Allosteric modulation of the catalytic VYD loop in Slingshot by its N-terminal domain underlies both Slingshot auto-inhibition and activation

Slingshots are phosphatases that modulate cytoskeleton dynamics, and their activities are tightly regulated in different physiological contexts. Recently, abnormally elevated Slingshot activity has been implicated in many human diseases, such as cancer, Alzheimer's disease, and vascular diseases. Therefore, Slingshot-specific inhibitors have therapeutic potential. However, an enzymological understanding of the catalytic mechanism of Slingshots and of their activation by actin is lacking. Here, we report that the N-terminal region of human Slingshot2 auto-inhibits its phosphatase activity in a noncompetitive manner. pH-dependent phosphatase assays and leaving-group dependence studies suggested that the N-terminal domain of Slingshot2 regulates the stability of the leaving group of the product during catalysis by modulating the general acid Asp361 in the catalytic VYD loop. F-actin binding relieved this auto-inhibition and restored the function of the general acid. Limited tryptic digestion and biophysical studies identified large conformational changes in Slingshot2 after the F-actin binding. The dissociation of N-terminal structural elements, including Leu63, and the exposure of the loop between α-helix-2 and β-sheet-3 of the phosphatase domain served as the structural basis for Slingshot activation via F-actin binding in vitro and via neuregulin stimulation in cells. Moreover, we designed a FlAsH-BRET–based Slingshot2 biosensor whose readout was highly correlated with the in vivo phosphatase activities of Slingshot2. Our results reveal the auto-inhibitory mechanism and allosteric activation mechanisms of a human Slingshot phosphatase. They also contribute to the design of new strategies to study Slingshot regulation in various cellular contexts and to screen for new activators/inhibitors of Slingshot activity.

Slingshots are a small group of phosphatases that regulate multiple physiological functions, such as axon growth and pathfinding (1)(2)(3), neutrophil chemotaxis (4), innate immune responses (5), and mitosis (6,7), by dephosphorylating cofilin and LIMK 4 to modulate actin dynamics (6). Abnormally increased Slingshot phosphatase activity is found in many human diseases, including cancer (8,9), salmonella-related diseases (10), and Alzheimer's disease (11), as well as vascular diseases such as aortic aneurysms, atherosclerosis, and post-angioplasty restenosis (12,13), which is unsurprising, given their role as an important hub controlling actin dynamics. Reducing Slingshot expression or inhibiting Slingshot activity using small chemical compounds has been suggested as a new strategy for treating these diseases (14,15). Therefore, a comprehensive understanding of the regulation of Slingshot activity and function in different cellular and biochemical contexts will substan-tially help in developing new therapeutic methods targeting this small group of phosphatases.
Slingshots are known to be inactivated by phosphorylation and activated by binding to F-actin (6,16,17). In mammals, the Slingshot family contains three members, Slingshot1-3 (SSH1-SSH3). SSH1 and SSH2 are structurally similar and exhibit good activity toward phospho-cofilin (6). In contrast, SSH3 shows minimal C-terminal sequence identity with SSH1 and SSH2 and exhibits very low activity toward phospho-cofilin (6). Whereas phosphorylation of the SSH1 at Ser 978 by PKD and phosphorylation of the SSH2 on its N-terminal Ser residues by GSK3 dampen their activities toward phospho-cofilin (18,19), F-actin binding to SSHs significantly increases their phosphatase activities (16). Until now, the detailed enzymological understanding and structural characterization of Slingshot catalysis and activation had not yet been established.
In the current study, we screened a series of SSH1 and SSH2 constructs for expression and purification in Escherichia coli. An SSH2-1-490 construct was readily expressed and was purified to homogeneity; the large quantity of the resulting protein facilitated further enzymological and biophysical studies. Our enzymological studies revealed that the N-terminal domain of SSH2 (including a pleckstrin homology-like domain; residues 90 -197) serves as an auto-inhibitory module of SSH2 phosphatase activity by modulating a key catalytic step, the stabilization of the product leaving group. Actin binding to Slingshot relieves the auto-inhibitory effect of the Slingshot N-terminal domain. Biochemical analysis using limited tryptic digestion and biophysical studies using the fluorescence spectroscopy enabled us to identify the key structural features, such as a motif encompassing Leu 63 , in the auto-inhibition of Slingshot phosphatase activity by its N-terminal domain and its subsequent activation after F-actin association. Moreover, the observed conformational changes during Slingshot activation derived from the in vitro results enabled us to design FlAsH-BRET probes of full-length Sling-shot2 that reveal the structural rearrangement and activity of Slingshot2 in cells in response to neuregulin-1␤ (NRG) stimulation. This new biosensor will be a useful tool not only for dissecting regulatory mechanisms and functions of Slingshot in different physiological and pathological contexts, but also for potentially contributing to the screening of small compounds that regulate Slingshot activity in an allosteric manner.

