Molecular mechanism of Afadin substrate recruitment to the receptor phosphatase PTPRK via its pseudophosphatase domain

  1. Iain M Hay
  2. Katie E Mulholland
  3. Tiffany Lai
  4. Stephen C Graham
  5. Hayley J Sharpe  Is a corresponding author
  6. Janet E Deane  Is a corresponding author
  1. Cambridge Institute for Medical Research, University of Cambridge, United Kingdom
  2. Signalling Programme, Babraham Institute, United Kingdom
  3. Department of Pathology, University of Cambridge, United Kingdom

Abstract

Protein tyrosine phosphatase receptor-type kappa (PTPRK) is a transmembrane receptor that links extracellular homophilic interactions to intracellular catalytic activity. Previously we showed that PTPRK promotes cell–cell adhesion by selectively dephosphorylating several cell junction regulators including the protein Afadin (Fearnley et al, 2019). Here, we demonstrate that Afadin is recruited for dephosphorylation by directly binding to the PTPRK D2 pseudophosphatase domain. We mapped this interaction to a putative coiled coil (CC) domain in Afadin that is separated by more than 100 amino acids from the substrate pTyr residue. We identify the residues that define PTP specificity, explaining how Afadin is selectively dephosphorylated by PTPRK yet not by the closely related receptor tyrosine phosphatase PTPRM. Our work demonstrates that PTP substrate specificity can be determined by protein–protein interactions distal to the active site. This explains how PTPRK and other PTPs achieve substrate specificity despite a lack of specific sequence context at the substrate pTyr. Furthermore, by demonstrating that these interactions are phosphorylation-independent and mediated via binding to a non-catalytic domain, we highlight how receptor PTPs could function as intracellular scaffolds in addition to catalyzing protein dephosphorylation.

Editor's evaluation

This Research Advance follows up on the authors' recent eLife article where they reported the identification of a series of pTyr protein targets for the protein–tyrosine phosphatase (PTP) activity of the receptor-like PTP, PTPRK. Here they identify a docking site in the catalytically inactive D2 pseudophosphatase domain that promotes substrate dephosphorylation by the D1 phosphatase domain. The evidence for the docking interaction is convincing, and the study is important because it suggests that the D2 pseudophosphatase domains of other RPTPs may similarly function in substrate recruitment and selectivity. The article will be of interest to biochemists studying protein phosphorylation–dephosphorylation and signal transduction mechanisms.

https://doi.org/10.7554/eLife.79855.sa0

Introduction

Tyrosine phosphorylation is a key post-translational modification in cellular communication, enabling cells to dynamically adapt their behavior in response to external cues. Kinases (PTKs) and phosphatases (PTPs) act in concert to maintain cellular phosphotyrosine (pTyr) levels. Dysregulation of this balance is associated with disease, including tumor progression (Arora and Scholar, 2005; Hunter and Sefton, 1980). The importance of these pathways is exemplified by the clinical use of PTK inhibitors and efforts to therapeutically target PTPs. It has been challenging to develop PTP inhibitors due to their highly charged active sites, as well as their lack of selectivity at the peptide level (Barr et al., 2009). Despite a lack of specific consensus sequences for pTyr substrates (Barr et al., 2009; Ren et al., 2011), PTPs show marked substrate specificity in vivo (Fearnley et al., 2019; Saxena et al., 1999). Thus, understanding how PTPs engage their substrates could lead to the development of alternative strategies to target them in the clinic.

Protein tyrosine phosphatase receptor-type kappa (PTPRK) is a member of the R2B family of transmembrane receptor (R)PTPs that supports cell adhesion and has been implicated as a tumor suppressor (Chang et al., 2020; McArdle et al., 2001; Shimozato et al., 2015). Like other receptor PTPs, PTPRK possesses two intracellular PTP domains, one catalytically active proximal to the plasma membrane (D1) and a distal inactive, pseudophosphatase (D2). We previously showed that for several PTPRK substrates the active D1 PTP domain itself appears not to be the sole determinant of substrate binding. Instead, the inactive D2 domains of PTPRK and its paralog PTPRM were shown to have critical roles in substrate recruitment (Fearnley et al., 2019). Specifically, we identified Afadin (MLLT4/AF6) as a PTPRK substrate and identified that it is recruited via the PTPRK D2 domain (Fearnley et al., 2019). In a similar paradigm, structural studies of the PTPRC-ICD suggested a potential role for the D2 domain in substrate recruitment (Nam et al., 2005). Indeed, it has been shown that the presence of the PTPRC-D2 is critical for the recruitment and dephosphorylation of several proposed substrates, including the T cell receptor (Kashio et al., 1998) and the Lck Src-family tyrosine kinase (Felberg et al., 2004).

