Dissecting protein tyrosine phosphatase signaling by engineered chemogenetic control of its activity

Protein tyrosine phosphorylation mediates a variety of cellular processes by affecting enzyme activity, protein localization, and the formation of specific signaling complexes. Protein tyrosine kinases have been studied in depth for their roles in physiological and pathological conditions. However, protein tyrosine phosphatases (PTPs) have proven more enigmatic while remaining highly important for regulation of critical tyrosine phosphorylation (Tiganis and Bennett, 2007; Tonks, 2013; Tonks and Neel, 2001). Particularly challenging is identification of dynamic events triggered by a phosphatase immediately after its activation. Conventional approaches, such as expression of a constitutively active or catalytically dead phosphatase, lack temporal control and can only provide information on the long-term effects of these phosphatases. While specific and bioavailable phosphatase inhibitors have been developed (Liu et al., 2020; Liu et al., 2019; Ran et al., 2016; Vazhappilly et al., 2019), these too have limitations. Specific inhibitors provide temporal control over activity but reveal only how the cell responds to a loss of phosphatase activity. Thus, the current toolset for protein phosphatases remains limited and dissecting the dynamic effects of phosphatase signaling still presents a challenge.

Src homology-2 (SH2) domain-containing protein tyrosine phosphatase 2, Shp2, is a ubiquitously expressed phosphatase containing two tandem SH2 domains, a phosphatase domain, and a flexible C-terminal tail (Sun et al., 2013; Guo and Xu, 2020). Regulation of Shp2 within cells is dictated by the autoinhibitory interaction between the N-SH2 domain and the phosphatase domain. This autoinhibition is disrupted through N-SH2 binding of phosphorylated tyrosine on Shp2-interacting partners (Anselmi and Hub, 2020; Benjamin et al., 2003). The canonical function of Shp2 is activation of the mitogen-activated protein kinase (MAPK) signaling cascade, Ras-Raf-MEK-ERK. This pathway is dysregulated in many RASopathies such as Noonan syndrome, Noonan syndrome with multiple lentigines, and juvenile myelomonocytic leukemia (Matozaki et al., 2009; Schaeper et al., 2000; Serra-Nedelec et al., 2012; Tajan et al., 2015; Yu et al., 2013). Shp2 also contributes to the regulation of cell migration and cell adhesion to extracellular matrix (Cao et al., 2019; Grosskopf et al., 2015; Hartman et al., 2013; Huang et al., 2012; Lee et al., 2013; Mañes et al., 1999; Sausgruber et al., 2015). However, these processes are dynamic and often transient in nature. Thus, dissection of Shp2-mediated cell migration, adhesions, and protrusions requires tools that enable tight temporal control of its activity. Such tools will uncover short-term and long-term effects of Shp2 activation in regulation of physiological and pathological processes. In addition to the critical role of timing, functions of Shp2 are also dictated by the plethora of signaling complexes it forms. Defining the role of specific Shp2-mediated signaling complexes is needed to understand how Shp2 regulates different cellular processes. However, these studies are challenging due to paucity of tools for manipulation of Shp2 signaling.

Engineered allosteric regulation offers unique advantages in studying cell signaling due to its temporal and specific control of a protein of interest (Hongdusit et al., 2020; Karginov et al., 2010; Karginov et al., 2014). A chemogenetic approach, the Rapamycin Regulated (RapR) system, has been successfully applied to regulate specific kinase activity to investigate dynamic signaling events (Chu et al., 2014; Karginov et al., 2010; Karginov et al., 2014; Klomp et al., 2019). The RapR system utilizes rapamycin-mediated dimerization of FRB and an insertable FKBP12 (iFKBP) to modulate the activity of a protein of interest. Inserting a highly dynamic iFKBP switch into the catalytic domain of a protein of interest results in allosteric disruption of critical catalytic residues (Karginov et al., 2010). Stabilization of iFKBP upon rapamycin and FRB binding recovers catalytic function and acutely activates the protein of interest. Precise temporal control of activation enables investigation of dynamic signaling events downstream of a specific protein.

We sought to overcome the limitations of current tools for studying phosphatases by developing a method for manipulation of phosphatase activity with tight temporal control and ability to control phosphatases within specific signaling complexes. Utilizing the RapR system, we engineered and optimized allosterically regulated Shp2, PTP1B, and PTP-PEST tyrosine phosphatases. We then implemented RapR-Shp2 to investigate dynamic Shp2 signaling and the specific role of Shp2 in cell migration and spreading. Using this system, we revealed transient morphological changes following the activation of RapR-Shp2 that could not be recapitulated with constitutively active (CA) Shp2. This tool also enabled interrogation of possible parallel Shp2-dependent signaling pathways mediating these dynamic morphological changes.

We aimed to develop an allosterically regulated Shp2 using rapamycin-induced dimerization of iFKBP and FRB (Fig. 1 A; Karginov et al., 2010). Insertion of the highly dynamic iFKBP switch within the catalytic domain disrupts the catalytic function in the absence of rapamycin. Activity is recovered following the addition of rapamycin and stabilization of iFKBP by FRB/rapamycin binding (Fig. 1 A). To eliminate autoinhibitory regulation and dependence on upstream endogenous pathways of Shp2, we employed a previously described activating mutation (D61A), this ensured activation solely by rapamycin (Chichger et al., 2015; Qiu et al., 2014). Three potential insertion sites were identified in flexible loops within the Shp2 catalytic domain: Val⁴⁰⁶, Thr²⁸⁸, and Lys³¹⁷ (Fig. 1 B). The insertion sites were selected for their structural coupling to catalytically relevant regions (WPD for Val⁴⁰⁶, substrate-binding loop for Thr²⁸⁸ and Lys³¹⁷), while also avoiding potential steric disruption of SH2 domain-mediated protein–protein interactions. The difference in length of the unstructured insertion loops allowed us to test how varying degrees of flexibility affect allosteric control by iFKBP. Both Thr²⁸⁸ and Val⁴⁰⁶ are in shorter unstructured loops while Lys³¹⁷ is in a long unstructured loop. In the shorter Val⁴⁰⁶ and Thr²⁸⁸ insertion loops, an additional secondary structure disruptor, Gly residue, was added at the junction of Shp2 and iFKBP. Previous studies indicated that insertion of short linkers such as Gly or G-S/P-G can improve the regulation of catalytic function through decreased disruption of protein structure by iFKBP insertion (Karginov et al., 2010). Using a cell-free in vitro assay with immunoprecipitated Shp2 and purified phosphorylated N-terminal fragment of paxillin, we identified Val⁴⁰⁶ insertion site as the best candidate for allosteric regulation of Shp2 phosphatase activity (Fig. 1, C and D). Insertion of iFKBP at the Thr²⁸⁸ site failed to fully activate the phosphatase with rapamycin treatment (Fig. 1, C and D). The Lys³¹⁷ site retained phosphatase activity in the absence of rapamycin (Fig. 1, C and D). Thus, the length of the insertion loop and proximity to catalytic residues are correlated with the dynamic range of regulation, where shorter loops enabled tighter regulation of catalytic activity.

Engineered Shp2 with iFKBP insertion at Val⁴⁰⁶ retained low residual activity in the absence of rapamycin as compared to dominant negative (DN) Shp2 (Fig. 1, C and D). Thus, to optimize the regulation of this construct, we further modified the linkers connecting iFKBP to Shp2 at Val⁴⁰⁶. Previous work using the RapR system in kinases demonstrated the importance of these linkers in optimized control of catalytic function (Karginov et al., 2010). The flexibility of the linkers was modified by insertion of an additional Gly-Ser-Gly fragment and truncation of Shp2 amino acids 405 and 407 on each side of iFKBP insert (Fig. 2 A). Coupling a longer GSG linker with deletion of amino acids 405–407 resulted in the best disruption of phosphatase activity in the absence of rapamycin and best recovery of activity in the presence of rapamycin (Fig. 2, B and C). For simplicity, this optimized version, with the replacement of residues 405–407 by iFKBP with GSG linkers, will be referred to as RapR-Shp2. Evaluation of substrate dephosphorylation over time revealed that RapR-Shp2 exhibits kinetics comparable to CA-Shp2 (Fig. 2, D and E). Our analysis estimated 1.2-fold increase in K_m and 2.2-fold increase in V_max suggesting only a small reduction in substrate binding affinity while catalytic efficiency was retained.