Auto-inhibition by the N-terminal domain of Slingshot2
To investigate how regions other than the phosphatase domain contribute to Slingshot catalytic activity, we generated a series of Slingshot2 truncations (Table S1). A construct (SSH2-1-490) encompassing the N-terminal region (Fig. 1A), catalytic domain, and an additional 40 amino acids at the C terminus was readily expressed in Escherichia coli and purified to homogeneity ( Fig. 1A and Fig. S1). Five constructs with N-or C-terminal deletions were subsequently generated and assayed for their phosphatase activity toward a variety of substrates (Fig. 1A). Stepwise truncation of the N-terminal region caused a gradual increase in SSH2 activity, and the construct with the N terminus deleted (residues 305-490) showed the highest phos-phatase activity toward small artificial substrates, including p-nitrophenyl phosphate (pNPP), 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP), and 3-O-methylfluorescein phosphate (OMFP), regardless of the number of rings in the substrates or the different pK a values of the leaving groups (Fig.  1B and Table S2). Interestingly, further deletion of the C-terminal 40 amino acids (residues 305-450) slightly decreased SSH2 phosphatase activity toward pNPP but significantly decreased activity toward DiFMUP and OMFP, indicating that these residues participate in the recognition of the small multiple ring substrates of SSH2 (Fig. 1B).
Importantly, including the N-terminal 232 amino acids or deleting the C-terminal 35 amino acids of SSH2(1-490) impaired its phosphatase activity toward the protein substrate phospho-cofilin in vitro, suggesting the auto-inhibitory role of the N-terminal 1-232 region and the essential role for the 455-490 region of SSH2 in phospho-cofilin recognition ( Fig. 1C and Fig. S4). Notably, the N-terminal 233-304 region is required for SSH2 phosphatase activity toward phospho-cofilin protein but not essential for the recognition of small artificial substrates, indicating that more structural modules are required for protein substrate recognition by Slingshots. Notably, the gel filtration result indicated that the SSH2(1-490) was a monomer (Figs. S2 and S3). Therefore, the above observed auto-inhibitory effect was mainly intramolecular.

Noncompetitive inhibition of SSH2 phosphatase activity by its N-terminal region
To dissect the auto-inhibitory mechanism of the N-terminal region toward Slingshot phosphatase activity, we purified the N-terminal 1-227 region of Slingshot and examined its effect on the intrinsic phosphatase activity of SSH2(233-490). Lineweaver-Burk plots indicated that the addition of increasing concentrations of N-terminal SSH2(1-227) had no effect on K m but significantly decreased V max (Fig. 1E, left), suggesting noncompetitive inhibition (20,21). Using pNPP as substrate, the K i for the N-terminal 1-227 region toward SSH2(233-490) was 1.6 M Ϯ 0.13. A similar but weaker mode of inhibition by the SSH2(1-227) on the SSH2 catalytic domain (residues 305-490) (K i was 5.41 Ϯ 0.37 M) was also observed (Fig. 1E, right), suggesting that SSH2(1-227) modulates SSH2 phosphatase activity by directly interacting with key catalytic elements located in the phosphatase domain (residues 305-490), slowing its important hydrolysis steps. The weaker inhibition also suggested that the 1-227 region interacts with the 233-305 region of SSH2. Consistently, the N-terminal 1-227 region of the Slingshot2 exhibited a noncompetitive inhibition mode toward its catalytic domain using physiological substrate phospho-cofilin. K i was 5.84 M Ϯ 0.30 (Fig. 1F). A proposed model is illustrated in Fig. 1G.

The SSH2 N-terminal domain modulates the stability of the product leaving group during the catalysis
We next carried out enzymology analysis to delineate how the N-terminal domain of SSH2 modulates catalysis. Similar to all other known tyrosine phosphatases, the pH dependence of the SSH2-hydrolyzed pNPP reaction follows a bell-shaped curve, suggesting the involvement of a general acid-base catalytic mechanism (Fig. 2, A and B). The pH-independent maximum first-order rate (k cat ) lim and second-or-der rate constants (k cat /K m ) lim of SSH2, which are more accurate measurements of intrinsic SSH2 phosphatase activity, were on the order of SSH2(305-490) Ͼ SSH2(233-490) Ͼ SSH2(1-490) (Fig. 2, A and B). Moreover, the basic limbs (pH Ͼ6) of both of the k cat and the k cat /K m of SSH2(1-490) were much shallower than those of SSH2(233-490) and SSH2(305-490) (Fig. 2, A and B), suggesting that the N-terminal domain of SSH2 might involve in the breaking of the P-O bond.

Allosteric regulation of Slingshot
During tyrosine phosphatase catalysis, a conserved general acid was identified in most PTP members to stabilize the negative charge developed on the leaving-group phenol-OH after breakage of the P-O bond by nucleophilic attack of the catalytic cysteine. All three SSH members contain a conserved VYD loop, which is structurally equivalent to the classic WPD loop of tyrosine phosphatases and harbors the general acid (Asp 361 in SSH2). Protonation of Asp 361 in SSH2 is predicted to form a hydrogen bond with the phenolic oxygen of the product, a key step for efficient substrate hydrolysis that appears as the basic limb on the pH-dependence curve. The difference in the basic limb of the pH-dependence curve suggested a less effective general acid function in SSH2(1-490) than in the truncation SSH2(233-490). We therefore used Brønsted plots to investigate the effect of the SSH2 N-terminal domain on the general acid catalysis by measuring the dependence of catalysis on the characteristics of the leaving group (22) (Fig. 2, C and D). Whereas the ␤ 1g values for both k cat and k cat /K m are very similar in SSH2(233-490) and SSH2(305-490), the ␤ 1g value for SSH2(1-490) is significantly lower, confirming the impaired general acid catalysis for the intrinsic activity and the reduced stability of the leaving group during the catalytic process of SSH2(1-490).

