PTPN11 (Shp2) Mutations in LEOPARD Syndrome Have Dominant Negative, Not Activating, Effects*

Multiple lentigines/LEOPARD syndrome (LS) is a rare, autosomal dominant disorder characterized by Lentigines, Electrocardiogram abnormalities, Ocular hypertelorism, Pulmonic valvular stenosis, Abnormalities of genitalia, Retardation of growth, and Deafness. Like the more common Noonan syndrome (NS), LS is caused by germ line missense mutations in PTPN11, encoding the protein-tyrosine phosphatase Shp2. Enzymologic, structural, cell biological, and mouse genetic studies indicate that NS is caused by gain-of-function PTPN11 mutations. Because NS and LS share several features, LS has been viewed as an NS variant. We examined a panel of LS mutants, including the two most common alleles. Surprisingly, we found that in marked contrast to NS, LS mutants are catalytically defective and act as dominant negative mutations that interfere with growth factor/Erk-mitogen-activated protein kinasemediated signaling. Molecular modeling and biochemical studies suggest that LS mutations contort the Shp2 catalytic domain and result in open, inactive forms of Shp2. Our results establish that the pathogenesis of LS and NS is distinct and suggest that these disorders should be distinguished by mutational analysis rather than clinical presentation.

Shp2 is a ubiquitously expressed, non-receptor protein-tyrosine phosphatase (PTP) comprising two N-terminal SH2 domains, a catalytic (PTP) domain, and a C terminus with tyrosyl phosphorylation sites and a proline-rich stretch (14). Via its SH2 domains, Shp2 binds directly to some growth factor receptors, such as the platelet-derived growth factor (PDGF) receptor, as well as to several scaffolding adapters, including IRS, FGF receptor substrate (FRS), and Gab proteins. Formation of such complexes is required for full activation of the Ras/Erk cascade in most, if not all receptor-tyrosine kinase, cytokine receptor, and integrin signaling pathways. Consequently, Shp2 plays an important role in mediating multiple downstream biological responses, such as proliferation and/or survival, adhesion, and migration.
All known biological functions of Shp2 require its catalytic activity (14). Shp2 activity is tightly regulated by an elegant "molecular switch" mechanism that couples activation with recruitment of Shp2 to its binding proteins (14 -16). In the basal state (i.e. in unstimulated cells when Shp2 is found primarily in the cytoplasm), the "backside loop" of the N-SH2 domain (the side opposite to its phosphotyrosyl peptide binding pocket) interacts with the PTP domain, preventing substrate access to the active site (15). Binding to a phosphotyrosyl (Tyr(P)) protein ligand (e.g. in a receptor-tyrosine kinase or scaffolding adapter) alters the conformation of the N-SH2 domain, rendering it unable to bind the PTP domain and activating the enzyme (Fig. 1A). Earlier studies showed that mutating key contacts between the N-SH2 and the PTP domains led to biochemically and biologically "activated mutants" of Shp2 (17).
A large number of NS mutants have been identified (1,4). Most of these disrupt key contacts between the N-SH2 and PTP domains, resulting in increased basal and stimulated PTP activity (18 -20). NS mutants (when co-transfected with the scaffolding adapter Gab1), enhance Erk mitogen-activated protein kinase activation in transient transfection assays (18). Erk is also hyperactivated in selected embryonic tissues from a mouse model of NS (21). Neoplasia-associated PTPN11 mutations affect many of the same residues but typically are less conservative, resulting in greater catalytic activation (5,7,19,20). When expressed in hematopoietic cells, such mutants enhance basal and cytokine-evoked Erk, Akt, and Stat 5 activation (22,23).
Because NS and LS share multiple phenotypic features and are caused by PTPN11 mutations, they have been viewed as overlap syndromes (1). We examined the enzymatic properties of LS mutants and their effects on receptor-tyrosine kinase signaling. Surprisingly, despite its phenotypic and genetic similarities to NS, we found that LS is caused by catalytically defective, loss-of-function mutations in PTPN11.