To interrogate the potential mechanistic effect of iFKBP on the catalytic domain of Shp2, we performed molecular dynamics simulations of RapR-Shp2 with and without FRB and rapamycin, and the wild-type Shp2 catalytic domain. These simulations facilitated the analysis of the potential effect of the iFKBP insertion on substrate recognition or substrate coordinating loops within Shp2. In the absence of rapamycin, the highly dynamic iFKBP insertion increases the root-mean-square fluctuation (RMSF) throughout the catalytic domain, suggesting increased structural plasticity that could disrupt activity (Fig. 3, A and B). Simulations of the RapR-Shp2-rapamycin-FRB complex reveal an overall decrease in RMSF, suggesting stabilization of this domain and potential recovery of catalytic function (Fig. 3, A and C). The substrate binding loop containing the KNRY sequence shows the greatest difference between wild-type Shp2 and unbound RapR-Shp2, while the WPD loop and catalytic residues do not appear to be affected by the insertion (Fig. 3, A–D). The RMSF of the catalytic domain of RapR-Shp2-rapamycin-FRB complex, including the substrate binding loop, closely resembles that of the wild-type catalytic domain of Shp2 (Fig. 3, C and D). We observed the loop immediately adjacent to the substrate binding loop exhibited elevated RMSF even after complexation with FRB and rapamycin. This may contribute to reduced substrate binding in the RapR-Shp2 as was observed in our kinetics analysis. These results reveal a potential mechanism of allosteric regulation of Shp2 through dynamic disruption of the substrate binding regions of the catalytic domain as well as provide information on the effect of the RapR insertion on catalytic efficiency.

Shp2 is well established to activate the Ras-Raf-MEK-ERK pathway downstream of growth factor signaling (Benjamin et al., 2010; Benjamin et al., 2003; Meng et al., 2005; Vazhappilly et al., 2019). Overexpression of CA-Shp2 (D61A) increases levels of active ERK1/2 (Yang et al., 2019). To validate that RapR-Shp2 can emulate Shp2 signaling, we activated overexpressed RapR-Shp2 in HEK293T cells (Fig. S1, D and E). Following activation for an hour, RapR-Shp2 induced ERK phosphorylation comparable to the effect of CA-Shp2 (Fig. 4 A and Fig. S1 A). In the absence of rapamycin, RapR-Shp2 expressing cells showed basal levels of ERK phosphorylation (Fig. 4 A and Fig. S1 A). Thus, our engineered RapR-Shp2 tool can induce canonical Shp2 signaling in live cells.

Activation of ERK 1/2 indicates that RapR-Shp2 likely recognizes endogenous substrates and forms canonical protein-protein interactions. To further validate these functions, we assessed whether RapR-Shp2 mediates dephosphorylation of known Shp2 substrates in live cells. Here, we used A431 cells because of their enhanced EGFR activity. Activation of RapR-Shp2 in these cells resulted in dephosphorylation of several established Shp2 substrates: EGFR (Y⁹⁹²), FAK (Y⁵⁷⁶/Y⁵⁷⁷), PLCγ (Y⁷⁸³), ROCK II (Y⁷²²), and Gab1 (Y⁶²⁷ and Y⁶⁵⁹; Fig. 4 B and Fig. S1 B, D, and E; Agazie and Hayman, 2003; Cunnick et al., 2001; Lee and Chang, 2008; Ren et al., 2011; Tsutsumi et al., 2006; Vidyasiri Vemulapalli et al., 2019,Preprint; Witsenburg et al., 2013). In addition, we ensured that insertion of iFKBP into the catalytic domain did not disrupt the function of SH2 domains of Shp2. This was established by evaluating protein-protein interactions of RapR-Shp2 with Gab1, a known binding partner for both Shp2 SH2 domains, as well as a known substrate and binding partner EGFR (Fig. 4 C and Fig. S1 C; Agazie and Hayman, 2003; Cunnick et al., 2001; Hartman et al., 2020). These data validated the ability of RapR-Shp2 to emulate wild-type Shp2 phosphatase activity and protein-protein interactions in living cells.

Shp2 is implicated in cell migration and spreading both through upstream signaling and direct regulation of focal adhesions and cytoskeletal rearrangements (Cao et al., 2019; Grosskopf et al., 2015; Hartman et al., 2013; Huang et al., 2012; Lee et al., 2013; Mañes et al., 1999; Mizutani et al., 2007; Ren et al., 2004; Sausgruber et al., 2015; Tan et al., 2019; Tsutsumi et al., 2006; Wu et al., 2016; Yu et al., 1998; Zhang et al., 2013). However, due to the approaches employed in these studies, the specific transient effects of Shp2 activation on cell morphology were not established. Application of RapR-Shp2 allowed us to overcome these limitations. We evaluated the effects of Shp2 activation on several morphological processes defining cell morphodynamics: formation and retraction of broad lamellipodia-like protrusions, cell spreading, and cell migration. Activation of RapR-Shp2 in HeLa cells led to immediate stimulation of cell spreading and protrusive activity (Fig. 5, A and B; Fig. S1, D and E; and Fig. S2 A). Cell spreading stops 30 min after activation (Fig. 5, A and C) coinciding with an increase in cell retraction (Fig. 5 D and Fig. S2 B). These events were accompanied by a transient increase in cell migration which returned to basal levels after 3 h of RapR-Shp2 activity in HeLa cells (Fig. 5 E and Video 1). HeLa cells expressing only FRB did not exhibit these morphological changes in response to rapamycin (Fig. 5, A–E; and Fig. S2, A and B). Furthermore, cells expressing CA-Shp2 (D61A; Fig. S2 E) did not exhibit increased basal migration or significant changes in spreading or migration after rapamycin treatment (Fig. S2, C and D). RapR-Shp2 was showing same diffused localization pattern as CA-Shp2 (Fig. S2 E). Thus, RapR-Shp2 revealed a series of transient morphological changes which can be observed only through acute Shp2 activation.

Activation of ERK has previously been established to contribute to cell spreading and migration (Tanimura and Takeda, 2017). As Shp2 canonically activates ERK, we reasoned that Shp2-induced spreading and migratory phenotypes may be mediated through activation of ERK. Inhibition of MAPK signaling through the use of a MEK or ERK1/2 inhibitors, trametinib (Rutkowski et al., 2015), and FR180204, respectively (Ohori et al., 2005), resulted in ablation of RapR-Shp2-induced cell spreading (Fig. 6, A and B). However, some cell migration persisted in the absence of ERK activity (Fig. 6 C). In addition, increased protrusion and retraction activity are observed despite ERK inhibition, though trametinib-treated cells exhibit a 10-min delay in induction of protrusive activity (Fig. 6, D–F). These results suggest a parallel ERK-independent pathway for Shp2-mediated protrusions and cell migration.

Shp2 catalytic function is a critical mediator of signaling in cells. However, the specificity of individual pathways downstream of Shp2 is often mediated by its SH2 domains (Anselmi et al., 2020; Barua et al., 2007; Benjamin et al., 2010; Meng et al., 2005; Paardekooper Overman et al., 2014; Sha et al., 2013; Guo and Xu, 2020; Wu et al., 2016; Yu et al., 2013). We interrogated the role of each SH2 domain in Shp2-induced morphodynamic changes by introducing previously described R32L and R138L mutations into the SH2 domains of RapR-Shp2, ablating phospho-tyrosine binding of the N- and C-SH2 domains, respectively. This disruption of protein–protein interactions prevents downstream signaling (Sun et al., 2013). Both RapR-Shp2 mutants failed to activate ERK in HEK293 cells, confirming that both SH2 domains are required for stimulation of MAPK pathway (Fig. 7 A and Fig. S3 A; Sha et al., 2013). Activation of the SH2 domain mutants of RapR-Shp2, however, revealed a specific role for the N-SH2 domain in ERK-independent migration. Loss of phospho-tyrosine binding by the N-SH2 domain (R32L) abolished both the spreading and migratory phenotypes while significantly reducing protrusive and retractive activity (Fig. 7, B–F and Fig. S3, B and C). In contrast, the C-SH2 domain mutant (R138L) still induced cell migration, although to a lesser extent than that of wild-type RapR-Shp2 (Fig. 5 E and Fig. 7 D). Furthermore, R138L RapR-Shp2 was incapable of inducing cell spreading (Fig. 7, B and C). These results suggested that the N-SH2 domain alone was sufficient to mediate ERK-independent cell migration. To further confirm this ERK independence, we activated R138L RapR-Shp2 in the presence of trametinib and again observed induction of cell migration, but not cell area increase, following activation (Fig. S3, E–G).