F-actin activates SSH2 through relieving auto-inhibition by its N-terminal domain
Although full-length Slingshot exhibited relatively low phosphatase activity toward both small artificial substrates and its physiological substrate phospho-cofilin, a dramatic increase in Slingshot activity has previously been observed following incubation with F-actin (16). The N-terminal Arg 96 , the Leu 185 -Lys 187 region, and the C-terminal Trp 457 of SSH1 were identified as being directly involved in actin binding (16,24). We therefore examined the effect of actin on the phosphatase activity of different SSH2 truncations. Incubation with 4 M F-actin markedly increased the dephosphorylation of phospho-cofilin by SSH2(1-490) but has little effect on SSH2(233-490) or SSH2(305-490) (Fig. 3, A and B). Similarly, incubation of F-actin significantly improved the activity of SSH2(1-490) and SSH2(1-455) toward pNPP, the two constructs encompassing residues 1-232, but had little effect on SSH2(233-490), SSH2(305-490), SSH2(305-450), or SSH2(233-450) (Fig. 3C). In particular, the intrinsic activity of SSH2  in the presence of F-actin was approximately the same as that of SSH2(233-490) toward pNPP. These data suggested that the binding of F-actin disabled the auto-inhibitory function of the SSH2 N-terminal domain.
We then examined the effect of actin on the leaving-group dependence of SSH2-catalyzed pNPP hydrolysis using Brønsted plots. Interestingly, the slope of the Brønsted plot was much shallower in the presence of actin than the plot of SSH2(1-490) alone (Fig. 3, D and E). The ␤ 1g values for the k cat /K m and k cat of SSH2  in the presence of the actin were Ϫ0.08 and Ϫ0.23, respectively, similar to those for SSH2(233-490) and much higher than the ␤ 1g value of SSH2(1-490) alone, suggesting a functional role of restoring the function of the general acid by F-actin association during SSH2 catalysis (Fig. 3F). These results suggested that the binding of actin relieves the auto-inhibitory role of the SSH2 N-terminal domain and restores general acid function during catalysis ( Fig. 3G) (25).

F-actin association leads to SSH2 conformational rearrangements
To investigate the structural basis of SSH2 activation by actin, we next employed limited tryptic proteolysis to monitor conformational changes in SSH2 after association with actin. In general, protein conformational changes will mask or unmask trypsin cleavage sites that produce different trypsin digestion patterns. Because trypsin only recognizes and cleaves proteins at the carbonyl site after Arg or Lys residues, analysis of tryptic digestion patterns has been used to glean structural information in many biochemical studies (26 -28). As illustrated in Fig.  4A, SSH2(1-490) appeared at 68 kDa, and tryptic digestion produced bands with apparent molecular masses of 40 and 28 kDa. Following incubation with F-actin, a new fragment of SSH2 appeared at 44 kDa, and the 28 and 40 kDa bands were not generated (Fig. 4, A and B). Moreover, limited trypsin digestion patterns of SSH2(1-490) and SSH2(1-455) were identical, sug-

Allosteric regulation of Slingshot
gesting that digestion at Lys 450 occurred at the C terminus in both constructs (Fig. 4, C and D). Mapping of the Arg and Lys positions indicated that SSH2 alone was cleaved by trypsin at Arg 66 , Lys 291 , and Lys 450 , whereas SSH2 incubated with actin was mainly cleaved at Arg 66 and Arg 332 (Figs. S5 and S6). Taken together, these results suggested that the linker region (residues 233-304) between the N-terminal auto-inhibitory domain and the phosphatase domain, which encompasses Lys 291 , is more accessible by trypsin in SSH2(1-490) and is protected after F-actin association. In contrast, the loop between ␣-helix-1 and ␤-sheet-3 of the phosphatase domain is protected in SSH2 (1-490) alone (Fig. S6) and is exposed to trypsin after F-actin binding (Fig. 4, E and F). These conformational changes are remote from the active site of the Slingshot catalytic domain, consistent with the noncompetitive inhibition of Slingshot by its N-terminal domain and subsequent activation by an allosteric mechanism. Taken together, these results indicate significant conformational changes after association of actin with SSH2(1-490), including the protection of the linker region between the N-terminal domain and the phosphatase domain and the exposure of the loop between ␣-helix-2 and ␤-sheet-3 of the phosphatase domain ( Fig. 3G and Fig. S6). for the k cat profiles and using Equation 3 for the k cat /K m profiles. The kinetic parameters of k cat lim , pK a , and pK b and pK E1 , pK E2 , and (k cat /K m ) lim were generated with the data presented in the top panel (bottom). Leaving-group dependence of K cat (C) and K cat /K m (D) for SSH2(1-490) (Ⅺ), SSH2(233-490) (ƒ), and SSH2(305-490) (E) was determined at 30°C in 50 mM DMG buffer, pH 7.0, using pNPP (pK a ϭ 7.14), 4-methylumbelliferyl phosphate (pK a ϭ 7.80), ␤-naphthyl phosphate (pK a ϭ 9.38), and O-phospho-L-tyrosine (pK a ϭ 10.07) as substrates. The lines were generated by linear least-squares fitting to a plot of log(K cat ) (C) and log(K cat /K m ) (D) versus leaving-group pK a . E, the ␤ 1g values for K cat and K cat /K m of different SSH2 truncations were obtained by fitting Brønsted plots from C and D. All values were obtained from three independent experiments, and the data are expressed as the mean Ϯ S.E. (error bars).

