Off‐target inhibition by active site‐targeting SHP2 inhibitors

Due to the involvement of SHP2 (SH2 domain‐containing protein‐tyrosine phosphatase) in human disease, including Noonan syndrome and cancer, several inhibitors targeting SHP2 have been developed. Here, we report that the commonly used SHP2 inhibitor NSC‐87877 does not exhibit robust inhibitory effects on growth factor‐dependent MAPK (mitogen‐activated protein kinase) pathway activation and that the recently developed active site‐targeting SHP2 inhibitors IIB‐08, 11a‐1, and GS‐493 show off‐target effects on ligand‐evoked activation/trans‐phosphorylation of the PDGFRβ (platelet‐derived growth factor receptor β). GS‐493 also inhibits purified human PDGFRβ and SRC in vitro, whereas PDGFRβ inhibition by IIB‐08 and 11a‐1 occurs only in the cellular context. Our results argue for extreme caution in inferring specific functions for SHP2 based on studies using these inhibitors.

SHP2, encoded by PTPN11, is a classic protein-tyrosine phosphatase (PTP), comprising two tandem SH2 domains (N-SH2 and C-SH2, respectively), followed by a catalytic (PTP) domain and a C-terminal tail [1]. In the absence of appropriate cellular stimuli, SHP2 resides in an auto-inhibitory 'closed' structure, as a consequence of intramolecular interactions between the N-SH2 and PTP domains [2]. Upon stimulation of cells with appropriate growth factors, such as plateletderived growth factor (PDGF) or epidermal growth factor (EGF), the cognate receptor tyrosine kinase (RTK) is activated, resulting in receptor trans-phosphorylation [3,4], as well as phosphorylation of scaffolding proteins, such as GAB family members [5]. SHP2 binds to phosphorylated RTKs and/or scaffold proteins via SH2 domain/phosphotyrosyl (pY) interactions, which are incompatible with intramolecular N-SH2/PTP interaction and result in an enzymatically activated 'open' structure [1,2].
Once activated, SHP2 functions as a key positive regulator of RTK-evoked signal transduction, acting upstream of RAS in the ERK MAP kinase pathway [6,7]. Phosphatase activity is required for SHP2 action [8], although the key substrate(s) remain controversial. In addition, in response to most agonists, SHP2 itself undergoes tyrosyl phosphorylation on two C-terminal tyrosines, which then function as GRB2 binding sites. This 'adapter' function for SHP2 enhances RAS/ERK signaling [9].
Protein-tyrosine phosphatase family members share a conserved PTP catalytic domain [25], so development of active site SHP2 inhibitors has focused largely on achieving specificity versus other PTPs. The above active site-targeting inhibitors, except NSC-87877, reportedly are highly specific for SHP2, compared with other PTPs, including SHP1, which has the highest similarity to SHP2 [21][22][23]. However, potential off-target effects of these molecules against other enzyme families (e.g., kinases) have, in general, not been reported.
We tested the effects of the commonly used active site SHP2 inhibitors NSC-87877, IIB-08, 11a-1, and GS-493 on PDGF signaling in fibroblasts and found that they either failed to effectively inhibit SHP2 function in cells (NSC-87877) or had off-target effects on PTKs. These results argue strongly that such agents cannot be used alone to infer SHP2 functions in health or disease.

Growth factors and chemical compounds
Recombinant human PDGF-BB and EGF were purchased from Peprotech. 4-OHT was purchased from Sigma. IIB-08 and 11a-1 were provided by Z.Y. Zhang (Purdue University). GS-493 was provided by J. Rademann (Freie Univer-sit€ at Berlin) or was purchased from Bioduro. NSC-87877 was purchased from Millipore. SHP099 was purchased from Alputon Inc.

Expression constructs, infection, and sorting
Retroviral expression vectors for wild-type SHP2 (WT SHP2) and the mutant SHP2 C459E were generated by subcloning human PTPN11 cDNA into pMSCV-IRES-EGFP (Clontech). Ptpn11 fl/fl MEFs expressing WT or mutant SHP2 were generated by retroviral infection, according to the manufacturer's protocol (pMSCV retrovirus system; Clontech), followed by fluorescence-activated cell sorting (FACS) for EGFP-positive cells.

In vitro kinase assays
In vitro kinase assays were performed by the SelectScreen TM Biochemical Kinase Profiling Service with the Z'-LYTE Kinase Assay (Thermo, Waltham, MA, USA). Purified recombinant human PDGFRb or human SRC was incubated with Z'-LYTE kinase substrate in the presence of serial dilutions of each inhibitor in kinase reaction buffer (50 mM HEPES, pH 7.5, 0.01% BRIJ-35, 10 mM MgCl 2 , 2 mM MnCl 2 , 1 mM EGTA, 1 mM DTT) with 100 lM ATP (PDGFRb) or 50 lM ATP (SRC) at room temperature for 1 h. Relative inhibition for each condition (average of technical duplicates) was calculated by setting the activity in the absence of inhibitor as 0%. IC 50 s were obtained by fitting the data to sigmoidal dose-response curves.