Shp2 can signal through small GTPase RhoA by directly regulating its downstream effector RhoA-dependent kinase, ROCK II, through dephosphorylation of an inhibitory phospho-tyrosine (Lee and Chang, 2008; Xu et al., 2019). Our analysis shows that RapR-Shp2 mediates dephosphorylation of ROCKII in living cells (Fig. 4 B). Thus, this pathway presents a potential mechanism for ERK-independent Shp2-mediated regulation of cell migration (Lee and Chang, 2008; Xu et al., 2019). Indeed, inhibition of ROCK II blocked ERK-independent migration induced by the C-SH2 domain mutant of RapR-Shp2 (R138L; Fig. 7 D). R138L RapR-Shp2 still induced protrusive and retractive activity in the presence of ROCK II inhibitor (Fig. 7, E and F; and Fig. S3, D and G). These data suggest that the N-SH2 domain of Shp2 independently mediates the necessary signaling complexes to induce ROCK-dependent migration following activation of Shp2 phosphatase activity.

The morphological changes observed with the individual SH2 domain mutants suggest that cell spreading and migration are mediated by Shp2 signaling in different protein complexes. Thus, we investigated Shp2 activation within specific signaling complexes that may contribute to these different morphological changes. One advantage of the RapR system is that the regulated protein of interest can be specifically targeted to a known interacting partner using the RapRTAP method (rapamycin-regulated targeted activation of pathways; Karginov et al., 2014). By fusing FRB to a specific binding partner, the RapR protein can be activated within a designated complex (Fig. 8 A). Thus, we employed the RapRTAP system to dissect the role of Shp2 complex-specific signaling in cell spreading and migration.

Shp2 has been well established to bind GRB2-associated binding (Gab) proteins 1 and 2 leading to Shp2 activation and downstream signaling (Meng et al., 2005; Ren et al., 2004; Schaeper et al., 2000; Zhang et al., 2013). Shp2 is also known to play a role in focal adhesion complexes, with studies identifying Shp2 interactions with a variety of focal adhesion proteins (Hartman et al., 2013; Huang et al., 2012; Lee et al., 2013; Mizutani et al., 2007; Paardekooper Overman et al., 2014; Tan et al., 2019; Tsutsumi et al., 2006; Yu et al., 1998). Furthermore, Shp2 signaling through Gab1, Gab2, and in focal adhesion complexes, has been implicated in regulation of cell spreading and migration (Hartman et al., 2013; Meng et al., 2005; Ren et al., 2004; Zhang et al., 2013). However, the specific contributions of each of these complexes in dynamic morphological changes remain to be dissected. Thus, we interrogated the role of each of these pathways in Shp2-induced morphodynamic changes.

FRB was fused to Gab1 and Gab2 at the N-terminus, enabling rapamycin-mediated recruitment and activation of RapR-Shp2 (Fig. 8 A). We introduced Y628F/Y660F and Y604F/Y634F mutations into Gab1 and Gab2, respectively, to eliminate known Shp2 binding sites and prevent interaction with endogenous protein (Cunnick et al., 2001; Liu et al., 2001). The R32L mutant of RapR-Shp2 was used for targeting since it cannot induce any morphodynamic changes (Fig. 7, B–D). This design ensured that engineered Shp2-Gab signaling pathway will be regulated only by rapamycin and prevented any influence by endogenous components. Targeting R32L RapR-Shp2 to either mutant Gab1 or Gab2 stimulated an increase in cell area (Fig. 8, B and C; and Fig. S4, G and H). Following the initial increase in protrusions (Fig. 8 D; and Fig. S4, A and C), R32L RapR-Shp2 targeted to Gabs initiated cell migration, showing a significant increase in the path length over the first 2 h after activation (Fig. 8 E; and Videos 2 and 3). This migration occurred concurrently with increases in retractive activity (Fig. S4, B and D). Migration induced by R32L RapR-Shp2 targeted to Gabs was attenuated relative to unmodified RapR-Shp2 activation (Fig. 5 E), suggesting that localization of active Shp2 to Gabs constitutes only a part of Shp2 signaling involved in these morphological changes.

Next, we interrogated the effects of Shp2 signaling in focal adhesions on cell spreading and migration. One of the proposed binding partners and substrates for Shp2 in focal adhesions is FAK, which is phosphorylated at Y³⁹⁷ to facilitate Shp2 association (Chichger et al., 2015; Hartman et al., 2013). Thus, we employed a FAK construct containing FRB inserted at Y³⁹⁷ (FAK(397)-FRB) to recruit RapR-Shp2 to focal adhesions and activate it. Activation of R32L RapR-Shp2 in complex with FAK induced robust cell spreading (Fig. 8, B and C; and Fig. S4 I and Video 4). However, it did not induce cell migration (Fig. 8 E). This signaling pathway stimulated a transient increase in protrusive activity, which occurred concurrently with cell spreading. It failed to induce prolonged protrusive and retractive activity associated with cell migration (Fig. 8 D; and Fig. S4, E and F). These data indicate that Shp2 regulates cell spreading by signaling in focal adhesion complexes.

Our results demonstrate that Shp2-mediated cell spreading depends on ERK activity (Fig. 6, A and B). Activation of R32L RapR-Shp2 in complex with FAK resulted in a transient increase in ERK phosphorylation 10–30 min after the addition of rapamycin (Fig. S5 A), concurrent with the observed cell spreading. This suggests that Shp2 activation in focal adhesions induces cell spreading, at least in part, through activation of ERK. Indeed, inhibition of MEK activity by Trametinib significantly reduced cell spreading triggered by targeting R32L RapR-Shp2 to FAK(397)-FRB (Fig. S5 B). This evidence suggests that Shp2 phosphatase activity within focal adhesions induces cell spreading in a MAPK pathway-dependent manner.

Our data show that targeting the N-SH2 mutant of RapR-Shp2 (R32L) to FAK stimulates ERK activation and cell spreading, but not cell migration (Fig. 8, B, C, and E; and Fig. S5 A). By contrast, the C-SH2 mutant of RapR-Shp2 (R138L) signals through N-SH2 domain to mediate ERK-independent cell migration through ROCK II, but not cell spreading (Fig. 7, B–D). Thus, we hypothesized that the combination of the N-SH2-mediated pathway with Shp2 signaling in focal adhesions should induce cell spreading and migration (Fig. 9 A). Indeed, targeting R138L RapR-Shp2 to FAK(397)-FRB stimulated robust and immediate spreading and protrusions (Fig. 9, B and D) as well as induced cell migration (Fig. 9 C and Video 5). Unlike R32L RapR-Shp2 targeted to FAK, elevated protrusive activity was sustained after the initial spike in protrusions (Fig. 9 D). Retraction activity is also briefly stimulated above basal levels, contrasting the phenotypes observed with R32L RapR-Shp2 (Fig. 9 E). These data indicate that combined N-SH2-mediated protein-protein interactions and phosphatase activity in focal adhesion protein complexes drives both cell spreading and cell migration.