Allosteric regulation of Slingshot F-actin activates SSH2 by specifically modulating the interaction of the N-terminal region of SSH2 with the VYD loop
Enzymology studies indicated that actin activated SSH2 phosphatase activity by modulating the general acid catalytic process. By examining the available SSH2 crystal structure (29), the general acid of the SSH2 is likely residue Asp 361 (Fig. S7), which is located in the VYD loop of the SSH2 phosphatase domain. To further dissect the structural basis of how actin binding regulates SSH2 activity by modulating the VYD loop, we generated a series of modified SSH2(1-490) fluorescence reporters and utilized the bimane-tryptophan technique (30,31) along with fluorescence spectroscopy to monitor confor-Allosteric regulation of Slingshot mational changes in SSH2 after incubation with F-actin. Mutating all Cys residues to Ser other than the catalytic Cys 392 had no effect on intrinsic phosphatase activity (Fig. S8A), suggesting that these mutations do not perturb the structural integrity of SSH2(1-490). We therefore introduced specific cysteines in the N terminus of SSH2 into a minimal-cysteine SSH2(1-490) background (Fig. 5A) and specifically labeled these single cysteines with monobromobimane. The labeled Slingshot proteins were subsequently purified with GSH-Sepharose 4B to remove unbound bimane, and SDS-PAGE was used to monitor protein purity (Fig. S9). No substitutions or labeling affected phosphatase activity (Fig. S8B).
Taken together, the fluorescence spectroscopy results suggested that Cys 19 and Cys 146 are distal to the VYD loop, whereas Cys 63 , Cys 78 , Cys 98 , and Cys 161 are close by in the resting state. Upon F-actin association, a structural element encompassing Cys 63 moves away from the VYD loop, whereas a structural segment containing Cys 19 moves closer to the VYD loop. Thus, specific conformational rearrangements, including the association and dissociation of specific structural elements that participate in coordinating VYD loop function for effective catalysis, underlie the activation of Slingshot phosphatase activity by F-actin in vitro (Fig. 5F).

A conformational biosensor for Slingshot2 indicates allosteric activation of this protein in cells
To extend our hypotheses regarding the inhibition of Slingshot by its N-terminal domain in a resting state and the release of its N terminus from the VYD loop after subsequent activation in a more physiological context, we prepared a series of FlAsH-BRET probes (32) for Slingshot2 by inserting the sixamino acid motif CCPGCC and the donor Renilla luciferase (rLuc) into full-length Slingshot2 and expressed the resulting modified proteins in MCF-7 cells (Fig. 6A). The FlAsH motif was incorporated into the catalytic VYD loop of full-length Slingshot2, and we assumed that the observed changes in the efficiency of BRET signaling between FlAsH and the inserted rLuc would yield a conformational signature indicating the relationship between the probe's VYD loop and specific positions in the N terminus in a cellular context (Fig. 6A). Both FlAsH-BRET probes (SSH2-23-FlAsH and SSH2-63-FlAsH) retained their functional integrity; in particular, both probes promoted cofilin dephosphorylation to a similar extent as that observed for WT Slingshot2 in response to NRG stimulation (Fig. 6, B-D). These probes consistently exhibited similar phosphatase activities toward the substrates pNPP and phosphocofilin in vitro (Fig. S11, A-D). We then tested whether NRG stimulation produced a conformational change in Slingshot2 that could be captured by intramolecular rLuc-Slingshot2-FlAsH BRET signals from the two probes based on biochemical results. Intriguingly, NRG treatment induced an increase in the BRET signal from SSH2-23-FlAsH and a decrease in the BRET signal from SSH2-63-FlAsH, suggesting that residue 23 moves closer to the catalytic VYD loop of the Slingshot2 but that residue 63 moves away from this catalytic structure; these findings exhibit good agreement with observations of conformational changes of allosteric activation of Slingshot2 in vitro (Figs. 5F and 6D).
Based on these consistent findings that the N-terminal region of Slingshot2 that includes position 63 moved away from the catalytic VYD loop, we postulated that SSH2-63-FlAsH could be used to report Slingshot2 activity changes in cells. As expected, the ⌬BRET values determined using SSH2-63- buffer, and then pNPP (pK a ϭ 7.14), 4-methylumbelliferyl phosphate (pK a ϭ 7.80), ␤-naphthyl phosphate (pK a ϭ 9.38), and O-phospho-L-tyrosine (pK a ϭ 10.07) were added for activity determination. F, the ␤ 1g values were obtained by fitting Brønsted plots from Fig. 5 (D and E). G, a speculated schematic description of the relief of auto-inhibition by the SSH2 N-terminal domain upon F-actin binding. F-actin binding relieves auto-inhibition by the SSH2 N-terminal domain, resulting in substrate access to the catalytic site. For all statistical analyses, data from at least three independent experiments were quantified and presented as the mean Ϯ S.E. (error bars). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.005. In A and C, trypsin was incubated with SSH2 or F-actin-activated SSH2 at a ratio of 1:1000 (w/w) in 20 mM Tris, pH 7.0, 100 mM NaCl at 37°C for the indicated times. Aliquots (ϳ2 g of protein sample) were subjected to SDS-PAGE. E and F, schematic description of F-actin activation leading to SSH2 conformational rearrangements. In the case of SSH2(1-490) alone (E), the initial tryptic proteolysis generated two relatively large fragments with molecular mass of 40 and 28 kDa, respectively. F-actin binding leads to SSH2 conformational rearrangements, with the exposure of multiple accessible trypsin cleavage sites (F), which is responsible for the generation of only one visible fragment ϳ4 kDa larger than the corresponding fragment formed by digestion of SSH2(1-490) in the absence of F-actin.