Afadin is a conserved junctional plaque protein, important for linking cell surface adhesion molecules to the cytoskeleton, and is regulated by tyrosine phosphorylation (Niessen and Gottardi, 2008; Yu and Zallen, 2020). Afadin deletion in mice is embryonically lethal and leads to dysregulated junctions and polarity (Zhadanov et al., 1999). It is a large (>200 kDa) protein possessing multiple folded domains and extended regions of predicted disorder. The annotation of these domains, as described in UniProt (UniProt ID: P55196), is in good agreement with predictions of ordered/structured regions as calculated by AlphaFold2 (AF2) and IUPred3 (Erdős et al., 2021; Jumper et al., 2021; Figure 1A, Figure 1—figure supplement 1). This modular domain structure allows Afadin to engage in multivalent interactions through distinct domains. Afadin is an effector of GTP-bound Ras (Goudreault et al., 2022; Smith et al., 2017), placing it downstream of activated receptor tyrosine kinases. It can also play a structural role, linking adhesion receptors, such as nectins, to the actin cytoskeleton (Takai and Nakanishi, 2003). It participates in additional scaffolding functions, for example, its PDZ domain binds the plekstrin homology (PH) domain-containing adherens junction protein PLEKHA7 (Kurita et al., 2013). Its interaction with LGN, a protein involved in spindle orientation during cytokinesis, was characterized structurally and showed binding to a C-terminal disordered region of Afadin, adjacent to its F-actin binding region (Carminati et al., 2016). This structure clearly demonstrates that these disordered regions of Afadin can engage in specific interactions via formation of elongated peptide interfaces along extended surfaces of partner proteins. Within one of these extended, disordered regions of Afadin lies residue Y1230, the tyrosine that was identified as the putative target for dephosphorylation by the PTPRK D1 domain as it was hyperphosphorylated in PTPRK knockout (KO) cells (Fearnley et al., 2019). Phosphorylation of Afadin is required for its interaction with SHP-2 and is important for downstream signaling by platelet-derived growth factor (Nakata et al., 2007). Thus, Afadin is an essential regulator of adherens junctions that mediates phosphorylation-dependent and multivalent protein–protein interactions.

Figure 1 with 3 supplements see all
PTPRK interacts directly with the Afadin C-terminal coiled-coil (CC) region.

(A) Top: schematic of full-length human Afadin, with domain annotation based on UniProt ID P55196, showing Ras-association (RAS, yellow), CC (green), forkhead-associated (FHA, orange), dilute (DIL, blue), PDZ (pink), F-actin binding (magenta) domains, and LGN-binding peptide motif (black). The putative PTPRK target site Y1230 is also highlighted (red). Bottom: predicted local distance difference test (pLDDT) for the Afadin AlphaFold2 (AF2) prediction from the AlphaFold Protein Structure Database (P55196, retrieved 7/2/2022). A pLDDT score <50 is predictive of protein disorder (Jumper et al., 2021). (B) Immunoblot analysis of pervanadate-treated MCF10A cell lysates incubated for 45 min at 4°C with 0.3 µM of the indicated recombinant PTP domains. Note: the total Afadin antibody is sensitive to phosphorylation likely at Y1230 (antigen: Afadin 1091–1233) and therefore indicates dephosphorylated Afadin. A full time course is quantified in Figure 1—figure supplement 2E. (C) Immunoblot analysis of streptavidin bead-conjugated PTPRK-ICD pull-downs from wheat germ lysate containing full-length Afadin. (D) Top: schematic of different Afadin C-terminal truncations used in initial mapping experiments, colored as in (A). Bottom: immunoblot analysis of streptavidin bead-conjugated PTPRK-ICD pull-downs from wheat germ lysates containing C-terminal Afadin truncations. Prey proteins enriched on both beads-only and PTPRK pull-downs were considered to be nonspecific interactions. (E) Top: schematic of different Afadin C-terminal GST-fusion constructs used for interaction mapping. Bottom: pull-downs using streptavidin bead-conjugated PTPRK-ICD with purified GST-Afadin fusion proteins, followed by SDS-PAGE and Coomassie staining. (F) Top: sequence of Afadin-CC showing predicted helical region (green block) as observed in the full-length Afadin AF2 prediction. This region contains a high number of charged residues, which have been highlighted (basic, blue; acidic, red). Bottom: pull-downs using streptavidin-conjugated-PTPRK-ICDs with either GST or GST-Afadin-CC in the presence of increasing NaCl concentrations (10, 100, 250, 500 mM; left to right) followed by SDS-PAGE and Coomassie staining. Gels and blots shown in this figure are representative of n 3 independent experiments.

Using whole-cell lysates, we previously reported that the interaction between Afadin and PTPRK-D2 was independent of phosphorylation. This raises the question of whether the interaction is mediated by a region of Afadin that does not encompass the phospho-peptide or is mediated via an additional binding partner. Interestingly, the paralogous PTPRM did not recruit and dephosphorylate Afadin, indicating distinct specificities even within the same family of receptors. In contrast, the D2 domain of an inactive paralog PTPRU was sufficient to recruit Afadin to the PTPRK D1 for dephosphorylation (Hay et al., 2020). To define the molecular details of R2B PTP substrate specificity, the interaction of PTPRK with its substrate Afadin was investigated.