EXPERIMENTAL PROCEDURES
Structural Interpretation and Molecular Modeling-The Shp2 structure 2SHP (Protein Data Bank), comprising the two SH2 domains plus the PTP domain and corresponding to the basal, inactive form of the enzyme in the absence of Tyr(P) peptide (15), and the predicted effects of various mutants were visualized using the PyMol Molecular Graphic System, Version 0.97 (DeLano Scientific).
Expression Constructs and Protein Purification-Point mutations were introduced using the QuikChange kit (Stratagene). Wild type Shp2 (WT) and mutant Shp2 in pGEX-4T-2F, a modified version of pGEX-4T (Amersham Biosciences) that generates proteins with N-terminal GST-and C-terminal FLAG tags, were produced and purified on glutathione-Sepharose (Amersham Biosciences), as described previously (20). Protein concentrations were determined by densitometric analysis of Coomassie-stained SDS-PAGE gels using bovine serum albumin as the standard. Purified enzymes were stored in the presence of 33% glycerol at Ϫ80°C.
PTP Assays-PTP assays using 32 P-labeled carboxamido-methylated and -maleylated lysozyme (RCML) (typical specific activity ϳ4000 cpm/ pmol) as substrate were conducted in assay buffer (50 mM Hepes (pH 7.4), 150 mM NaCl, 0.1 mg/ml bovine serum albumin, 10 mM dithiothreitol, 5 mM EDTA, 2 M RCML, and 5 nM GST-enzyme) with or without IRS-1 Tyr(P)-1172 (SLNpYIDLDLVK; pY is Tyr(P)) or Tyr(P)-1222 (LSTpYASINFQK) peptides as described (20). Assays using para-nitrophenyl phosphate (pNPP, obtained from Sigma) as substrate were carried out in 30 mM Hepes (pH 7.4), 120 mM NaCl, 5 mM dithiothreitol, 10 mM pNPP, and 5 nM enzyme with or without various concentrations of Tyr(P)-peptide at 30°C for 10 min and terminated with 0.2 N NaOH. Phosphate release was determined by measuring A 410 . Immune complex PTP assays were performed on lysates of transiently transfected 293 cells. Forty-eight hours after transfection, lysates from serum-starved or epidermal growth factor (EGF)-stimulated (50 ng/ml) cells were prepared, and Shp2 proteins were immunoprecipitated by using anti-Shp2 polyclonal antibodies (Santa Cruz) coupled to protein A-Sepharose. Shp2 immune complexes were washed 3 times in 1% Nonidet P-40 lysis buffer (see below) without sodium orthovanadate and once in wash buffer (30 mM HEPES (pH 7.4), 120 mM NaCl without pNPP). PTP assays were performed at 37°C in 50 l of pNPP assay buffer as described above. Recovered immune complexes were boiled in 2ϫ SDS-PAGE sample buffer, resolved by SDS-PAGE, and immunoblotted for Shp2 to control for equal Shp2 expression. All assays were carried out in triplicate.
Stable Cell Lines-Retroviruses expressing WT or mutant Shp2 were generated by transient co-transfection of 293T cells with pBABE-puro constructs and AmphoPac (Invitrogen). Viral supernatants were collected 48 h post-transfection, passed through a 0.45-m filter, and used to infect fresh 293T cells in the presence of 4 g/ml Polybrene (Sigma). Forty-eight hours later cells were selected in 1 g/ml puromycin (Sigma). Cell stimulations and analyses were performed as described above.