Allosteric regulation of Slingshot
FlAsH exhibited dose dependence and were well correlated with cofilin dephosphorylation levels in response to NRG stimulation (R 2 ϭ 0.9; Fig. 6 (E-G) and Fig. S11E). This strong linear correlation between ⌬BRET and Slingshot2 activity indicated that SSH2-63-FlAsH served as an effective reporter of Slingshot activity in response to allosteric regulation in cells treated

Allosteric regulation of Slingshot
with NRG. A previous report (19) has indicated that the Ser 21 , Ser 25 , Ser 32 , and Ser 36 in the N-terminal region of SSH2 were phosphorylated by GSK3␤, which attenuated the activity of SSH2 toward dephosphorylation of phospho-cofilin. To extend our identified structural alteration and auto-inhibitory mechanism of SSH2 in this specific cellular context, we mutated the Ser 21 and Ser 25 to Asp and Ser 32 and Ser 36 to Glu to create an SSH2-DDEE phosphomimetic mutant. In response to NRG treatment, the SSH2-DDEE-63-FlAsH ⌬BRET signal was significantly lower compared with the SSH2-63-FlAsH (Fig. 6H). This result suggested that the inhibitory mechanism of SSH2 phosphorylated by GSK in the N-terminal SSH2 Ser 21 -Ser 25 -Ser 32 -Ser 36 residues was potentially due to the locking of the SSH2 in an auto-inhibitory mechanism, impairing its ability to dislodge the N-terminal region from the catalytic site in response to actin binding.

Discussion
The reversible regulation of protein phosphorylation states is coordinated by two groups of enzymes, protein kinases and protein phosphatases. Although auto-inhibition and subsequent allosteric activation serve as important regulatory mechanisms for many protein tyrosine kinases to achieve their precise function in different cellular contexts (33)(34)(35), few protein phosphatases are known to possess an auto-inhibitory mechanism (36,37). Whereas recently we have identified the allosteric inhibition of calcineurin by structural studies (36), the bestcharacterized tyrosine phosphatase with an auto-inhibitory mechanism is SHP2, an important oncogenic PTP involved in Noonan syndrome and other types of cancers (21,(37)(38)(39)(40). Similar to SHP2, activation of Slingshots promotes cell motility (6,21,41). Therefore, an auto-inhibitory mechanism is important to retain low phosphatases activity in the resting state, and abnormal activation of these phosphatases is related to the progression of diseases such as cancers (9,21,37).
In contrast to SHP2, whose N-terminal SH2 domain inserts into the catalytic center and competitively inhibits phosphatase activity (21,40), here we found that the N-terminal domain inhibited SSH2 in a noncompetitive manner. Our enzymological studies showed that the N-terminal domain inhibited SSH2 activity by modulating the stability of the product leaving group, an important catalytic step that requires the general acid Asp 361 in the VYD loop of Slingshot. Adding F-actin relieved this auto-inhibition, and biophysical studies suggested that a large conformational change underlies F-actin-mediated SSH2 activation. Therefore, Slingshot activity is regulated by F-actin through delicate structural arrangements, which may account for its exceptional substrate specificity.
Previous studies found that the N-terminal regions are required for F-actin-mediated Slingshot activation and that residues Arg 96 , Leu 185 , His 186 , Lys 187 , and Trp 457 of SSH1 may directly interact with F-actin (16,24). Our biophysical studies found that the 98-C-bimane probe underwent significant fluorescence quenching after incubation with F-actin, confirming that the loop between the ␤1 and ␤2 sheets of the predicted PH-like domain (42) is one of the critical interaction sites of Slingshot with F-actin. Moreover, fluorescence spectroscopy revealed that the N-terminal region encompassing Leu 63 is not only proximal to the VYD loop of SSH2 in the resting state but also a potential docking site for F-actin binding. After F-actin binding, Leu 63 becomes dislodged from the VYD loop of SSH2. Therefore, the structural motif harboring Leu 63 may be a determinant for Slingshot activation by inhibiting SSH2 activity through its interaction with the VYD loop in the resting state and subsequent dissociation from the VYD loop following F-actin binding. Importantly, consistent with the aforementioned findings, the release of a segment encompassing Leu 63 of the Slingshot N terminus from the catalytic VYD in response to NRG stimulation was verified in cells. Moreover, we developed a biosensor that reflects Slingshot2 activity and conformational changes associated with allosteric regulation. This type of biosensor could serve as an important tool to investigate the activation mechanism of Slingshot in different physiological and pathological processes. For example, the phosphorylation of SSH2 in its N-terminal by GSKs was known to attenuate its activity toward phospho-cofilin. Using our FlAsH BRET assay, we were able to dissect the mechanism underlying this specific cellular event. These results indicated that the phosphorylation of SSH2 by GSK3 locked SSH2 in an auto-inhibitory configuration, impairing the dislodging of the N-terminal region of SSH2 from its catalytic site in response to activation.
Recently, an allosteric SHP2 inhibitor was identified that concurrently targets the interface of both the N-terminal inhibitory domain and phosphatase domain and was shown to exhibit strong therapeutic potential (21). Therefore, an inhibitor targeting both the N-terminal Leu 63 region and the VYD loop of Slingshot is conceivable to drive the development of novel allosteric Slingshot inhibitors. A fluorescence assay that utilizes our newly developed L63C-bimane and L63C-bimane-Y360W probes or a BRET assay with our SSH2-63-FlAsH biosensor may also be useful to screen for allosteric Slingshot regulators.
In conclusion, we have shown that the N-terminal region of Slingshot plays an auto-inhibitory role via allosteric regulation of the catalytic VYD loop and restriction of the stability of the