Results

PTPRK dephosphorylates Afadin Y1230 in a D2 domain-dependent manner

Previous work has shown that the Afadin residue Y1230 is hyperphosphorylated in PTPRK KO epithelial cells, suggesting that this is a tyrosine targeted for dephosphorylation by PTPRK (Fearnley et al., 2019). To test this, we generated an antibody against a synthetic tyrosine phosphorylated peptide corresponding to amino acids 1224–1235 of Afadin (Figure 1—figure supplement 2A). Consistent with our previous tyrosine phosphoproteomics data, a band at the predicted molecular weight of the longest isoform of Afadin is detected at higher levels in PTPRK KO compared to WT MCF10A cells (Figure 1—figure supplement 2B). This band is increased by pervanadate (Huyer et al., 1997), a cell treatment that globally increases protein tyrosine phosphorylation, and is decreased by knockdown of Afadin using siRNA (Figure 1—figure supplement 2C). It should be noted that phosphorylation appears to interfere with the signal when using an anti-Afadin (epitope: 1091–1233) antibody from BD Transduction Laboratories on immunoblots, resulting in decreased intensity when Afadin is phosphorylated (Figure 1—figure supplement 2A and C). To further validate the site specificity of the Afadin-pY1230 antibody, we generated tyrosine to phenylalanine (YF) mutants of mScarlet-tagged Afadin. The antigenic phosphopeptide used as an epitope possesses two tyrosine residues, Y1226 and Y1230, which we mutated alone or in combination. mScarlet-Afadin phosphorylation, as detected by our antibody, increased with pervanadate treatment and was completely reversed by calf intestinal phosphatase treatment (Figure 1—figure supplement 2D). However, there was no detectable increase in signal upon pervanadate treatment of mScarlet-Afadin-Y1230F mutants, supporting that this is the predominant phosphosite recognized by our Afadin phosphoantibody. We previously demonstrated that the PTPRK ICD, but not the PTPRK D1 domain alone, dephosphorylates Afadin (Fearnley et al., 2019). This is recapitulated using the pY1230 Afadin antibody after incubation of recombinant domains with pervanadate-treated, DTT-quenched, MCF10A lysates (Figure 1B, Figure 1—figure supplement 2E). Over a time course, PTPRK-ICD shows greater activity against Afadin pY1230 than PTPRK-D1. In contrast, p120Cat was dephosphorylated equally by the two domains, with minimal dephosphorylation of the non-substrate control, paxillin-pY118. Finally, the observed hyperphosphorylation of Afadin Y1230 in PTPRK KO MCF10A cell lysates could be reduced by doxycycline-induced expression of PTPRK to wildtype levels, but not by a catalytically inactive mutant (Figure 1—figure supplement 2F and G). Together, these data confirm that this new antibody is specific to Afadin phosphorylated at Y1230 and that pY1230 is targeted for dephosphorylation by PTPRK in a D2 domain-dependent manner. This pTyr residue is located in the C-terminal half of Afadin within an extended region of predicted disorder (Figure 1A). We next sought to determine how Afadin is recognized by the PTPRK D2.

Afadin binds directly to the PTPRK intracellular domain

The large size and predicted disorder of Afadin make it poorly suited to recombinant expression in Escherichia coli. Therefore, to assay direct binding of Afadin to PTPRK, recombinant full-length Afadin was expressed in vitro using a wheat germ-based cell-free translation system. Bacterially expressed and biotinylated PTPRK intracellular domain (ICD, Figure 1—figure supplement 3) was immobilized on streptavidin beads for pull-downs from wheat germ lysate containing full-length Afadin. We observed clear binding of full-length Afadin to the PTPRK-ICD (Figure 1C), consistent with a direct interaction. To determine whether PTPRK binds a region of Afadin incorporating tyrosine 1230, pull-downs were performed using C-terminal truncations of Afadin produced in wheat germ lysate. Deletion of the Afadin C-terminal disordered tail was sufficient to abolish binding to PTPRK (Figure 1D, aa. 1–1097). However, additional truncations showed substantial nonspecific binding to the streptavidin resin hindering further interaction mapping using this strategy (e.g., aa. 1–972, Figure 1D).

We next mapped PTPRK binding to the Afadin C-terminus in more detail using GST-fusion proteins (Figure 1E). A GST-fusion construct encompassing the entire C-terminal disordered region of Afadin, shown to be important in the wheat germ system, also showed direct binding to PTPRK, supporting the identification of this region as sufficient for binding (Figure 1E). Several truncation constructs within this region were tested for binding to PTPRK, clearly identifying that only constructs containing a short, 63-residue predicted coiled-coil (CC) region spanning residues 1393–1455 was able to bind PTPRK (Figure 1E). Therefore, this CC region is both necessary and sufficient for binding to the PTPRK-ICD. The sequence composition of this region of Afadin (Afadin-CC) includes a high proportion of charged residues (Figure 1F), suggesting that the PTPRK–Afadin interaction could be electrostatically mediated. In support of this, the PTPRK–Afadin interaction was sensitive to salt concentration, with less GST-Afadin-CC present in PTPRK pull-downs at increasing NaCl concentrations (Figure 1F). Further, PTPRK-ICD and Afadin-CC have complementary predicted isoelectric points (pI = 5.3 and 10.2, respectively), suggesting that these fragments would retain charge complementarity at physiological pH. Together, these data show that the PTPRK-ICD interacts directly and specifically with the C-terminal Afadin-CC region.

Afadin-CC and PTPRK D2 form a 1:1 complex

CC domains consist of heptad repeats (designated a–g), in which residues ‘a’ and ‘d’ are generally nonpolar, forming a hydrophobic seam that facilitates oligomerization (Truebestein and Leonard, 2016). The primary sequence of the Afadin-CC region is highly charged and does not maintain this pattern of hydrophobic residues (Figure 1F), suggesting that it may not be a bona fide CC. To test whether the Afadin-CC forms higher oligomers, the GST tag was removed (Figure 2—figure supplement 1) and the Afadin-CC alone was analyzed by size-exclusion chromatography coupled to multi-angle light scattering (SEC-MALS). This technique allows for the direct measurement of the molar mass of a particle in solution and demonstrated that the Afadin-CC is a monomer, with a molecular mass of ~8 kDa (predicted mass = 8.3 kDa, Figure 2A). The chromatography trace also revealed a small amount of truncated Afadin-CC, suggesting that following removal of the GST tag this region may be prone to some degradation. To further characterize the interaction of the PTPRK-ICD with Afadin-CC, isothermal titration calorimetry (ITC) experiments were performed to measure the binding affinity and stoichiometry. Afadin-CC (without the GST tag) forms an equimolar complex with PTPRK-ICD with a binding affinity in the low micromolar range (KD = 3.2 ± 0.6 µM, Figure 2B and E). This binding affinity is consistent with a reversible, transient protein–protein interaction as expected for an enzyme-substrate complex. The high ΔH of this interaction (–17.6 ± 0.7 kcal/mol) indicates it is enthalpically driven, consistent with the highly charged nature of the Afadin-CC region.