LS Mutations
Are Catalytically Inactive-In contrast to NS mutations, which are scattered throughout the Shp2 molecule, LS mutants are confined to the PTP domain (11,12,23) (Figs. 1, B-D). Based on structural and enzymologic studies of other PTPs (e.g. PTP1B (24,25)), the six mutations specifically linked to LS (Y279(C/S), T468M, A461T, G464A, Q506P, Q510P) involve residues predicted to affect catalysis (26). Three (Ala-461, Gly-464, and Thr-468) reside within the "signature motif " ((I/V)HCXAGXGR(S/T)GT) that defines the PTP family (26) and comprises the "PTP loop" in PTP structures (24,25). The conserved cysteinyl residue within this motif carries out a nucleophilic attack on the substrate phosphotyrosyl residue, generating a thiophosphate intermediate that subsequently undergoes hydrolysis by a bound water molecule. Other signature motif residues are important for lowering the pK a (i.e. enhancing the nucleophilicity) of the catalytic cysteine, orienting the substrate for nucleophilic attack, maintaining the structural integrity of the catalytic pocket, and/or neutralizing the charge of the phosphotyrosine in the substrate (26). Shp2 Tyr-279 is cognate to Tyr-46 in PTP1B, which sets the depth of the catalytic cleft and, thus, confers specificity for phosphotyrosine-containing substrates (24). Gln-506 or possibly Gln-510 is the Shp2 analog of Gln-262 in PTP1B, which helps position the water molecule to hydrolyze the thiophosphate intermediate (25,27).
Based on their locations in the PTP catalytic cleft, we hypothesized that the LS mutations might have fundamentally different biochemical and biological properties than other disease-associated PTPN11 mutants. To begin to test this hypothesis, we expressed and purified GST-tagged wild type Shp2 (WT), the NS mutation D61G, the leukemia-associated mutant E76K, and several LS mutants as recombinant proteins in bacteria, as described previously (20). Purified proteins were assayed against the artificial substrates pNPP and reduced carboxamido-methylated and -maleylated lysozyme (RCML). As expected (17)(18)(19)(20), WT Shp2 exhibited low basal activity that was stimulated after the addition of a Tyr(P) peptide from IRS-1 (Tyr(P)-1172) that binds the N-SH2 domain ( Fig. 2A). D61G had increased basal activity (compared with WT) that was further enhanced upon the addition of Tyr(P)-1172, whereas E76K was maximally activated even in the absence of Tyr(P) peptide. In contrast, none of the LS mutants exhibited detectable PTP activity against either substrate, even in the presence of saturating amounts of Tyr(P)-1172 ( Fig. 2A) or over a wide range of doses of another activating phosphopeptide, Tyr(P)-1222, of IRS-1 (Fig. 2B).
To exclude the possibility that the recombinant LS proteins produced in bacteria folded inappropriately and/or were unstable, mammalian expression constructs for (untagged) WT Shp2 and the indicated mutants were transiently transfected into 293T cells, which were then starved and either left unstimulated or stimulated with EGF. Endogenous (i.e. empty vector-transfected) Shp2 activity, as measured by immune complex PTP assay, was enhanced slightly upon growth factor stimulation (Fig. 2C). Cells transfected with WT Shp2 showed slightly higher basal and stimulated Shp2 activity, whereas those expressing the leukemia-associated E76K mutant showed markedly enhanced and constitutive Shp2 activity. However, transient expression of LS mutants led to no increase in PTP activity even though these mutants were expressed at levels comparable with WT Shp2 or E76K (Fig. 2C, bottom  panel). Instead, LS mutants tended to cause decreased EGF-stimulated Shp2 activity (as measured by immune complex assay), raising the possibility that such mutants have dominant negative effects.