Allosteric regulation of Slingshot
leaving group of the product during catalysis in the resting state. Movement of the segment encompassing Leu 63 away from the VYD loop is important for the allosteric activation of Slingshot2 both in vitro due to F-actin binding and in cellular contexts in response to NRG stimulation. A fluorescence assay in vitro and a BRET assay in cells were developed; these assays could be used to dissect regulatory mechanisms of Slingshot proteins in various contexts or to screen for allosteric Slingshot inhibitors or activators.

Allosteric regulation of Slingshot
lalanyl chloromethyl ketone-treated trypsin (catalogue no. T1426) was ordered from Sigma. Monobromobimane probe was obtained from Thermo Fisher Scientific. Cell culture medium and fetal bovine serum were ordered from Hyclone GE Healthcare. Lipofectamine 2000 transfection reagent and the TC-FlAsH II In-Cell tetracysteine detection reagent were purchased from Invitrogen. NRG was ordered from Novoprotein. The coelenterazine was obtained from Promega.

Plasmid construction
Human full-length SSH1 and SSH2 were originally kind gifts from Dr. Lefkowitz (Duke University) and Dr. Zhong-Yin Zhang (Purdue University). Their E. coli expression constructs and derivative truncations (see Table S1 for detailed information) were subcloned into pET-15b and pGEX-6P-2, respectively, resulting in plasmid coding for N-terminally His 6 -tagged or GST fusion proteins. For fluorescent probe monobromobimane (mBBr) labeling, SSH2 single cysteine mutants were constructed by introducing a cysteine residue into specified positions of SSH2 cysteine minus mutant in which all potential reactive cysteines were replaced by serines (see Fig. 5A). We subcloned SSH2(1-1423) into pCDNA3.1 vector with an His 6 tag at the N terminus. Then we added the rLuc into the 23-and 63-positions. The two plasmids were constructed by introducing an amino acid motif, CCPGCC, following amino acid residue 361 of SSH2 (see Fig. 6A). All SSH2 point mutations were obtained by PCR amplification following the standard operating procedure of the QuikChange site-directed mutagenesis kit (Stratagene) using plasmid coding for GST-tagged SSH2 as the template. The sequences of WT and mutant SSH2 were verified by DNA sequencing. Oligonucleotides used in this study are listed in Table S3.

Expression and purification of SSH2
Plasmids coding for His-tagged WT and mutant SSHs were transformed in E. coli BL21(DE3) cells and followed by isopropyl-1-thio-␤-D-galactopyranoside (IPTG) induction as described previously (46). When absorbance at 600 nm for cell cultures grown in lysogeny broth medium was between 0.6 and 0.8, IPTG was added to a final concentration of 0.1-0.3 mM, and the culture was further induced at 23°C for 12-16 h. The His 6 -tagged proteins were purified by Ni-NTA resin as described previously (44). After a 2-h incubation of lysates with Ni-NTA-agarose a 4°C by gentle rotation, the agarose was washed with His buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl) 3-4 times. This washing step can get rid of most nonspecific proteins. After the wash, different concentrations were prepared of imidazole elution buffer, which contains 20, 50, 100, 200, or 300 mM imidazole. We normally concentrated protein to a final concentration between 1 and 10 mg/ml and used the centrifugal filter (Millipore) by centrifuging at 4°C. During the concentration, the protein was changed to the HN (20 mM Hepes, pH 7.5, 100 mM NaCl, 1 mM DTT) buffer to eliminate the imidazole. SDS-PAGE was used to determine the purity of the protein. The proteins were further purified by size-exclusion chromatography. The purity of the His-SSH2 was over 95% as examined.
The procedure to purify GST-tagged protein was similar to that described previously (43). Plasmids coding for GST tag SSHs and mutations were transformed in E. coli BL21(DE3) cells and induced by 300 M IPTG for 16 h at 23°C until OD reached 0.8 -1.0. The GST-tagged proteins were purified by GSH-Sepharose 4B. Bacteria were resuspended in chilled GST buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Triton X-100, 5% glycerol, 1 mM EDTA, 2 mM DTT). The cells were broken, and then the lysis was transferred into tubes to spin at 12,000 rpm for 30 min. The supernatant was removed and incubated with GSH beads for 2 h at 4°C by gentle rotation. Beads were washed at least 3 times with GST buffer and then resuspended in GST elution buffer (10 mM reduced GSH prepared with GST buffer). Beads were spun at a lower speed, always 800 rpm, for 5 min. The supernatant was transferred into a centrifugal filter (Millipore) column to spin at 4000 rpm. The protein concentration was determined by A 280 and Coomassie Blue staining.