Figure 2 with 1 supplement see all
PTPRK:Afadin-CC forms an equimolar complex with micromolar affinity.

(A) Size-exclusion chromatography coupled to multi-angle light scattering (SEC-MALS) analysis of Afadin-CC. The SEC elution profile (normalized differential refractive index [dRI]; thin red line) and weight-averaged molecular mass (red thick line) are shown. The dashed horizontal line indicates the predicted mass of monomeric Afadin-CC after removal of the GST affinity tag. (B) Isothermal titration calorimetry (ITC) titration curves of the interaction between Afadin-CC and PTPRK-ICD. Left: baseline-corrected differential power (DP) plotted over time. Right: normalized binding curve showing the integrated change in enthalpy against the molar ratio. (C) PTPRK pull-downs using streptavidin bead-conjugated ICD, D1 or D2 domains against GST-Afadin-CC followed by SDS-PAGE and Coomassie staining. Gel is representative of n 3 independent experiments. (D) ITC titration curves of the interaction between Afadin-CC and PTPRK-D2. Data presented as described for (B). (E) Table showing the dissociation constant (KD), enthalpy (ΔH), and number of binding sites (N) for the ITC experiments performed in this study. Data represents the mean ± SEM of n = 2 independent experiments.

Previous work highlights a critical role for the D2 pseudophosphatase domain in the specific interaction of Afadin with R2B family receptors (Fearnley et al., 2019; Hay et al., 2020). To investigate the contribution of the D2 domain to Afadin binding, pull-downs against Afadin-CC were performed using biotinylated PTPRK ICD, D1 and D2 domains. The PTPRK-D2 domain alone was sufficient to bind Afadin-CC to similar levels observed for the full ICD (Figure 2C). No interaction of Afadin-CC with the PTPRK-D1 domain was observed, highlighting that the D2 domain is necessary and sufficient for the interaction with this part of Afadin. To confirm this, ITC experiments were performed monitoring the binding of the Afadin-CC to the PTPRK-D2 domain. These data produced very similar results to those for the full ICD, specifically the Afadin-CC forms an equimolar complex with the PTPRK-D2 domain with a binding affinity in the low micromolar range (KD = 6.3 ± 0.3 µM, Figure 2D and E). These data suggest there is only a very modest (twofold) reduction in binding affinity for the D2 domain when compared with the ICD, consistent with the D2 domain being the primary Afadin binding surface.

Structure prediction of the PTPRK–Afadin interaction

Having mapped the regions of Afadin and PTPRK required for binding, we sought to understand the structural basis of the interaction, but extensive efforts to crystallize the Afadin-CC:PTPRK-D2 complex were unsuccessful. Therefore, we exploited recent advances in deep learning neural network techniques by using AlphaFold2 (AF2)-Multimer to model the complex based on the truncation mapping data. Initial structural predictions were performed using the PTPRK-D2 domain (aa. 1154–1446) with Afadin-CC (aa. 1393–1455) as inputs. The AF2 model of PTPRK-D2 in complex with the full Afadin-CC fragment used for biochemical characterization highlighted that many residues at both the N- and C-termini of the Afadin-CC may not be contributing to the putative interaction interface (Figure 3—figure supplement 1). Therefore, further mapping experiments to refine the binding region of the Afadin-CC were conducted. Both N- and C-terminal truncations of GST-Afadin-CC were expressed, purified, and tested in pull-downs for their ability to bind PTPRK (Figure 3A). For N-terminal truncations of Afadin-CC, residues 1393–1400, which are predicted to be unstructured and outside the core helix (Figure 3—figure supplement 1), were replaced with an equivalent glycine/serine linker sequence to avoid the possibility of steric hindrance with the N-terminal GST-tag during mapping experiments (Figure 3A). In agreement with the initial AF2 model, regions at both the N- and C-termini of Afadin-CC were dispensable for the interaction, while disruption of the core sequence encompassing the charged residues (aa. 1408–1448) resulted in a lack of detectable binding (Figure 3A). Interestingly, this core charged region of Afadin is almost completely conserved across vertebrates and diverges in Drosophila melanogaster, which lacks a full-length PTPRK ortholog (Figure 3—figure supplement 2; Hatzihristidis et al., 2015).

Figure 3 with 3 supplements see all
Structural prediction of the PTPRK-D2:Afadin-CC complex.