We aimed to demonstrate the broad applicability of this approach for regulation of PTPs. Using sequence and structural homology, we identified potential sites for insertion of iFKBP in tyrosine phosphatases PTP1B and PTP-PEST (Fig. 10 A). As with RapR-Shp2, we evaluated flexible linkers of varying lengths to optimize regulation of each of these PTPs. The shortest flexible linker, a single Gly, disrupted phosphatase activity with and without rapamycin for both PTP1B and PTP-PEST (Fig. 10, B–E). Increasing the length of the flexible linker to GPG resulted in improved dynamic range of regulation for both PTP1B and PTP-PEST (Fig. 10, B–E). In both PTPs, this linker promoted disruption of phosphatase activity in the absence of rapamycin and recovery of activity with the addition of rapamycin. RapR-PTP1B and RapR-PTP-PEST with the longest linker showed dramatically reduced dynamic range of regulation (Fig. 10, B–E). These results demonstrate the broad applicability of this approach in regulation of protein tyrosine phosphatases as well as the tunability of this regulation through the modification of the flexible linkers.

We developed a temporally regulated tool to study Shp2 signaling on a physiologically relevant time scale. The RapR-Shp2 tool utilizes a highly dynamic allosteric switch in the Shp2 catalytic domain to disrupt catalytic activity and recover activity following the addition of FRB and rapamycin (Fig. 3, B and C). Furthermore, we validated that insertion of the iFKBP domain does not disrupt canonical functions of Shp2. RapR-Shp2 induced ERK activity, recognized and dephosphorylated established Shp2 substrates, and formed previously described protein-protein interactions mediated by SH2 domains (Fig. 4). We also demonstrated the general applicability of this approach to other PTPs (Fig. 10, B and C). Other tools developed to study phosphatases, such as an optogenetically controlled PTP1B developed by Hongdusit et al. (2020), possess many of the same advantages of our tools but lack facile implementation in a variety of PTPs and the capability to reconstruct individual downstream pathways (Hongdusit et al., 2020). We also described an approach for optimization of the tool by varying insertion site flexibility, thus providing a template for the future design of allosterically regulated tools.

The importance of Shp2 in cell migration, polarization, and protrusions has been repeatedly established (Cao et al., 2019; Grosskopf et al., 2015; Hartman et al., 2013; Hu et al., 2014; Lee et al., 2013; Sausgruber et al., 2015; Zhang et al., 2015). However, due to the limitations of the tools, the timing, extent, and sequence of events specifically induced by Shp2 activation has not been fully determined. We overcome the limitations of conventional molecular biology approaches to demonstrate that acute Shp2 activation can recapitulate known Shp2 signaling and morphological changes with added temporal and spatial resolution. Using RapR-Shp2, we provided direct evidence that Shp2 activity induces cell spreading, protrusions, and migration even in the absence of parallel signaling pathways induced by growth factor stimulation (Fig. 5, A–E). These dynamic behaviors showed dependency on the duration of Shp2 signaling, with an initial increase in cell spreading and protrusions, delayed initiation of retraction, and transient induction of migration. These transient events could not be detected using CA-Shp2 in HeLa cells. Our results show the importance of a temporally regulated tool to study the intricacies of Shp2 signaling on a physiologically relevant time scale.

Previous reports suggested that increased Shp2 activity leading to hyperphosphorylation of ERK contributes to migratory phenotypes, and that ERK activity stabilizes protrusions at the leading edge of a cell (Mendoza et al., 2015; Mizutani et al., 2007). We found that in the absence of MEK/ERK signaling, Shp2 still induces moderate cell migration but not sustained cell spreading (Fig. 6, A–C). Using specific SH2 domain mutants to selectively block phospho-tyrosine binding by each SH2 domain, we observed that protein-protein interactions mediated by the N-SH2 domain of Shp2 are sufficient to induce ERK-independent cell migration (Fig. 7 D). Further, our data suggest that this ERK-independent migration is mediated through Shp2-regulated ROCK II activity (Fig. 7 D), revealing multiple Shp2-dependent pathways for the induction of cell migration.

Unexpectedly, Shp2 regulation of cell retraction was unaffected by MAPK or ROCK II inhibition (Fig. 6 D and Fig. 7 E). Only the loss of the N-SH2 domain binding resulted in a reduction of cell retraction (Fig. 7 E). This suggests an additional Shp2 signaling pathway responsible for driving cell contraction, likely mediated through interaction with the N-SH2 domain. The specific and temporal regulatory capabilities of the RapR-Shp2 tool provide a unique opportunity to further dissect these pathways to better understand Shp2-mediated morphological changes.

In addition, we used RapRTAP to probe the importance of Shp2 signaling within specific protein complexes. We identified morphological changes caused by different Shp2 signaling complexes: those mediated by Gab1 or Gab2 scaffolding proteins and focal adhesion protein complexes. Targeted activation of RapR-Shp2 in complex with Gab1/2 demonstrated transient cell migration with robust protrusive and retractive activity (Fig. 8, D and E). This morphology may potentially be mediated by Shp2-Gab-dependent activation of Src family kinases (Ren et al., 2004). Activation of this pathway was effective at stimulating necessary pathways for migration, but a loss of direct Shp2 activity in focal adhesions likely prevented the formation of nascent focal adhesions and local signaling. Hence, Gab targeting only partially reconstructed Shp2-mediated morphological changes.

In contrast, targeted activation of Shp2 within focal adhesions induced transient spreading and protrusions in an ERK-dependent manner (Fig 8, B–D and Fig. S5 B). While we demonstrated the necessity of ERK activity in this transient spreading event, it remains to be seen if this is due to local pools of active ERK within focal adhesions. Notably, R32L RapR-Shp2 activation within focal adhesions is incapable of inducing cell retraction or migration. This suggests a possible loss of signaling required for migration which could be, in part, due to an inability to induce polarized contractility mediated by ROCK II. Shp2-mediated migration and protrusions can be reconstituted, potentially through ROCK II, in combination with cell spreading by targeting R138L RapR-Shp2 to focal adhesions (Fig. 9). Application of RapRTAP-Shp2 tool will enable the identification of specific players mediating these alternative pathways in future studies.

In summary, we presented a protein engineering tool that enables allosteric regulation of protein tyrosine phosphatases. This method provides high specificity and tight temporal control of phosphatase activation. We demonstrated that this approach allows us to mimic Shp2 signaling in living cells revealing transient effects of Shp2 activation and parallel signaling pathways mediating cell spreading and migration. Targeting activation of the engineered phosphatase to a specific protein complex allowed us to reconstruct individual signaling pathways downstream of Shp2 and determine their role in regulation of cell morphodynamics. Thus, this approach will provide a powerful tool for dissection of physiological and pathological functions of tyrosine phosphatases.

The following antibodies were used: anti-FLAG (cat. no. F3165; Millipore-Sigma, mouse), anti-FLAG (cat. no. MAB8529; Cell Signaling, mouse), anti-mCherry (cat. no. M11217; Invitrogen, rabbit), anti-GFP (cat. no. 632381; Clontech, mouse), anti-Myc (cat. no. 05–724; Millipore-Sigma, mouse), anti-paxillin (cat. no. BDB612405; Thermo Fisher Scientific, mouse), anti-phospho-paxillin (Tyr¹¹⁸; cat. no. 44–722G; Invitrogen, mouse), anti-p44/42 (ERK1/2; cat. no. 9102; Cell Signaling, rabbit), anti-p44/42 (ERK1/2; Thr²⁰²/Tyr²⁰⁴; cat. no. 9101; Cell Signaling, rabbit), anti-EGFR (cat. no. 2232; Cell Signaling, rabbit), anti-phospho-EGFR (Tyr⁹⁹²; cat. no. 2235; Cell Signaling, rabbit), anti-FAK (cat. no. 06-543; Millipore-Sigma, mouse), anti-phospho-FAK (Tyr^576/577; cat. no. 3281; Cell Signaling, rabbit), anti-PLCγ (cat. no. 5690; Cell Signaling, rabbit), and anti-phospho-PLCγ (Tyr⁷⁸³; cat. no. 14008; Cell Signaling, rabbit), anti-Gab1 (cat. no. 3232; Cell Signaling, rabbit), anti-phospho-Gab1 (Tyr⁶²⁷; cat. no. 3231; Cell Signaling, rabbit), anti-phospho-Gab1 (Tyr⁶⁵⁹; cat. no. 12745; Cell Signaling, rabbit), anti-ROCK II (cat. no. 8236; Cell Signaling, rabbit), anti-phospho-ROCK II (Tyr⁷²²; cat. no. SAB4301564; Sigma-Aldrich, rabbit), anti-phospho-MLC 2 (Ser¹⁹/Thr¹⁸; cat. no. 3674; Cell Signaling, rabbit), and anti-Shp2 (cat. no. 3397; Cell Signaling, rabbit).