LS Mutations Disrupt the Catalytic Pocket of Shp2-Examination of the Shp2 crystal structure (Fig. 1D) provided potential explanations for these observations. Because Tyr-279 sets the depth of the catalytic cleft, substitution by cysteine (or serine as in another LS patient (12)) should alter the depth of the cleft and might also perturb the orientation of the catalytic cysteine (Figs. 3, A and B). Ala-461 is a highly conserved (see ptp.cshl.edu) PTP loop residue that also contacts Tyr-279; placing a bulkier threonyl substitution here should contort the catalytic site and interfere with substrate phosphotyrosine binding (Figs. 3, C and D). Gly-464 is an invariant signature motif residue in PTP superfamily members. Its replacement by alanine also should disrupt the PTP loop and sterically clash with the phosphate group of an incoming substrate phosphotyrosyl residue (Figs. 3, E and F). Thr-468 is a buried residue on the ␣G (␣-4 in PTP1B) helix. Methionine substitution here probably would destabilize the entire PTP catalytic domain, with associated loss of enzymatic activity (Figs. 3, G and H).  MARCH 10, 2006 • VOLUME 281 • NUMBER 10 LS Mutants Are Dominant Negative-We next examined the effects of LS mutants on receptor-tyrosine kinase signaling. Expression constructs for WT or mutant Shp2, Gab1, and HA epitope-tagged Erk1 were co-transfected into 293T cells, as described previously (18). Tran-siently transfected cells were starved or stimulated with EGF, and total Tyr(P) (data not shown) and Erk activation was monitored by antiphospho-Erk immunoblotting (Fig. 4A). As expected, EGF stimulation activated Erk in WT-expressing cells, and Erk activation was enhanced to a much greater extent in cells expressing E76K or D61G. Expression of a mutant of the essential catalytic cysteinyl residue (C459S) inhibited Erk activation, as reported (28). Notably, the two most common LS mutants, Y279C and T468M, also had dominant negative effects, strongly inhibiting EGF-evoked Erk activation (Fig. 4A). Likewise, in stable cell pools expressing each mutant, D61G enhanced, whereas LS mutants strongly impaired, EGF-evoked Erk activation (Fig. 4B). Similar to the effects of Shp2 deficiency in fibroblasts (29 -31), the major effect of LS mutants was to inhibit sustained EGF-evoked Erk activation (Fig.  4C). LS mutants also impaired Erk activation in response to FGF or PDGF stimulation (Fig. 4, D and E), consistent with a general ability of these mutants to impair receptor-tyrosine kinase signaling.

LS Mutants Preferentially
Associate with Gab1-At first glance, expression of one normal and one catalytically inactive version of Shp2 might be expected to have the same effect as loss of a normal Shp2 allele, yet hemizygous Shp2 mutant mice have no apparent cardiac, facial, or size abnormalities (29). However, complexes between Shp2 and a relevant binding protein are the active signaling entities, because Shp2 must  293T cells were transiently transfected with HA-Erk1, WT or mutant Shp2, and Gab-1 expression constructs as indicated and then starved and either stimulated with EGF (50 ng/ml) or left unstimulated. Erk activation was assessed by anti-phospho-Erk immunoblotting of total cell lysates. The lower panels show reprobes for total Erk-1 and Shp2 levels. In contrast to the enhanced Erk activation evoked by the NS mutant D61G and the leukemia mutant E76K, LS mutations, like the catalytically inactive mutant C459S, impair Erk activation. B, stable expression assays. Retrovirally transduced pools of 293T cells stably expressing WT Shp2 or the indicated mutants were either starved or stimulated with EGF (50 ng/ml), lysed, and immunoblotted with phospho-Erk1/2 antibodies. The lower panels depict the same lysates blotted for total Erk1/2 and Shp2. C, time course of Erk activation in cells expressing Shp2 mutants. 293T cells stably expressing vector control (Vec), WT Shp2, or the indicated mutants were starved or stimulated with EGF (50 ng/ml) for various times, as indicated, then lysed and immunoblotted with anti-phospho-Erk1/2 antibodies. The lower panel shows equivalent loading. Note that LS mutants preferentially impair sustained Erk activation. D and E, Erk activation in response to other growth factors is also inhibited by LS mutations. 293T cells stably expressing WT Shp2 or the indicated mutants were starved and then stimulated with FGF (20 ng/ml) (D) or with PDGF (100 ng/ml) (E). Erk activation was assessed as above.
bind an upstream scaffolding adapter (or other Tyr(P) protein) to become activated. A catalytically inactive (i.e. LS) mutant should compete for binding with the endogenous WT Shp2 allele. Mathematical modeling reveals that if the concentration of the upstream adapter is limiting, there will be a less active Shp2/binding partner complex in cells expressing one LS allele and one WT allele than in an Shp2 hemizygote even if the affinity of LS and WT Shp2 for their binding proteins are the same (see supplemental material).