(A) Top: schematic of different GST-Afadin-CC truncations used for further interaction mapping experiments. Regions within the predicted helix are marked by a green bar. Dashed lines indicate deleted regions in N-terminal truncations. Bottom: pull-downs using streptavidin bead-conjugated PTPRK-ICD against GST-Afadin-CC truncations followed by SDS-PAGE and Coomassie staining. (B) The top model generated by AF2-Multimer of the PTPRK-D2 domain (blue) in complex with Afadin-CC (green). (C) Prediction quality analysis for the top PTPRK-D2:Afadin-CC complex model. Top: plot of predicted local distance difference test (pLDDT) for Afadin-CC (green) and PTPRK-D2 (blue). Bottom: predicted aligned error (PAE) plot for the PTPRK-D2:Afadin-CC complex. Quality analyses for all five generated models are available in Figure 3—figure supplement 3. (D) PTPRK pull-downs using streptavidin bead-conjugated PTPRK-ICD against GST-Afadin-CC point mutants followed by SDS-PAGE and Coomassie staining. Residue numbering of the Afadin-CC sequence is shown below. Residues that were mutated are in bold underline, with mutations that alter Afadin-CC binding highlighted in yellow. (E) Molecular surface representation of PTPRK-D2 (blue) in complex with Afadin-CC (green ribbons). The sidechains of Afadin-CC residues that were mutated in (D) are shown in stick representation, with residues that were shown to be critical for PTPRK binding highlighted in yellow. Gels shown in this figure are representative of n 3 independent experiments.

Based on these additional mapping data, a new AF2 model was generated using the PTPRK-D2 domain with Afadin-CC minimal region (aa. 1408–1448) as inputs. Of the five models calculated, the top four had the Afadin-CC placed in the same position on the surface of the PTPRK-D2 domain (Figure 3—figure supplement 3), strongly suggesting that the CC domain adopts a linear alpha-helical structure that stretches across one face of the D2 domain (Figure 3B). The top AF2 model (ipTM + pTM score = 0.58) predicts the overall fold of each domain with high confidence and their relative orientations with moderate to high confidence (Figure 3C). To test the validity of this model and further explore the residues involved in Afadin binding to PTPRK, a series of single and double mutations were introduced into this region of the Afadin-CC domain. Mutations of Afadin used in these pull-down assays included residues predicted to be at the interaction interface as well as residues away from the interface (Figure 3D and E). Double alanine mutation at Afadin residues W1418-Y1419 or R1432-K1433, which contribute large sidechains to the interaction interface, abolishes binding to PTPRK, while E1429-R1430 to alanine reduces but does not abolish binding. Importantly, additional mutations to residues distal from the interface (R1410, Q1443, T1446) do not interfere with binding to PTPRK. These mutational data strongly support the validity of the predicted structure for how Afadin binds to the PTPRK D2 domain.

Substrate specificity of PTPRK recruitment and dephosphorylation of Afadin

Our previous work demonstrated that both PTPRK and PTPRU bind Afadin while PTPRM cannot (Fearnley et al., 2019; Hay et al., 2020). Therefore, residues on the D2 domain that are critical for Afadin binding should be conserved in PTPRK and PTPRU but divergent in PTPRM. Based on a multiple-sequence alignment of PTPRK, PTPRU, and PTPRM (Figure 4—figure supplement 1), the positions of unique PTPRM residues were mapped onto the AF2 model of PTPRK-D2 (Figure 4A, the predicted position of the D1 domain is illustrated for clarity and orientation). Interestingly, a cluster of four residues mapped to a region of the PTPRK-D2 surface that is predicted to be the binding site for Afadin-CC (Figure 4A) and are located adjacent to several Afadin-CC residues that we have shown to be necessary for complex formation (Figure 3D and E). This binding site is located on the opposite face relative to the cysteine of the would-be ‘active’ site in the PTPRK-D2 domain (Figure 4A). Although these residues of PTPRK (Y1270, G1273, E1332, and L1335) are not all highly charged, within the context of the PTPRK model they form an acidic pocket into which the Afadin-CC is predicted to bind (Figure 4B and C). To investigate potential molecular differences at this site, a PTPRM-D2 (aa. 1160–1452) AF2 model was generated to allow comparison to PTPRK-D2. Strikingly, in a model of the PTPRM-D2 the altered residues in this region result in a different surface topology and charge, having lost the acidic pocket observed for PTPRK-D2 (Figure 4B and C). These changes in surface properties may therefore account for the differential binding to Afadin.

Figure 4 with 4 supplements see all
PTPRK binds Afadin-CC via an acidic pocket distal from the D2 ‘active’ site.