Results and Discussion
SHP2 catalytic activity is necessary for PDGFdependent MAPK activation but not for PDGFR phosphorylation Exposure of PDGF-BB to cells causes dimerization and activation of PDGFRb on the plasma membrane, resulting in trans-phosphorylation of multiple receptor tyrosine residues, including Y579, Y716, Y857, and Y1009 [4]. These pYs serve as binding sites for SH2 domain-or PTB domain-containing signaling proteins, such as SHP2, which initiate downstream signal transduction [4]. To evaluate the role of SHP2 catalytic activity in PDGF signaling, we employed immortalized mouse embryo fibroblasts (MEFs) from Ptpn11 fl/fl animals, which express Cre-ER Tam [7]. MEFs were reconstituted with wild-type (WT) SHP2 or a catalytically inactive SHP2 mutant (C459E) and then were treated with 4-OHT (1 lM) for 4 days to delete endogenous Ptpn11. Upon PDGF-BB-treatment (50 ngÁmL À1 ) after serum starvation, parental (undeleted) cells showed agonist-evoked phosphorylation/ activation of the ERK MAPK pathway components MEK1 and ERK1/2, whereas these were strongly suppressed by SHP2 depletion (Fig. 1). Suppressed ERK signaling in Ptpn11 knockout cells was rescued by reexpression of WT SHP2. By contrast, SHP2 C459E did not restore PDGF-dependent ERK phosphorylation, despite expression at levels comparable to endogenous SHP2 in control cells (Fig. 1), confirming a requirement for SHP2 catalytic activity in PDGFR-induced ERK activation in fibroblasts. Notably, overall PDGFR tyrosyl phosphorylation, as well as phosphorylation of multiple specific PDGFRb tyrosyl residues (pY579, pY716, pY857, and pY1009 of PDGFRb), was unaffected, or slightly increased, by the absence of catalytically active SHP2 in Ptpn11-knockout cells (Fig. 1). These data indicate that SHP2 does not enhance (and might inhibit some) PDGFRb tyrosyl phosphorylation events (at these time points) in this immortalized MEF line.
The other active site inhibitors tested efficiently abolished PDGF-evoked MEK1 phosphorylation, but also strongly suppressed ligand-dependent tyrosine phosphorylation of PDGFRb (Fig. 2D). As Ptpn11 knockout (Fig. 1) did not attenuate PDGFR phosphorylation, we tested these inhibitors in MEFs. In contrast to the effects of SHP2 depletion or replacement of SHP2 with a catalytically inactive SHP2 mutant, IIB-08, 11a-1, and GS-493 suppressed ligand-dependent PDGFRb phosphorylation in wild-type MEFs. Even more importantly, these inhibitors also blocked PDGFRb phosphorylation in cells lacking SHP2 (Fig. 2E). By contrast, treatment of MEFs with SHP099 suppressed PDGF-dependent MEK/ERK activation without suppressing PDGFRb phosphorylation (Fig. 2F). These observations show unambiguously that suppression of PDGFRb phosphorylation by these active site inhibitors at their recommended concentrations is independent of SHP2, and indicate that they have off-target effects, at least in fibroblasts.
Taken together, our data indicate that NSC-87877or at least versions of this compound available commercially-lacks clear inhibitory effects on the ERK MAPK pathway, at least in the context of PDGF or EGF stimulation of fibroblasts and HEK293T cells. By contrast, GS-493, IIB-08, and 11a-1 suppress PDGFRb phosphorylation independently of SHP2 and thereby block downstream signaling in PDGF-stimulated cells. GS-493 suppresses ligand-evoked transphosphorylation of PDGFRb, likely by directly inhibiting the PDGFRb kinase domain. GS-493 also inhibits SRC in vitro, and given the conservation of kinase domains, might well inhibits other PTKs as well. How IIB-08 and 11a-1 impair PDGFRb activation in cells, while not inhibiting the PDGFRb kinase domain in vitro, remains unclear. Potential mechanisms include (a) modulation of receptor affinity for PDGF, (b) inhibition of receptor dimerization or allosteric inhibition via another region of PDGFRb not included in the recombinant kinase domain, (c) activation of PTP(s) that dephosphorylates PDGFRb, or (d) PDGFRb inhibition via metabolite(s) generated within cells. As with GS-493, additional potential off-target effects of IIB-08 and 11a-1 on other PTKs cannot, and should not, be excluded.
Inhibiting either SHP2 or the PDGFR can inhibit RAS/ERK/MAPK signaling. Hence, the biological effects of the above inhibitors are difficult, if not impossible, to attribute to SHP2 inhibition. Nevertheless, these inhibitors have been used extensively to probe SHP2 action (reviewed in refs. [26,27]), often as major perturbants of SHP2 function. By contrast, the allosteric inhibitor SHP099 does not have off-target effects on SRC or other tyrosine kinases in vitro [24] or in the cellular contexts described here. Even so, comparing the effects of SHP099 and SHP2 depletion (via chemical degradation) indicates that SHP099 also can have offtarget effects in at least some settings [28].
In summary, the above study and our results indicate that reports using SHP2 inhibitors as the major means of inferring SHP2 function must be re-evaluated and potentially re-interpreted. For SHP099-like compounds, defined drug-resistant mutants can be used in rescue experiments to demonstrate specificity [24]. Current allosteric SHP2 inhibitors act as 'molecular glue' stabilizing the closed, inactive form of the enzyme, and thereby act like 'chemical nulls' that potentially block multiple, if not all, aspects of SHP2 function [29]. For this reason, it would certainly be useful to have validated 'on-target' catalytic inhibitors. The off-target effects revealed here argue that additional efforts are necessary to achieve this goal, and emphasize the need for careful control experiments and multiple lines of evidence to confidently assign functions to SHP2 using inhibitor approaches.