Allosteric regulation of Slingshot
adjusted to 0.15 M by NaCl) as described previously (43,46), unless otherwise specified. All reactions were initiated by the addition of appropriate amount of SSH2 into the reaction mixtures containing different substrates at various concentrations, except that, in the experiment to examine the effect of F-actin on activation of SSH2 phosphatase activity, SSH2 (0.4 M) was preincubated with F-actin (4 M) at room temperature for 30 min before use. For pNPP and OMFP, reaction was terminated by the addition of an equal volume of 1 M NaOH, and the hydrolyzed products were determined by measuring the absorbance at 405 and 477 nm, respectively. For DiFMUP and all pTyr-, pSer-, pThr-containing phosphopeptides, reactions were quenched by Biomol Green reagent, and the production of P i was detected by monitoring the absorbance at 620 nm as described previously (47). Kinetic parameters were fitted to the Michaelis-Menten equation (Equation 1) as described previously (48). To examine the catalytic activity of SSH2 toward physiological substrate p-cofilin, SSH2 (0.4 M) was incubated with 0.4 mM p-cofilin in a total volume of 50 l of assay buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 2 mM DTT). The reaction was stopped at different time intervals, and the phosphorylation level of cofilin was examined by immunoblotting with anti-phospho-cofilin antibody. Quantification of signal intensities from three immunoblots was done using ImageJ software (National Institutes of Health). The data were analyzed by GraphPad Prism version 5 (GraphPad Software).

P i assay of p-cofilin protein dephosphorylation by SSH2 in vitro
The dephosphorylation was performed in a reaction buffer containing 20 mM Tris-HCl, pH 7.0, 100 mM NaCl, 2 mM DTT at 37°C. The initial p-cofilin concentration was 2 M, and then it was diluted hole by hole in a 96-well plate with or without SSH2. The reaction was terminated by Biomol Green (Enzo Life Sciences) after that, with the end-point reading at 620 nm.

The inhibition assay
The inhibition mode for SSH2 N-terminal region was determined by examining the effects of SSH2(1-227) on SSH2 (catalytic domain)-catalyzed pNPP or p-cofilin protein hydrolysis in DMG buffer at 30°C. The data were fitted to the Lineweaver-Burk equation (Equation 2) as described previously (48).

pH profile assay
Effects of pH on SSH2-catalyzed hydrolysis of pNPP were performed in the following buffers: 50 mM succinate, pH 5.0 -6.5; 50 mM DMG, pH 6.6 -7.5; 50 mM Tris, pH 7.5-8.5. All of the above-mentioned reaction buffers contained 1 mM EDTA, 2 mM DTT, and the ionic strengths were adjusted to 0.15 M using NaCl. The k cat and k cat /K m values for different SSH2 truncations were measured in various buffers using pNPP as the substrate. Then the data were fitted to the Michaelis-Menten equation (Equation 1) as described previously. After that, k cat /K m versus pH profile values and the k cat versus pH profile values were fitted to Equations 3 and 4 as described previously (46,49), where H is the proton concentration.

Leaving-group dependence
The assays were conducted in 50 mM DMG buffer (pH 7.0, 1 mM EDTA, 2 mM DTT with a 0.15 M ionic strength) at 30°C. Small molecular substrates with different pK a (pNPP, 7.14; MUP, 7.8; 2-naphthyl phosphate sodium salt, 9.38; O-phospho-L-tyrosine, 10.04) were used as substrates in this assay (50,51). The k cat and k cat /K m data were fitted to the Michaelis-Menten equation (Equation 1) as described previously (46,52). Log(k cat ) and log(k cat /K m ) values were plotted against the leaving-group pK a of the substrates to derive the leaving-group dependence curve, and the ␤ 1g values were acquired. All data were analyzed by GraphPad Prism version 5 (GraphPad Software).

Preparation of F-actin
Skeletal G-actin was extracted from rabbit hind leg muscle and stored in G buffer (5 mM Tris-HCl, pH 8.0, 0.2 mM ATP, 1 mM DTT, 0.1 mM CaCl 2 , and 1 mM NaN 3 ) at a concentration of 10 M as described previously (53). To convert G-actin to F-actin, MgCl 2 was added to G-actin solution to a final concentration of 2 mM, and KCl was added to a final concentration of 0.8 M. The reaction mixture was further incubated at 4°C for 1 h with stirring, and polymerized F-actin was collected by centrifugation at 100,000 ϫ g for 1 h. F-actin pellets at the bottom of tubes were resuspended with F buffer (15 mM HEPES, pH 7.5, 50 mM KCl, 1 mM MgCl 2 , 0.1 mM EGTA, 0.005% NaN 3 ).