(A) Conservation mapping of the PTPRK-D2:Afadin-CC interface. PTPRK-D2 is shown (blue surface representation) in complex with Afadin-CC (green ribbons). Residues that are conserved in both PTPRK/PTPRU but not PTPRM are potentially involved in Afadin binding specificity and are highlighted on the PTPRK-D2 surface in red. Cα atoms of Afadin-CC residues identified as critical for PTPRK binding (see Figure 3D and E) are highlighted by yellow spheres. For clarity and orientation, the D1 domain of PTPRK has been modeled in transparent surface representation (white) and both D1 and D2 ‘active’ sites highlighted in purple and dotted circles. Three orientations rotated by 90° are shown. (B) Inset shows the molecular detail of key residues at the highlighted region on the PTPRK-D2 surface. Equivalent residues for the highlighted region are shown for PTPRK (blue) and PTPRM (wheat) in stick representation. (C) Electrostatic properties of PTPRK (left) and PTPRM (right) D2 domains, colored by electrostatic potential (–5 to +5 kT, as red and blue, respectively). Domains are oriented as shown in (B). The key PTPRM substitutions highlighted in (B) result in altered surface topology and electrostatic potentials compared to PTPRK. (D) Pull-downs using streptavidin bead-conjugated WT, G1273H, or L1335R PTPRK-D2 domains against GST-Afadin-CC followed by SDS-PAGE and Coomassie staining. (E) Pull-downs using streptavidin bead-conjugated WT and G1273H/L1335R double mutant (DM) PTPRK-ICD and D2 domains against GST-Afadin-CC followed by SDS-PAGE and Coomassie staining. (F) Isothermal titration calorimetry (ITC) data showing a lack of interaction between Afadin-CC and PTPRK-D2 DM. Left: baseline-corrected differential power (DP) plotted over time. Right: normalized binding curve showing the integrated change in enthalpy against the molar ratio. To highlight lack of binding, DM data (black) is shown superimposed onto the data for the WT D2 domain (gray), which is also shown in Figure 2D. (G) Immunoblot analysis of streptavidin bead-conjugated PTPRK pull-downs from wheat germ lysate containing full-length Afadin. Both WT and DM PTPRK-ICD and D2 domains were assayed for their ability to bind full-length Afadin. (H) GST pull-downs using GST-Afadin-CC against PTPRK-ICD and D2 WT, DM, F1225A, and M-loop, followed by SDS-PAGE and Coomassie staining. Gels and blots shown in this figure are representative of n 3 independent experiments.

To further test the validity of the predicted PTPRK-D2:Afadin-CC complex, and to better understand the binding specificity, residues in the PTPRK-D2 acidic pocket were mutated to their PTPRM equivalents. Specifically, PTPRK-G1273 was mutated to histidine (G1273H) and L1335 to arginine (L1335R). These changes to PTPRM-equivalent residues are predicted to cause steric hindrance (G1273H) or introduce a basic residue into the acidic pocket (L1335R). Each of these single-point mutations were sufficient to abolish the interaction of the PTPRK-D2 with the GST-Afadin-CC (Figure 4D). Furthermore, the introduction of both mutations (double mutant [DM]) into the PTPRK-D2 or full PTPRK-ICD prevents binding to Afadin-CC (Figure 4E). To quantify the loss in affinity due to these point mutations, ITC binding experiments were repeated using the PTPRK-D2-DM with Afadin-CC and no binding was detected, confirming loss of the interaction (Figure 4F). To further confirm that these mutations could interfere with Afadin binding in the context of the full-length protein, wheat germ lysate pull-downs were repeated with WT and DM PTPRK constructs. While PTPRK-ICD and PTPRK-D2 bind Afadin, introduction of these two mutations into PTPRK reduces Afadin binding to background levels (Figure 4G). Two additional mutations were introduced into the PTPRK-D2 domain to test the predicted complex model. The first, F1225A, alters the PTPRK residue that is predicted to interact with the W1418-Y1419 residues in Afadin that were essential for binding (Figure 3D, Figure 4—figure supplement 2A). This single-point mutation (F1225A) in PTPRK-D2 vastly reduces binding to the Afadin-CC domain (Figure 4H). This residue is conserved across the R2B family so although it contributes to the binding affinity and validates the complex model, it does not contribute to specificity. Another mutation (labeled the ‘M-loop’) was introduced at the edge of the interface, changing residues CEE in PTPRK (aa. 1372–1374) to the PTPRM-equivalent YNG (aa. 1378–1380) (Figure 4—figure supplement 2B). This mutation retains binding, suggesting that this loop does not contribute to specificity of Afadin binding to PTPRK versus PTPRM (Figure 4H). The cysteine residue of the M-loop is distal to Afadin in the predicted PTPRK complex and the EE to NG substitution would not introduce significant steric hinderance or charge repulsion, consistent with retained binding. For all mutant forms of the PTPRK-D2 domain, correct folding was confirmed by monitoring their thermal stability and comparing melting temperatures (Tm) to wildtype PTPRK-D2 (Figure 4—figure supplement 3). Wildtype and mutant PTPRK-D2 domains possessed Tm values between 39 and 41°C, supporting that any loss of binding was not driven by misfolding. These data strongly validate the predicted model for the PTPRK-D2:Afadin-CC complex and identify two key residues within PTPRK D2 that are crucial for binding to full-length Afadin and, based on their lack of conservation in PTPRM, are likely to be important for determining binding specificity.

To determine whether the PTPRK-D2:Afadin-CC interaction is necessary for PTPRK-mediated dephosphorylation of Afadin, dephosphorylation assays were performed using both WT and DM PTPRK-ICDs. Since we and others have observed allosteric modulation of RPTP D1 domains by D2 domains (Fearnley et al., 2019; Wen et al., 2020), we tested whether introducing D2 domain mutations affects PTPRK-ICD phosphatase activity. The WT and DM PTPRK-ICDs were compared in activity assays using the generic phosphatase substrate 4-nitrophenyl phosphate (pNPP), which confirmed that both proteins have identical activity profiles (Figure 5A). Dephosphorylation assays were then performed using PTPRK-ICD WT and DM proteins incubated with quenched, pervanadate-treated lysates. Afadin was dephosphorylated at Y1230 by WT PTPRK-ICD, but not PTPRK-ICD DM (Figure 5B, Figure 5—figure supplement 1). Comparatively, another PTPRK substrate, p120-Catenin, which has been shown previously to be dephosphorylated efficiently by the PTPRK-D1 domain alone, was dephosphorylated by both proteins – independent of D2 domain mutation status. Results using our Afadin-pY1230 antibody were also confirmed by immunoprecipitation (IP) of pTyr proteins from WT and DM dephosphorylation assays (Figure 5C). Afadin was depleted in pTyr-IPs from lysates treated with WT PTPRK-ICD, showing dephosphorylation had occurred, while the levels of phosphorylated Afadin remained unchanged in PTPRK-ICD DM-treated samples (Figure 5C). Again, both WT and DM PTPRK were equivalently able to dephosphorylate p120-Catenin. These data show that D2 domain recruitment of Afadin is critical for efficient dephosphorylation by the D1 domain, and that PTPRK uses distinct surfaces to recruit different substrates for dephosphorylation.