The following cell lines were used: HeLa cells (cat. no. CCL-2; ATCC), human embryonic kidney HEK293T cells (cat. no. CRL-3216; ATCC), HEK293FT (cat. no. R70007; Thermo Fisher Scientific), and A431 epidermoid carcinoma cells (cat. no. CRL-1555; ATCC). All cell lines were cultured at 37°C and 5% carbon dioxide in DMEM medium supplemented with 10% vol/vol fetal bovine serum and 1% vol/vol GlutaMAX. All experiments were performed with cells grown for <20 passages after thawing. All cells were tested negative for Mycoplasma contamination. Cell line identity was confirmed by the supplier using STR analysis. DH5α bacteria cells (cat. no. C2987H; NEB) were used for GST-paxillinN-C3 production.

The following reagents were used: IgG-coupled agarose beads (cat. no. IP04-1.5ML; Millipore-Sigma), Leupeptin hemisulfate (cat. no. L-010–5; Gold Biotechnology), Aprotinin (cat. no. A-655–25; Gold Biotechnology), Y-27632 dihydrochloride (cat. no. Y0503; Millipore-Sigma), 2X Laemmli Sample buffer (cat. no. 161–0737; Bio-Rad), 2-Mercaptoethanol (cat. no. 60-24-2; Fisher Chemical), trypsin (cat. no. V5113; Promega), Fetal Bovine Serum (cat. no. FB-01; Omega Scientific), and CellMask Deep Red Plasma Membrane Stain (cat. no. C10046; Thermo Fisher Scientific), GlutaMAX (cat. no. 35050061; Thermo Fisher Scientific). Coverslips for live imaging were purchased from Thermo Fisher Scientific (cat. no. 25CIR-1.5).

Dominant Negative Shp2-flag and Constitutively Active Shp2-flag plasmids were a gift from the Graeme Carnegie lab (UIC). PTP1B and PTP-PEST were a gift from Nicholas Tonks (Cold Spring Harbor Laboratory). iFKBP was inserted at the designated insertion sites in CA-Shp2, WT-PTP1B, and PTP-PEST using a modified site-directed mutagenesis method as described by Karginov et al. (2010). An mVenus or Cerulean fluorescent tag was inserted at the C-terminus of Shp2. R32L and R138L point mutations were introduced using site-directed mutagenesis. FRB-GFP-Gab1 and FRB-GFP-Gab2 were developed from Gab1 and Gab2 plasmids which were a gift from Benjamin Neel (NYU). FRB was inserted in pGFP-c1 vector using the modified site-directed mutagenesis method with using an FRB megaprimer. Gab1 and Gab2 genes were amplified using primers generating XhoI and BamHI restriction sites at the ends of the genes. These restriction sites were used to insert Gab1 or Gab2 into the FRB-GFP vector. Y/F mutations were then introduced within both of these constructs at Tyr⁶²⁸/Tyr⁶⁶⁰ in Gab1 and Tyr⁶⁰⁴/Tyr⁶³⁴ in Gab2 using site-directed mutagenesis. To generate FAK (Y397) FRB-mVenus, FRB was inserted in FAK-mVenus using modified site-directed mutagenesis, replacing the Tyr³⁹⁷ site with FRB. Venus-ipep-FRB and mCherry-ipep-FRB and adenoviral mCherry-FRB as described by Klomp et al. (2019) were used. LeGO-iV2 was a gift from Boris Fehse (plasmid # 27344; Addgene; http://n2t.net/addgene:27344; RRID:Addgene_27344; Weber et al., 2008), pMD2.G was a gift from Didier Trono (plasmid # 12259; Addgene; http://n2t.net/addgene:12259; RRID:Addgene_12259), psPAX2 was a gift from Didier Trono (plasmid # 12260; Addgene; http://n2t.net/addgene:12260; RRID:Addgene_12260), GST-paxillinN-C3 construct was a gift from Michael Schaller (Lyons et al., 2001). In order to prepare lentiviral RapR-Shp2, RapR-Shp2-mVenus-flag was PCR amplified with primers to introduce NotI and BsiWI on the 5′- and -3′ ends. The PCR product, along with Lego-iV2 plasmid, was digested with NotI and BsrGI then ligated with generate lentiviral Lego-iV2 RapR-Shp2-mVenus-flag.

5′-CTA​TAC​GCT​AAG​AGA​ACT​TAA​ACT​TTC​AAA​GGG​TAC​CTG​CGT​GGT​GCA​CTA​CAC​CGG​GAT​GCT​TGA​AGA​TGG​AAA​GAA​ATT​TGA​TTC​CTC​CCG​GGA​CAG​AAA​CAA​GCC​CTT​TAA​GTT​TAT​GCT​AGG​CAA​GCA​GGA​GGT​GAT​CCG​AGG​CTG​GGA​AGA​AGG​GGT​TGC​CCA​GAT​GAG​TGT​GGG​TCA​GAG​AGC​CAA​ACT​GAC​TAT​ATC​TCC​AGA​TTA​TGC​CTA​TGG​TGC​CAC​TGG​GCA​CCC​AGG​CAT​CAT​CCC​ACC​ACA​TGC​CAC​TCT​CGT​CTT​CGA​TGT​GGA​GCT​TCT​AAA​ACT​GGA​AGG​ACA​AGG​AAT​ACG​GAG​AGA​ACG​GTC​TGG​C-3′.

Forward: 5′-TTT​ACG​AGA​GCT​CAA​ACT​CTC​TAA​AGT​CGG​CTC​TGG​TGG​TAC​CTG​CGT​GGT​GCA​CTA​CAC-3′.

Reverse: 3′-GAC​GGT​TCT​CTC​TGT​GTT​TCC​CTG​GCC​ACC​AGA​GCC​CTC​CAG​TTT​TAG​AAG​CTC​CAC​AT-5′.

Forward: 5′-TAC​ACT​TTA​CGA​GAG​CTC​AAA​CTC​TCA​GGA​AGC​GGT​ACC​TGC​GTG​GTG​CAC​TAC​ACC​G-3′.

Reverse: 3′-CTG​CCA​GAC​GGT​TCT​CTC​TGT​GTT​CCC​CGA​ACC​GCC​TTC​CAG​TTT​TAG​AAG​CTC​CAC-5′.

Forward: 5′-CCG​CCC​ATG​ACT​ACA​CTT​TAC​GAG​AG-3′.

Reverse: 3′-GAA​AGT​GGT​ACT​GCC​AGA​CGG​TTC​TCT-5′.

Forward: 5′-GCC​CAT​GAC​TAC​ACT​TTA​CGA​GAG​CTC​AAA​CTC​TCA​GGT​ACC​TGC​GTG​GTG​CAC​TAC​ACC-3′.

Reverse: 3′-AAA​GTG​GTA​CTG​CCA​GAC​GCT​TCT​CTC​TGT​GTT​ACC​GCC​TTC​CAG​TTT​TAG​AAG​CTC​CAC-5′.

5′-GCA​AAT​ATC​ATC​ATG​CCT​GAA​TTT​GAA​ACC​GGT​ACC​TGC​GTG​GTG​CAC​TAC​ACC​GGG​ATG​CTT​GAA​GAT​GGA​AAG​AAA​TTT​GAT​TCC​TCC​CGG​GAC​AGA​AAC​AAG​CCC​TTT​AAG​TTT​ATG​CTA​GGC​AAG​CAG​GAG​GTG​ATC​CGA​GGC​TGG​GAA​GAA​GGG​GTT​GCC​CAG​ATG​AGT​GTG​GGT​CAG​AGA​GCC​AAA​CTG​ACT​ATA​TCT​CCA​GAT​TAT​GCC​TAT​GGT​GCC​ACT​GGG​CAC​CCA​GGC​ATC​ATC​CCA​CCA​CAT​GCC​ACT​CTC​GTC​TTC​GAT​GTG​GAG​CTT​CTA​AAA​CTG​GAA​TCA​AAG​CCC​AAA​AGA​GTT​ACA​TTG​CCA​CAC-3′.