Nevertheless, structural analysis suggested that the LS mutants tested also should alter the interaction between the N-SH2 and PTP domains in the closed (inactive) form of the enzyme (Fig. 1D). For example, Tyr-279 contacts Tyr-62 and Lys-70 in the N-SH2 domain (Figs. 5, A  and B). Substitution of a bulky threonyl residue for Ala-461 should cause steric clash with Asp-61 and Tyr-62 of the N-SH2 domain (Figs. 5, C and  D), whereas the contortion of the PTP loop expected in the G464A mutant also should displace Asp-61 in the N-SH2 away from the catalytic cleft (Figs. 5, E and F). The destabilized PTP loop expected in the T468M mutant should, in addition to impairing catalysis, displace the N-SH2 domain (Figs. 5, G and H). Thus, in addition to inactivating Shp2 activity, LS mutations should favor the "open" form of the enzyme, thereby enabling the LS mutants to bind upstream activators preferentially (compared with WT Shp2). Consistent with this prediction, LS mutants as well as the NS mutant D61G, which is known to favor the open conformation (17,20), co-immunoprecipitated significantly with HA-tagged Gab-1 even under serum-starved conditions. In contrast, WT Shp2 bound Gab-1 primarily after growth factor stimulation (Fig.  6A). In addition, LS mutants (as well as the activating and "open conformation" mutants E76K and D61G) of Shp2 more readily formed complexes with Gab-1 in the presence of growth factor than did the WT Shp2 or the catalytically inactive but "closed conformation" mutant, C459S (Fig. 6B). These findings suggest that whereas NS mutations result in an "open and hyperactive" state of the enzyme, LS mutations are "open and inactive." An increased tendency for LS mutants to adopt the open conformation should confer more potent dominant negative effects.

DISCUSSION
Several lines of evidence show that whereas NS and neoplasia-associated PTPN11 mutants exhibit gain-of-function, LS mutants are, unexpectedly, catalytically impaired and dominant negative. Bacterially produced LS mutant proteins have markedly decreased PTP activity when assayed against standard artificial substrates. The same is true when these mutants are expressed transiently in mammalian cells and subjected to immune complex PTP assays. Under the same conditions, NS and neoplasia-associated mutants exhibit substantially increased catalytic ability. These in vitro biochemical differences in Shp2 properties are accompanied by (and presumably cause) diametrically opposite effects of NS/neoplasia-associated and LS mutants, respectively, on receptor-tyrosine kinase signaling. Transient or stable expression of NS/leukemia-associated mutants enhances, but LS mutants markedly impair Erk activation in response to multiple growth factors, including EGF, FGF, and PDGF. Consistent with their impaired catalytic activity observed in vitro and their dominant negative effects on receptor-tyrosine kinase signaling in vivo, molecular modeling studies predict that LS mutants should markedly alter the catalytic cleft of Shp2 and disrupt the basal inhibitory interaction between the N-SH2 and PTP domains, thereby giving rise to open, inactive forms of the enzyme. Mathematical modeling predicts that these open, inactive mutants should have more potent dominant negative effects (supplemental material).
Our findings raise an obvious conundrum; how do gain-of-function and dominant negative mutants cause similar disease phenotypes? Although the precise explanation for this apparent paradox remains to be determined experimentally, there are several possible explanations. Shp2 plays important roles downstream of multiple receptors and acts at several different times during development (14). We suspect that the NS and LS phenotypes result from differential effects of mutant Shp2 on different receptor-tyrosine kinase pathways at distinct developmental times. For example, previous studies suggest a possible explanation for the similar cardiac defects in LS and NS. Valves develop from specialized endocardial cushion endothelium, which undergoes epithelialmesenchymal transformation. The resultant mesenchymal cells proliferate and migrate and ultimately undergo morphogenesis, which entails substantial apoptosis (32). Epithelial-mesenchymal transformation and/or mesenchymal cell proliferation are promoted by ErbB2/ErbB3 (and possibly, ErbB3/4) complexes at ϳE10 -10.5 (33). The Ras/Erk pathway plays a pivotal role in epithelial-mesenchymal transformation/ proliferation (most likely acting downstream of ErbB2/3 and/or ErbB3/4 complexes), as illustrated by the markedly increased number of cushion mesenchymal cells in embryos lacking the RasGap NF1, the domain. This likely leads to the displacement of Asp-61, disrupting its ability to hydrogen bond to catalytic loop residues. G and H, T468M may interact with Gln-510, which normally binds to Gly-60. The more bulky methionyl residue should alter the relative position of Gly-60 and adjacent residues in the N-SH2 domain, thereby exposing the PTP domain.