Figure 5 with 1 supplement see all
Mutation of residues at the interaction interface inhibit dephosphorylation of Afadin by PTPRK.

(A) Time course of pNPP dephosphorylation by WT (black) and double mutant (DM, red) PTPRK-ICD. Error bars represent ± SEM of n = 2 independent experiments. (B) Immunoblot analysis of pervanadate-treated MCF10A lysates incubated for 1.5 hr at 4°C with 0.3 µM of PTPRK-ICD WT or DM. (C) Immunoblot analysis of pervanadate-treated MCF10A lysates incubated for 1.5 hr at 4°C with 0.3 µM of PTPRK-ICD WT or DM followed by pTyr immunoprecipitation (IP). Dephosphorylated proteins are depleted from pTyr-IPs and/or enriched in supernatants (sup.). Gels and blots shown in this figure are representative of n 3 independent experiments.

Discussion

Here we demonstrate direct and specific binding of the substrate Afadin to the receptor phosphatase PTPRK. This interaction is mediated via a single helix within an extended disordered region of Afadin that binds directly to the pseudophosphatase D2 domain of PTPRK. Experimental mapping of the binding interface involved truncation mapping of Afadin using different protein production strategies. The cell-free wheat germ-based system allowed for the efficient testing of large Afadin constructs and the rapid mapping down to a region more compatible with large-scale cell-based protein production, such as E. coli. This additional mapping allowed for the identification of relevant, smaller regions and domains of both Afadin and PTPRK. By gaining this information experimentally, it was then possible to exploit deep learning techniques via AF2 to make reliable predictions of the interaction interface. Interestingly, if larger regions of either protein, such as the full ICD of PTPRK or longer segments of Afadin, are used for AF2-based complex predictions, a range of different assemblies are predicted with lower confidence (Figure 4—figure supplement 4). Therefore, at present, these predictive tools are at their most valuable when coupled to experimental approaches that identify the relevant protein fragments to be used as inputs. When this requirement is satisfied, the power of these predictions is clearly demonstrated here by the identification of a testable interface that proved to be highly accurate in informing subsequent experimental assays.

SEC-MALS analysis shows the purified Afadin-CC region that binds PTPRK to be monomeric in solution (Figure 2A), demonstrating that binding to PTPRK does not compete with potential dimerization of Afadin via this region. The CC region of Afadin that is shown here to be critical for binding to PTPRK is distal to the tyrosine residue that is targeted for dephosphorylation (Y1230). We also demonstrate that the interaction with the CC region is critical for efficient dephosphorylation, showing that this interaction is important for bringing Afadin within close proximity to PTPRK but does not directly position the relevant pTyr in the active site of the PTPRK-D1 domain. This role for the D2 pseudophosphatase domain in specific substrate recruitment helps to explain the apparent lack of specificity of phosphatases for phosphopeptide substrates, indicating that it is not necessarily a motif near the pTyr that determines substrate targeting.

Our previous observation that Afadin binding is independent of its phosphorylation status (Fearnley et al., 2019) is consistent with the data presented here. Specifically, Afadin binds PTPRK via a site that is over 100 amino acids downstream of the target phosphosite and the interaction interface is distinct from the nonfunctional ‘active site’ on the D2 domain. These data support that the pseudophosphatase domain has been repurposed in evolution for specific protein–protein interactions. However, it is also possible that even active PTP domains mediate such interactions. Additional substrates, such as PARD3, showed some dependence on the D2 for dephosphorylation by PTPRK and PTPRM, whereas other substrates of PTPRK, such as p120Cat, do not require the D2 domain for recruitment (Fearnley et al., 2019). This raises interesting questions about simultaneous binding of substrates to PTPRK and potential competition for access to the active site. Furthermore, if the phosphorylation state is not critical for binding, PTPRK might play a scaffolding role connecting multiple junctional components as well as F-actin. Consistent with a scaffolding role for receptor PTPs, structural studies on the R2A receptor PTPRF revealed an interaction interface with Liprin-α spanning both the D1 and D2 domains, adding further complexity to potential binding modes (Xie et al., 2020). Interestingly, Liprin is not a PTPRF substrate, thus RPTPs could form multiprotein complexes independent of protein phosphorylation status. This might also help to explain reported phosphatase-independent functions of RPTPs (Juettner et al., 2019). Finally, the region of Afadin that interacts with PTPRK has been shown to also mediate binding to αE-catenin (Maruo et al., 2018), a core component of the E cadherin-catenin complex (Niessen and Gottardi, 2008). Further studies are required to understand how RPTPs regulate cell adhesion while recruiting core adhesion molecules, such as Afadin, potentially in competition with canonical adhesion complexes.