5′-TAG​ATA​TAA​AAA​CAT​CCT​GCC​CTT​TGA​TCA​TGG​TAC​CTG​CGT​GGT​GCA​CTA​CAC​CGG​GAT​GCT​TGA​AGA​TGG​AAA​GAA​ATT​TGA​TTC​CTC​CCG​GGA​CAG​AAA​CAA​GCC​CTT​TAA​GTT​TAT​GCT​AGG​CAA​GCA​GGA​GGT​GAT​CCG​AGG​CTG​GGA​AGA​AGG​GGT​TGC​CCA​GAT​GAG​TGT​GGG​TCA​GAG​AGC​CAA​ACT​GAC​TAT​ATC​TCC​AGA​TTA​TGC​CTA​TGG​TGC​CAC​TGG​GCA​CCC​AGG​CAT​CAT​CCC​ACC​ACA​TGC​CAC​TCT​CGT​CTT​CGA​TGT​GGA​GCT​TCT​AAA​ACT​GGA​AAG​GGT​TGT​CCT​ACA​CGA​TGG​TGA​TCC​CAA​TGA​G-3.′

5′-CAG​TGC​GAC​AGC​TAG​AAT​TGG​AAA​ACG​GTA​CCT​GCG​TGG​TGC​ACT​ACA​CCG​GGA​TGC​TTG​AAG​ATG​GAA​AGA​AAT​TTG​ATT​CCT​CCC​GGG​ACA​GAA​ACA​AGC​CCT​TTA​AGT​TTA​TGC​TAG​GCA​AGC​AGG​AGG​TGA​TCC​GAG​GCT​GGG​AAG​AAG​GGG​TTG​CCC​AGA​TGA​GTG​TGG​GTC​AGA​GAG​CCA​AAC​TGA​CTA​TAT​CTC​CAG​ATT​ATG​CCT​ATG​GTG​CCA​CTG​GGC​ACC​CAG​GCA​TCA​TCC​CAC​CAC​ATG​CCA​CTC​TCG​TCT​TCG​ATG​TGG​AGC​TTC​TAA​AAC​TGG​AAG​GTC​AAA​GGA​GTT​ACA​TTC​TTA​CCC​AGG​GC-3′.

5′-GAC​TAC​TTC​ATC​AGG​ACA​CTC​TTA​CTT​GAA​TTC​GGA​CCA​GGT​ACC​TGC​GTG​GTG​CAC​TAC​ACC​GGG​ATG​CTT​GAA​GAT​GGA​AAG​AAA​TTT​GAT​TCC​TCC​CGG​GAC​AGA​AAC​AAG​CCC​TTT​AAG​TTT​ATG​CTA​GGC​AAG​CAG​GAG​GTG​ATC​CGA​GGC​TGG​GAA​GAA​GGG​GTT​GCC​CAG​ATG​AGT​GTG​GGT​CAG​AGA​GCC​AAA​CTG​ACT​ATA​TCT​CCA​GAT​TAT​GCC​TAT​GGT​GCC​ACT​GGG​CAC​CCA​GGC​ATC​ATC​CCA​CCA​CAT​GCC​ACT​CTC​GTC​TTC​GAT​GTG​GAG​CTT​CTA​AAA​CTG​GAA​GGT​CCC​GGA​CAA​AAT​GAA​TCT​CGT​AGG​CTG​TAT​CAG​TTT​CAT​TA-3′.

Forward: 5′-GAG​GAG​TCG​ATG​GCA​GTT​TTT​TAG​CAC-3′.

Reverse: 3′-GAA​GTC​TCC​AGG​GTT​ACT​CTT​ACT​GGG-5′.

Forward: 5′-GGC​AAG​CAT​GGC​AGC​TTC​CTC​GTT​CTA-5′.

Reverse: 3′-GAG​AAC​GAA​GTC​TCC​GGG​GTG​GCT​CT-3′.

5′-CGA​GCG​GCC​GCC​ACT​GTG​CTG​GAT​CAT​GGC​TTC​TAG​AAT​CCT​CTG​GCA​TGA​GAT​GTG​GCA​TGA​AGG​CCT​GGA​AGA​GGC​ATC​TCG​TTT​GTA​CTT​TGG​GGA​AAG​GAA​CGT​GAA​AGG​CAT​GTT​TGA​GGT​GCT​GGA​GCC​CTT​GCA​TGC​TAT​GAT​GGA​ACG​GGG​CCC​CCA​GAC​TCT​GAA​GGA​AAC​ATC​CTT​TAA​TCA​GGC​CTA​TGG​TCG​AGA​TTT​AAT​GGA​GGC​CCA​AGA​GTG​GTG​CAG​GAA​GTA​CAT​GAA​ATC​AGG​GAA​TGT​CAA​GGA​CCT​CCT​CCA​AGC​CTG​GGA​CCT​CTA​TTA​TCA​TGT​GTT​CCG​ACG​AAT​CTC​AAA​GAC​TAG​TGG​TGG​CAG​TGG​AAC​CAG​CAA​GGG​CGA​GGA​GCT​GTT​CAC​CG-3′.

Forward: 5′-GAT​ACT​CTC​GAG​GTA​TGA​GCG​GCG​GCG​AAG​TGG​TTT​G-3′.

Reverse: 3′-GAC​ACC​CAC​CAA​GAA​TGT​GAA​GTG​AGG​ATC​CCA​TAA​G-5′.

Forward: 5′-GAA​TAA​ACC​CAA​AGG​AGA​CAA​ACA​AGT​CGA​ATT​CCT​GGA​TTT​AGA​CCT​AGA​TTC​TGG​GA-3′.

Reverse: 3′-TCC​CAG​AAT​CTA​GGT​CTA​AAT​CCA​GGA​ATT​CGA​CTT​GTT​TGT​CTC​CTT​TGG​GT-5′.

Forward: 5′-GCA​GCA​TGG​CAG​ACG​AGA​GGG​TGG​ATT​TCG​TTG​TGG​TGG​ACC​AAC​AGA​AGA​CTC-3′.

Reverse: 3′-GAG​TCT​TCT​GTT​GGT​CCA​CCA​CAA​CGA​AAT​CCA​CCC​TCT​CGT​CTG​CCA​TGC​TGC-5′.

Forward: 5′-GAT​ACT​CTC​GAG​GTA​TGA​GCG​GCG​GCG​GCG​GCG​ACG-3′.

Reverse: 3′-GAA​CCT​TCC​AAG​GGT​GCC​AAG​CTG​TAA​GGA​TCC​CAC​GAC-5′.

Forward: 5′-CAA​AGA​AGA​GTA​CTG​GCA​GTG​TGG​ATT​TTC​TCG​CCC​TGG​ACT​TCC​AGC​CGG​GCT​C-3′.

Reverse: 3′-GAG​CCC​GGC​TGG​AAG​TCC​AGG​GCG​AGA​AAA​TCC​ACA​CTG​CCA​GTA​CTC​TTC​TTT​G-5′.

Forward: 5′-CTG​TCA​CAT​CAG​ATG​AGA​AGG​TAG​ACT​TTG​TCC​AAG​TGG​ATA​AAG​AGA​AGA​CCC-3′.

Reverse: 3′-GGG​TCT​TCT​CTT​TAT​CCA​CTT​GGA​CAA​AGT​CTA​CCT​TCT​CAT​CTG​ATG​TGA​CAG-5′.