Limited trypsin proteolysis
The limited trypsin digestion reactions were performed as described previously (54), with minor modifications. A total of 500 g of SSH2(1-490) (or SSH2(1-455)) at 2 M, premixed with or without 4 M F-actin, or F-actin alone was incubated with tosylphenylalanyl chloromethyl ketone-treated trypsin at a trypsin/SSH2 mass ratio of 1:1000 in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl. The reaction mixtures were maintained at 37°C. Aliquots were collected at 0, 15, and 30 min and analyzed by SDS-PAGE to determine the effects of F-actin on the digestion pattern. For precise trypsin cleavage site analysis, samples were further subjected to protein sequencing using Edman degradation as described previously (28). The N-terminal protein sequencing was performed by the Beijing Agricultural Biology Monitoring Center.

Labeling of SSH2 with mBBr
Labeling of the SSH2 with mBBr was carried out using a protocol described previously (31,47). A noncysteine-GST tag was achieved by mutation of all four cysteines in GST to Ser.

Allosteric regulation of Slingshot
The resulted GST-noncysteine construct showed normal GST bead-binding activity. Briefly, GST-noncysteine-tagged single-cysteine SSH2 mutation proteins were buffer-exchanged three times using labeling buffer containing 10 mM MES, pH 6.5, 150 mM NaCl. The mutants were ultracentrifuged to a final concentration of 50 M and incubated with a 10-fold molar excess of mBBr (stock in 1% DMSO) at room temperature for 3 h with gentle rotation. The reaction mixtures were then centrifuged at 12,000 rpm for 30 min, and the supernatants were further incubated with GSH resin for 1 h. SSH2 bound to GSH resin was washed extensively with washing buffer (10 mM Hepes, pH 7.4, 250 mM NaCl) and eluted by 10 mM GSH. The mBBr-labeled SSH2 was further purified by size-exclusion chromatography using a buffer containing 10 mM Hepes, pH 7.4, 150 mM NaCl and concentrated to 20 M. The calculation of the labeling efficiency was performed as described previously (55).

Fluorescence spectroscopy
For the fluorescence spectroscopy assay, reactions were performed in a final volume of 100 l in reaction buffer (10 mM Hepes, pH 7.4, 150 mM NaCl) containing mBBr-labeled SSH2 mutants (2 M) with or without F-actin (4 M). The bimane fluorophore was excited at 390 nm, and emission was collected between 410 and 600 nm (2-nm step size, 0.5-s integration per point) as described previously (47).

Cell culture and transfection
MCF-7 cells were maintained in high-glucose minimum essential medium supplemented with 10% fetal bovine serum and 1% antibiotic/antimycotic solution. Transfections were performed using Lipofectamine 2000. The cells' confluence was ϳ90 -95% before transfection, and the signaling assays were performed with cell confluence at ϳ75%.

Intramolecular FlAsH BRET assay
MCF-7 cells were seeded in 6-well plates after transfection with AT1R-FlAsH (as a control), SSH2 WT, SSH2-23-FlAsH, and SSH2-63-FlAsH with rLuc inserted in a specific N-terminal site. Before the BRET assay, MCF-7 cells were starved with serum for 1 h. Then cells were digested, centrifuged, and resuspended in 500 l of BRET buffer (25 mM Hepes, 1 mM CaCl 2 , 140 mM NaCl, 2.7 mM KCl, 0.9 mM MgCl 2 , 0.37 mM NaH 2 PO 4 , 5.5 mM D-glucose, 12 mM NaHCO 3 ). The TC-FlAsH reagent was added at a final concentration of 2.5 M and incubated at 37°C for ϳ1 h. Subsequently, cells were washed with BRET buffer and then distributed into white-wall clear-bottom 96-well plates, with ϳ100,000 cells/well. The cells were treated with 50 ng/ml NRG at 37°C for 15-20 min, and then coelenterazinc was added at a final concentration of 5 M, followed immediately by checking of the luciferase (440 -480 nm) and TC-FlAsH (525-585 nm) emissions. The BRET ratio (emission enhanced yellow fluorescent protein/emission rLuc) calculated using a Berthold Technologies Tristar 3 LB 941 spectrofluorimeter. The procedure was modified from those described previously (32,56).

In vitro pulldown assay
SSH2 WT, SSH2-23-FlAsH, and SSH2-63-FlAsH with a His tag at their N termini were transfected into three 150-mm plates of MCF-7 cells for purification. Lysates were incubated with Ni-NTA-agarose at 4°C for about 2 h with gentle rotation. After centrifugation, pallets were washed and eluted with gradient imidazole. The fraction containing the Slingshot was analyzed by immunoblotting with His-antibody and collected accordingly. The purity of these proteins was examined by Western blotting. The enzyme activity was assayed using the substrate pNPP or phospho-cofilin protein (see Fig. 6 (D and E) and Fig. S11).

Statistics
All of the Western blots were performed independently at least three times, and the data are presented as the mean Ϯ S.D. All kinetic data are presented as the mean Ϯ S.E. of more than three independent experiments. Statistical comparisons were analyzed using analysis of variance with GraphPad Prism version 5 or GraphPad Prism version 7.