product of the neurofibromatosis type I gene, Nf1 (34,35). In contrast, heparin binding epidermal growth factor activates the EGF receptor (ErbB1) at ϳE13.5, which transmits signals that help terminate mesenchymal cell proliferation and promotes morphogenesis by inhibiting Smad family transcription factors (36,37). D61G/ϩ "knock-in" embryos exhibit excess cushion mesenchymal proliferation by E12.5 (21) 5 , a phenotype very similar to the effects of Nf1 deficiency. Likewise, adenoviral expression of the NS mutant Q79R in chicken cardiac cushion explants leads to increased Erk activation and to an Erk-dependent increase in cell proliferation (38). By contrast, Shp2 hemizygosity enhances the effects of EGFR hypomorphism on cardiac valvulogenesis (36). Although other explanations remain possible, we suspect that the predominant effect of NS mutants is to enhance epithelial-mesenchymal transformation/mesenchymal cell proliferation by increasing ErbB2/3 (and/or ErbB3/4) signaling, whereas LS mutants antagonize HB-EGF/ ErbB1 signaling at later times. The differential effects of NS and LS mutants on these two signaling pathways likely reflect distinct signaling thresholds for activating the Ras/Erk (or other) pathways downstream of ErbB2/3 and Erb1, respectively, which could result from different expression levels of Shp2 binding proteins (e.g. Gab proteins) and/or other pathway regulators.
Other phenotypes common to NS and LS (e.g. facial abnormalities and short stature) might involve defective migration and/or differentiation, which might result from increased or decreased signaling involving the same pathway (e.g. either abnormally strong or weak cell adhesion could impair migration) (39). In this regard, the altered ability of Shp2 to control activation of the small G protein Rho (17,40,41) may play an important role in the pathogenesis of NS and LS. We also cannot exclude the possibility that as yet undetermined or poorly understood functions of Shp2 underlie pathogenesis of one or both of these disorders. For example, a recent study reported that a small percentage of Shp2 resides within mitochondria (42). The function(s) of mitochondrial Shp2 and the effects of disease-associated mutants on this function(s) remain to be determined.
Resolving the detailed cellular basis for the similar defects in NS and LS will require mouse modeling of LS, which is currently under way in our laboratory. However, our data argue strongly that instead of classifying diseases caused by PTPN11 mutations by current clinical criteria, characterization should be based on the biochemical effects of a given PTPN11 mutation on Shp2 activity and signaling. FIGURE 6. LS mutations preferentially associate with Gab-1 in LS transiently expressing cells. A, 293T cells transiently expressing HA-tagged Gab-1 (2 g) and WT Shp2, the NS mutant D61G, the catalytically inactive C459S mutant, or the indicated LS mutants were either starved or stimulated with EGF (50 ng/ml), lysed, and immunoprecipitated (IP) with anti-HA antibodies. Immunoprecipitates were immunoblotted with anti-Shp2 or anti-Gab-1 antibodies, respectively. Note that the NS mutant D61G and the LS mutations bind Shp2 even in the absence of growth factor stimulation, suggesting a more open conformation of the enzyme in these mutants. Interestingly, the C459S mutant binds Shp2 much like the WT, suggesting that this catalytically inactive mutant does not preferentially reside in an open conformation. V, vector. B, 293T cells transiently expressing WT Shp2, the leukemia mutant E76K, the NS mutant D61G, the catalytically inactive C459S mutant, or the indicated LS mutants were either starved or stimulated with EGF (50 ng/ml), lysed, and immunoprecipitated with anti-Shp2 antibodies. Immunoprecipitates were immunoblotted with anti-Gab-1 or anti-Shp2 antibodies, respectively. Note that the E76K mutant and the D61G and the LS mutations bind Gab-1 more readily than either the WT Shp2 or the catalytically inactive C459S mutant, also suggesting that the E76K, D61G, and the LS mutants are more likely found in an open conformation.