Sticking It to Cancer With Molecular Glue For SHP2

Abstract

Much effort has been expended to develop inhibitors against protein-tyrosine phosphatases (PTPs), nearly all of it unsuccessful. A recent report, describing a highly specific, orally bioavailable inhibitor of the PTP oncoprotein SHP2 with in vivo activity, suggests that allostery might provide a way forward for PTP inhibitor development.

Tyrosine phosphorylation, controlled by protein-tyrosine kinases (PTKs) and protein-tyrosine phosphatases (PTPs), regulates cell survival, proliferation, migration and differentiation. Not surprisingly, aberrant tyrosine phosphorylation often causes disease: for example, activating mutations/amplifications of PTKs occur in multiple malignancies. PTK inhibitors [e.g., Imatinib (BCR-ABL, others), Erlotinib (EGFR) Lapatinib (HER2) and Sunitinib (VEGFR, others)] are mainstays of “precision oncology” and among the best-selling drugs worldwide (Wu et al., 2015).

Individual PTPs can have positive (signal-enhancing) or negative (signal-inhibiting) roles, and several are implicated in cancer (Labbe et al., 2012). However, the best-validated PTP oncogene is PTPN11, encoding SHP2 (Neel, 2009). One of two SH2 domain-containing PTPs (the other is SHP1, encoded by PTPN6), SHP2 features two SH2 domains (N-SH2/C-SH2), a catalytic (PTP) domain, and a C-terminal tail with two tyrosine phosphorylation sites. SHP2 toggles between closed (inactive) and open (active) states (Barford and Neel, 1998). In the closed state, the N-SH2 is wedged into the PTP domain, blocking substrate access. Phosphotyrosyl (pTyr) peptide binding to the N-SH2 disrupts auto-inhibition, activating the enzyme (Figure 1A). SHP2 binding sites are found in receptor tyrosine kinases (RTKs) and scaffolding adapters (GAB, IRS, FRS proteins), so this “molecular switch” ensures that SHP2 is activated only at proper cellular locales. In growth factor and cytokine signaling, SHP2 acts upstream of RAS to dephosphorylate (a) still hotly debated substrate(s) and enables full activation of the ERK/MAP kinase pathway (Figure 1B). In addition, the C-terminal tyrosines of SHP2 undergo phosphorylation in response to most agonists. Tyrosyl phosphorylated SHP2 recruits GRB2/SOS, contributing to RAS activation (Neel, 2009). Furthermore, SHP2 binds immune-inhibitory receptors, including PD-1 (Pardoll, 2012), and, often in concert with SHP1, inhibits signaling from activating immunoreceptors (e.g., TCR).

Figure 1.

Schematics illustrating key features of SHP2 regulation and function.

(A) SHP2 switches from a closed (inactive) to an “open” active form upon binding to specific phosphotyrosyl peptide motifs in RTKs/scaffold proteins.

(B) Role of SHP2 in growth factor/cytokine-dependent RAS/ERK pathway activation (left) and in mediating the inhibitory function of PD-1 (and other immune-inhibitory receptors) on TCR signaling (right).

(C, D) Effects of active site-targeting inhibitor (C) or SHP09 (D) on SHP2.

Malignancies hyper-activate SHP2 in two ways (Neel, 2009). In ~1/3 of juvenile myelomyelogenous leukemia (JMML) and less frequently (1–10%) in acute leukemias, neuroblastomas and carcinomas, somatic mutations, usually affecting the N-SH2 domain, disrupt auto-inhibition. Analogous, but less activating, PTPN11 mutations can occur in the germ line, and cause ~50% of Noonan Syndrome (NS) cases (Roberts et al., 2013). Constitutively active fusion-PTKs (e.g., BCR/ABL) or amplification/over-expression of growth factors, RTKs or scaffolding adapters also drive inappropriate SHP2 activation.

Clearly, a specific inhibitor for SHP2 might have therapeutic utility, but developing PTP inhibitors presents unique challenges (He et al., 2013). Substantial binding energy is contributed by the substrate phosphate residue, which is then targeted by the highly reactive catalytic cysteine. Therefore, catalytic inhibitors must mimic phosphotyrosine and have low reactivity. Unfortunately, conventional screens for PTP inhibitors have typically recovered reactive, polar, low affinity and/or cross-reactive compounds. A few SHP2 inhibitors have been reported to have substantial in vitro potency, PTP selectivity, and beneficial effects in animal models. However, collectively these molecules have poor bioavailability and/or troublesome pharmacophores for further drug development. In addition, none have been profiled extensively for off-target effects against other enzyme families. Furthermore, where in vivo efficacy has been reported, on-target activity has not been demonstrated convincingly.