5′-CAT​GCA​GTC​TCT​GTG​TCA​GAG​ACA​GAT​GAC​GGC​CCC​GGA​TGG​CAT​GAG​ATG​TGG​CAT​GAA​GGC​CTG​GAA​GAG​GCA​TCT​CGT​TTG​TAC​TTT​GGG​GAA​AGG​AAC​GTG​AAA​GGC​ATG​TTT​GAG​GTG​CTG​GAG​CCC​TTG​CAT​GCT​ATG​ATG​GAA​CGG​GGC​CCC​CAG​ACT​CTG​AAG​GAA​ACA​TCC​TTT​AAT​CAG​GCC​TAT​GGT​CGA​GAT​TTA​ATG​GAG​GCC​CAA​GAG​TGG​TGC​AGG​AAG​TAC​ATG​AAA​TCA​GGG​AAT​GTC​AAG​GAC​CTC​CTC​CAA​GCC​TGG​GAC​CTC​TAT​TAT​CAT​GTG​TTC​CGA​CGA​ATC​TCA​AAG​ACT​AGT​GGA​CCC​GGT​GCA​GAG​ATC​ATC​GAT​GAG​GAA​GAC​ACA​TAC-3′.

Forward: 5′-ATC​GTA​CTG​CGG​CCG​CAT​GAC​ATC​GCG​GAG​ATG​GTT​TCA​CC-3′.

Reverse: 3′-TGC​TAC​GAT​CGT​ACG​TCA​CTT​GTC​GTC​ATC​GTC​CTT​GTA​GT-5′.

For expression of the RapR phosphatase constructs and FRB constructs, HEK293T cells were grown to 70% confluency on a plastic dish coated with 1:200 poly-L-lysine and were transfected with Lipofectamine 2000. A transfection solution of 500 μl of Opti-MEM, 4 μl Lipofectamine 2000, and 2 µg of DNA was added to the cells and incubated overnight. Cells were stimulated with 1 µM Rapamycin in ethanol for 1 h or listed time points for the time course of ERK activation assay. Control cells were treated with an equivalent volume of ethanol. Cells were then lysed in 2X Laemmli sample buffer with 5% 2-mercaptoethanol, and proteins were resolved via sodium dodecyl sulfate-polyacrylimide gel electrophoresis (SDS/PAGE) and immunoblotted. Blots were developed using enhanced chemiluminescence.

An N-terminal fragment of paxillin was purified using a previously described procedure (Lyons et al., 2001). The GST-paxilliN-C3 construct was expressed in DH5α E. coli cells following inductions with 0.1 mM Isopropyl β-D-1-thiogalactopyranoside for 4 h. The bacterial pellet was resuspended in 30 ml of TETN buffer (20 mM TRIS pH 8, 10 mM NaCl, 1 mM EDTA, 0.5% [vol/vol] Triton X-100) then lysed by sonication. Using affinity chromatography with Glutathione Sepharose, GST-tagged paxillin was purified from the cleared lysates.

The purified GST-paxillin was then phosphorylated using an immunoprecipitated constitutively active c-Src. HEK239FT cells 70% confluent in a 10-cm dish were transfected with c-Src (Y527F)-mCherry. Cells were lysed with lysis buffer (20 mM HEPES-KOH, pH 7.8, 50 mM KCl, 1 mM EGTA, 1% NP-40, aprotinin 16 µg/ml, and Leupeptin hemisulfate 3.2 µg/ml). Lysates were cleared through centrifugation at 2683 RCF for 10 min at 4°C. Protein G Agarose beads were preincubated with anti-myc antibody (4A6 from Millipore-Sigma) for 1 h at 4°C. Cleared lysates were incubated with anti-myc conjugated beads for 1.5 h at 4°C. Beads were then washed twice with wash buffer (20 mM Hepes-KOH, pH 7.8, 100 mM NaCl, 50 mM KCl, 1 mM EGTA, 1% NP-40) and then with kinase reaction buffer (25 mM HEPES, pH 7.5, 5 M MgCl₂, 0.5 mM EGTA, 0.005% BRIJ-35). The beads were resuspended in kinase reaction buffer and incubated with 0.1 mM ATP and 0.05 mg/ml purified N-terminal fragment of paxillin (GST-paxillinN-C3) at 37°C for 10 min. The reaction buffer containing the phosphorylated paxillin fragment was then separated from the Agarose beads conjugated with c-Src(Y527F)-mCherry via centrifugation. Phosphorylation of paxillin was verified with SDS/PAGE and immunoblotting for paxillin phosphorylation.

HEK293T cells coexpressing the RapR Phosphatase constructs and mVenus-ipep-FRB or mCherry-ipep-FRB (Klomp et al., 2019) were treated with 1 µM Rapamycin for 1 h prior to lysis (Lyons et al., 2001). Cells were lysed in a HEPES sample buffer (20 mM HEPES-KOH pH 7.8, 50 mM KCl, 1 mM EGTA, and 1% NP-40 [vol/vol]). Lysates were cleared via centrifugation at 2683 g RCF for 10 min at 4°C. Protein G Agarose beads were preincubated with anti-FLAG or anti-GFP antibody (F3165; Sigma-Aldrich) for 1 h in lysis buffer. Lysates were incubated with anti-FLAG or anti-GFP conjugated beads at 4°C for 1.5 h. Following incubation, beads were washed twice with 1 ml of lysis buffer and twice with 1 ml of wash buffer (20 mM HEPES-KOH, pH 7.8, 100 mM NaCl, 50 mM KCl, 1 mM EGTA, 1% NP-40 [vol/vol]). Beads were resuspended in 40 μl Imidazole reaction buffer (25 mM Imidazole pH 7.2, 2.5 mM EDTA, 50 mM NaCl, 5 mM DTT) containing purified phosphorylated N-terminal fragment of paxillin. The immunoprecipitated RapR phosphatases were then incubated with purified phosphorylated paxillin for 30 min, or for the specified times in the time course of CA-Shp2 and RapR-Shp2 phosphatase activity, at 32°C. The reaction was quenched with the addition of 2X Laemmli sample buffer with 5% (vol/vol) 2-mercaptoethanol and boiling at 100°C for 5 min. Proteins were resolved via SDS/PAGE and immunoblotted.

Relative kinetic values, K_m and V_max, were determined using the approach described by Orsi and Tipton (1979) for substrate depletion over time. Briefly, plotting ([Substrate₀]−[Substrate_t])/time vs. (ln([Substrate_t]/[Substrate₀])/t), we generated a linear graph with a slope of −K_m and a y-intercept of V_max. Using immunoblot intensity, we established the zero time point for both CA-Shp2 and RapR-Shp2 as 1.0 and evaluated the disappearance of substrate through the change in relative blot intensity. This value was used as a proxy for substrate concentration to evaluate the relative kinetic values for CA-Shp2 and RapR-Shp2 and the differences in K_m and V_max. CA-Shp2 relative kinetic values: K_m = 0.85 R.I. and V_max = −0.012 R.I./min. RapR-Shp2 relative kinetic values: K_m = 1.04 R.I. and V_max = −0.026 R.I./min. Thus, RapR-Shp2 shows 1.2-fold increase in K_m and 2.2-fold increase in V_max over corresponding values for CA-Shp2.

Lego-iV2-RapR-Shp2-mVenus-flag was cotransfected with lentiviral envelope protein VSV-G in pMD2.G plasmid and a second-generation lentiviral packaging plasmid, psPAX2 in HEK293T cells. Conditioned media were collected 1–3 d after transfection, and the virus was concentrated using Lenti-X concentrator following the manufacturer’s protocol. A431 cells were then grown to 80% confluency and transduced using the concentrated virus.

A431 cells were transduced with lentiviral RapR-Shp2-mVenus-flag and mCherry-FRB adenovirus. Confluent stably expressing A431 cells were serum starved for 4 h prior to stimulation with 1 µM rapamycin for 2 and 4 h. Samples without rapamycin were treated with an equivalent volume of ethanol for 4 h. Nonexpressing A431 cells were treated either with rapamycin or with an equivalent volume of ethanol as controls.

