phosphorylation-dependent fashion. For instance, it interacts with stathmin, thereby controlling microtubule dynamics in migrating T cells,⁹⁴ contributes locally to neuronal axon function,⁹⁵ and is directly involved in collagen-induced platelet aggregation.⁹⁶ Notably, STAT3 is also present in mitochondria, where it is involved in the organization of cancer-related alterations in cell metabolism.⁹⁷ Obviously, these activities have direct potential influence on cell function in health and disease and it will be important to determine a potential regulatory role of PTPs on these processes. Moreover, it would be most interesting to determine if similar mechanisms of phosphorylation-dependent retention in the cytoplasm are also operative for STAT1. For this factor, the unphosphorylated form has been shown to transmigrate to the cell nucleus in its own right and to play a distinctive role in role in transcriptional regulation. Unphosphorylated STATs can provide specific contributions to the pattern of gene regulation.⁹⁸ Consequently, the degree of cytoplasmic dephosphorylation and, hence, the ratio of unphosphorylated vs. phosphorylated STAT1 can be expected to have a direct influence on the composition of the transcriptome.

Structural information on STAT–PTP interaction is sparse. As outlined above, it has been concluded from mutational studies that parallel pSTAT1 dimers have to undergo a conformational rearrangement to an antiparallel dimer to allow accessibility of the pTyrs for PTPs.^44,47 Structural studies on corresponding STAT–PTP complexes would be highly desirable to better understand the mechanism and regulation of these interactions.

Lastly, since the interdependence of tyrosine and serine phosphorylation with regard to STAT activity is of significance and, at the same time, poorly understood, a better insight into mechanisms governing STAT serine phosphorylation and dephosphorylation is highly warranted.

10.4161/jkst.28087