In a recent report, a group from Novartis circumvents these problems by developing allosteric SHP2 inhibitors (Chen et al., 2016). Cleverly capitalizing on SHP2 enzymology, they screen for small molecules that block activation by a pTyr-peptide but fail to inhibit the isolated PTP domain. Medicinal chemistry yields SHP099, which is potent (K_d=70 nM) and has no significant activity against a panel of other PTPs (including SHP1) and kinases. In addition, SHP099 has almost minimal activity against other enzyme systems typically associated with toxicity. An SHP099/SHP2 crystal structure shows that the inhibitor binds a heretofore unrecognized pocket in “closed” SHP2, acting like “molecular glue” to prevent N-SH2/loop/C-SH2 movements that presumably occur upon enzyme activation. SHP1 has an analogous, but larger pocket, explaining why it is not inhibited. Chen et al. also screen 250 well-annotated cancer cell lines with a deep-coverage shRNA library. Not surprisingly, lines with activated RTKs/PTK-fusions are preferentially sensitive to SHP2 depletion, while cells bearing RAS or BRAF mutations are resistant. Reassuringly, SHP099 has a similar inhibitory preference for lines with activated PTKs and blocks ERK activation. Providing unambiguous evidence of on-target action, an SHP2 mutant engineered to be drug-resistant rescues these effects. Most importantly, SHP099, administered orally, shows efficacy against an EGFR-driven cancer cell line xenograft and a FLT3-ITD-AML patient-derived xenograft (PDX). Remarkably, treated mice also had no evidence of toxicity.

These important findings provide a path to the clinic, although not without some caveats. Neoplasms driven by activated PTKs clearly represent prime targets for SHP2 inhibitor testing. SHP2 inhibition also could abrogate/prevent emergence of resistance to BRAF and MEK inhibitors, which is often caused by RTK activation (Prahallad et al., 2015). But while SHP099 caused regression of the AML PDX, it only arrested growth of the solid tumor xenograft, so drug combinations will likely be necessary for clinical efficacy. Chen et al. did not explore potential immune-stimulatory effects that could result from blocking immune-inhibitory receptor action. SHP2 probably does not mediate all of these effects, in which case an SHP2 inhibitor might be less efficacious than PD1/PDL1 antibodies (Pardoll, 2012). Alternatively, it could have greater/different efficacy because SHP2 binds several immune-inhibitory receptors. If its immune-stimulatory effects are significant, an SHP2 inhibitor could deliver a “double whammy” against PTK-driven tumors and/or have a broader spectrum of anti-neoplastic activity. Also unclear is whether SHP099-like agents can treat JMML/other neoplasms caused by PTPN11 mutations. Such mutations destabilize the closed form of SHP2, so compounds more potent than SHP099 might be required to provide sufficiently strong “molecular glue.” In this context, it is encouraging that the Novartis group has published patents describing additional inhibitors that act like SHP099, several with ~10-fold lower IC50s, which could also be tested in mutant PTPN11-driven disease models. Even if allosteric SHP2 inhibitors cannot target cancer-associated SHP2, they might inhibit the less activated mutants that cause NS, and be useful for treating severe cases of this disorder.

The results of many previous biological studies predict that the side effects of SHP099 and catalytic SHP2 inhibitors should be quite different. Catalytic inhibitors permit SHP2 recruitment, and should preserve any potential PTP-independent functions of SHP2 (Figures 1C and 1D). Indeed, active site inhibitors might have side effects similar to (or a subset of) phenotypes seen in NS with multiple lentigines (NS-ML), a PTPN11-associated developmental disorder caused by catalytically impaired SHP2 (Neel, 2009). SHP099, by blocking N-SH2 access, should act more like a “chemical null.” In fact, its apparent lack of toxicity is actually somewhat surprising, given that conditional Ptpn11 knockout mice display significant, often severe phenotypes (Neel, 2009). At efficacious doses, SHP099 probably does not impair SHP2 action as completely as Ptpn11 knockout, and some knockout phenotypes might reflect developmental functions. Nevertheless, it will be important to monitor for the toxicities predicted by mouse genetic analyses in the clinic.

Besides its obvious potential clinical benefits, the advent of a bona fide SHP2 inhibitor should help to resolve several remaining questions about this important enzyme. Unlike knockout, knockdown and dominant negative approaches, inhibitors act immediately, allowing interrogation of the direct effects of signaling components. Experiments using SHP099 could help identify key SHP2 substrates, as well as clarify whether the reported effects of SHP2 deficiency on pathways other than the RAS/ERK cascade are direct or compensatory.

PTP inhibitor development has sent many companies to the graveyard. Beyond SHP099, the results of Chen et al. suggest an obvious strategy for designing allosteric SHP1 inhibitors. In addition, recent elegant work by Tonk’s group shows that the natural product MSI-136, though not orally active, is an allosteric PTP1B inhibitor with in vivo activity (Krishnan et al., 2014). Allostery could provide a new way forward to developing a wide array of specific PTP inhibitors, one that exploits the striking diversity of PTP regulatory domains while avoiding the perils of the PTP active site.