A431 cells were transduced with lentiviral RapR-Shp2-mVenus and mCherry-FRB adenovirus, lentiviral CA-Shp2-mVenus, or mCherry-FRB adenovirus. Confluent, stably expressing, A431 cells were activated with 1µM rapamycin for 2 h. Protein G Plus Agarose (IP04; Sigma) beads were preincubated with anti-GFP antibody for 1 h in lysis buffer. Samples were lysed with HEPES sample buffer (20 mM HEPES-KOH pH 7.8, 50 mM KCl, 1 mM EGTA, and 1% NP-40 [vol/vol]) and centrifuged at 12,000 rpm for 10 min. Lysate was incubated with anti-GFP conjugated Protein G Plus beads for 1.5 h at 4°C. Following incubation, beads were washed twice with 1 ml of lysis buffer and twice with 1 ml of wash buffer (20 mM HEPES-KOH, pH 7.8, 250 mM NaCl, 50 mM KCl, 1 mM EGTA, 1% NP-40 [vol/vol]). Proteins were eluted with Laemmli sample buffer and resolved via SDS/PAGE and immunoblotting.

HeLa cells were grown to 60% confluency in a 35-mm dish and transfected with 6 μl of Fugene6 and 1 µg of total DNA and incubated overnight. These cells were seeded at 30% confluency on fibronectin-coated (5 mg/l) glass coverslips for 4 h prior to imaging and serum starved in Leibovitz L-15 imaging media for 2 h prior to imaging. All cells were stained with CellMask Deep Red plasma membrane stain (cat. no. C10046; Thermo Fisher Scientific) prior to imaging following manufacturer’s recommendation. Using epifluorescence imaging, we selected cells coexpressing all necessary constructs. Cells were imaged live at 37°C in an open heated chamber (Werner Instruments) using Olympus UPlanSAPO 40× (oil, N.A. 1.25) objective on Olympus IX-83 microscope controlled by Metamorph software and equipped with Xcite 120 LED (Lumen Dynamics) light source and Image EMX2 CCD camera (Hamamatsu). Time lapse images were acquired every 2 min over the duration of the imaging experiment, 4–4.5 h. In the experiments using Trametinib, cells were treated with 10 nM Trametinib for 1 h prior to imaging. HeLa cells used for the ROCK II inhibition studies were treated with 10 µM Y-27632 dichloride (cat. no. Y0503; Millipore-Sigma) for 1 h prior to imaging.

Epifluorescence images of CellMask Deep Red were used to analyze changes in cell morphology. The binary mask of the images was created using MovThresh 2014 (Tsygankov et al., 2014) software package in Matlab (version 2017a). Cell area, in pixels, was calculated in ImageJ for each timepoint. In order to determine the area change for each cell, the area at each time point was normalized to the average area of the cell in the time prior to stimulation with rapamycin. The average of these normalized values and 90% confidence interval were then calculated for all cells for each condition.

To analyze the changes in protrusive activity of the cells, the binary masks described above were analyzed using ProActive2014 script in Matlab (version 2017a; Tsygankov et al., 2014). Total cell area was used to normalize the area gained and lost, with a lag of one frame. The protrusive and retractive area for each time point was divided by the average of the protrusive or retractive area prior to rapamycin addition. The average and 90% confidence interval for the relative protrusive and retractive area were again calculated for all cells in each condition.

Analysis of cell migration was determined using ImageJ software. The centroid of the cell was determined for each time point. The distance traveled over each 2-min time point was calculated and the total path length was determined for 1-h intervals.

Plots for all imaging analyses were generated in OriginPro 2019.

In silico analysis of RapR-Shp2 was performed using NAMD v2.13 software (Phillips et al., 2020). Models of RapR-Shp2 were created using PDB accession no. 4DGP for Shp2 catalytic domain and PDB accession no. 4FAP for the FKBP-Rapamycin-FRB complex. The unresolved loops within the catalytic domain of Shp2 (residues 295–301 and 314–323) were created and refined using Modeller (Šali and Blundell, 1993). Insertion of iFKBP in Shp2 to generate these models was performed as previously described (Karginov et al., 2010). All simulations were conducted in explicit solvent at 310 K. Models were minimized and equilibrated prior to conducting a 50-ns MD simulation. All models were simulated in three independent runs with slightly different starting structures to increase the conformational space sampled. The first 10 ns of each run were discarded for analysis.

All in vitro data, imaging data, and Western blots are representative of at least three independent experiments. Western blots were quantified using ImageJ software. Error bars in plots are representative of a confidence interval of 90%. Box and whisker plots: Box indicates first through third quartile with median line, whiskers are 10–90%, open square represents mean. Significance was determined using ANOVA with appropriate post hoc analysis or two-tailed Student’s t test. Data distribution was assumed to be normal but this was not formally tested.

All in vitro phosphatase assays were analyzed using two-tailed Student’s t test. This test was selected to enable pairwise comparisons of each modification with and without rapamycin to evaluate statistically significant induction of phosphatase activity by each individual construct.

One-way ANOVA with Holm-Bonferroni post hoc analysis was used for comparison of multiple related means. This was applied to evaluate the influence of the RapR insertion on activation of ERK relative to a catalytically dead and constitutively active Shp2 or evaluate inhibitor effects on ERK or MLC phosphorylation.

Time courses an imaging data were evaluated using one-way ANOVA with repeated measures and post hoc Dunnett test to compare all time points to the zero point. Pairwise comparisons of multiple conditions at different time points, as with MAPK inhibition or comparison to controls, were performed using two-way ANOVA with repeated measures with post hoc Holm-Bonferroni pairwise comparisons between factors.

Fig. S1 provides quantitative analysis of Western blots from Fig. 3 as well as shows relative expression of RapR-Shp2 to endogenous Shp2. Fig. S2 analyzes the protrusive and retraction activity of RapR-Shp2 expressing cells and the effects of CA-Shp2 on cell morphology. Video 1 shows the cell mask movie of RapR-Shp2 expressing HeLa cells before and after activation. Fig. S3 analyzes the influence of Shp2 SH2 domains on ERK activation and protrusive and retractive area as well as the effects of ROCK II and ERK inhibitors on SH2 domain-mediated cell morphology. Fig. S4 shows the effects of RapRTAP Shp2 signaling complexes on cell protrusion and retraction as well as construct localization. Video 2 shows the cell mask movie of R32L RapR-Shp2 and FRB-GFP-Gab1 (Y628F/Y660F) expressing HeLa cells before and after activation. Video 3 shows the cell mask movie of R32L RapR-Shp2 and FRB-GFP-Gab2 (Y604F/Y634F) expressing HeLa cells before and after activation. Video 4 shows the cell mask movie of R32L RapR-Shp2 and FAK(397)FRB-mVenus expressing HeLa cells before and after activation. Fig. S5 demonstrates the effect of Shp2 activity within focal adhesions on ERK activity and cell spreading. Video 5 shows the cell mask movie of R138L RapR-Shp2 and FAK(397)FRB-mVenus expressing HeLa cells before and after activation.

The authors would like to thank Dr. Martin Brennan and Dr. Mark Shaaya for technical assistance and discussion and Dr. Benjamin Neel for discussion and advice.

National Institute of Health-NIGMS grant R01GM118582 (A.V. Karginov), National Institute of Health-NCI grant R21CA212907 (A.V. Karginov), National Institute of Health T32 Training grant HL007829-22 (J. Fauser), US Department of Energy, Office of Science DE-AC02-06CH11357 (Y. Alexeev).

The authors declare no competing financial interests.

Author contributions: Investigation: J. Fauser, V. Huyot, J. Matsche, P. Kota. Validation: J. Fauser, V. Huyot, J. Matsche, B.N. Szynal. Methodology: J. Fauser, V. Huyot, P. Kota. Formal Analysis: J. Fauser, P. Kota. Visualization: J. Fauser. Funding acquisition: A.V. Karginov. Resources: Y. Alexeev. Supervision: A.V. Karginov. Conceptualization: A.V. Karginov. Project Administration: A.V. Karginov. Writing—original draft: J. Fauser. Writing—review and editing: J. Fauser, A.V. Karginov, P. Kota, B.N. Szynal, Y. Alexeev.