Methods

Plasmids and constructs

Amino acid (aa) numbering used throughout is based on the following sequences: PTPRK, UniProt ID: Q15262-3; PTPRM, UniProt ID: P28827-1; Afadin, UniProt ID: P55196-4. All point mutations were generated by site-directed mutagenesis using polymerase chain reaction with Phusion Hot Start II DNA polymerase (Thermo Fisher Scientific). The pET-15b.His.TEV.AviTag expression vectors used for the generation of biotinylated PTPRK-ICD (aa. 864–1439), D1 (aa. 865–1157), and D2 (aa. 1154–1446) domains were generated in a previous study (Fearnley et al., 2019). PTPRK-ICD and D2 G1273/L1335R single and double mutants (DM) were generated in these His.TEV.AviTag vectors. For ITC experiments, PTPRK-ICD, D2, and D2-DM were subcloned into a pET-15b expression vector in frame with an N-terminal His6-tag only. For wheat germ cell-free expression, the full-length human Afadin cDNA and truncations (outlined in Figure 1C) were subcloned into the pF3A WG (BYDV) vector (Promega) with a modified multiple cloning site (SpeI-AvrII-EcoRI-NheI-SacI-KpnI downstream of an N-terminal Myc-epitope; pF3A-WG-Myc). For bacterial expression, Afadin C-terminal fragments (outlined in Figure 1D) were subcloned into the pGEX-6P-1 vector with an in-frame N-terminal GST tag followed by the human rhinovirus 3C protease recognition sequence. For transient overexpression experiments, full-length Afadin was subcloned into pmScarlet-C1 (gift from Z. Kadlecova) and mutated to generate Y1226F and Y1230F single and double (YFYF) mutants.

Antibodies

All antibodies, except Afadin-pY1230, were used at a 1:1000 dilution in TBS-T (20 mM Tris–HCl, pH 7.4, 137 mM NaCl, 0.2% [v/v] Tween-20) with 3% (w/v) bovine serum albumin (BSA) unless otherwise noted. Rabbit anti-Myc, rabbit anti-Afadin, mouse anti-His-tag, rabbit anti-pTyr, rabbit anti-phospho-p120-Catenin (Y228 and Y904), rabbit anti-paxillin, and rabbit anti-phospho-paxillin (Y118) primary antibodies were purchased from Cell Signaling Technologies. Mouse anti-p120-Catenin and mouse anti-Afadin primary antibodies were purchased from BD Transduction Laboratories. Mouse anti-α-tubulin primary antibody was purchased from Sigma-Aldrich. Mouse anti-RFP antibody was purchased from Thermo Fisher Scientific. Anti-mouse, anti-goat, and anti-rabbit HRP-conjugated secondary antibodies (1:5000 in TBS-T) were purchased from Jackson ImmunoResearch. The rabbit anti-PTPRK antibody was generated previously (Fearnley et al., 2019).

The Afadin pTyr-1230 antibody was generated by MRCPPU Reagents and Services, University of Dundee (https://mrcppureagents.dundee.ac.uk/) using a phosphorylated Afadin-Y1230 peptide (CTYTRE[pY]FTFPA) as the antigen. Briefly, the N-terminal Cys residue was used to conjugate the peptide to both keyhole limpet hemocyanin (KLH) and BSA. Sheep were immunized with the antigen followed by four further injections 28 days apart, with bleeds performed 7 days after each injection. Antibodies were affinity purified from serum using the Afadin-Y1230 phospho-peptide. For Western blotting, the antibody was used at 1 μg/ml in TBS-T with 3% milk, after pre-incubation with 10 μg/ml non-phosphorylated peptide (TYTREYFTFPA) for 30 min at room temperature (RT) or 4°C.

Protein expression and purification

Expression plasmids were transformed into competent E. coli BL21 (DE3) Rosetta cells and cultured in 2× TY medium supplemented with 50 µg/ml ampicillin and 34 µg/ml chloramphenicol at 37°C and 220 RPM to an optical density at 600 nm of 0.6. Cultures were then equilibrated to 20°C and protein expression induced with addition of 1 mM isopropyl-thio-β-d-galactopyranoside for 18 hr. For in vivo biotinylation of Avi-tagged proteins, cultures were supplemented with 200 µM D-biotin (Sigma-Aldrich) at the point of induction. Bacterial cultures were then harvested by centrifugation (4000 × g, 20 min) and cell pellets stored at –20°C. Bacterial cell pellets were thawed and resuspended in ice-cold lysis buffer (50 mM Tris–HCl, pH 7.4, 500 mM NaCl, 0.5 mM TCEP, 5% [v/v] glycerol, 1× EDTA-free protease inhibitors) prior to lysis by high-pressure disruption at 25 kpsi (Constant Systems Ltd). Cell lysates were then clarified by centrifugation (40,000 × g, 4°C, 30 min).

For GST-Afadin constructs, clarified cell lysates were incubated with washed glutathione sepharose 4B beads (Cytiva) for 1 hr at 4°C. Beads were packed into a gravity column and washed with 20 ml of GST wash buffer (50 mM Tris–HCl, pH 7.4, 500 mM NaCl, 0.5 mM TCEP) and protein eluted in 5 × 1 ml fractions of GST wash buffer supplemented with 50 mM reduced glutathione. Eluted proteins were further purified by SEC or dialysis. SEC was performed using a HiLoad Superdex 200 pg 16/600 column equilibrated in 50 mM Tris–HCl, 150 mM NaCl, 5% [v/v] glycerol, 5 mM DTT. Dialysis was performed using a Slide-A-Lyzer 20K MWCO cassette (Thermo Fisher Scientific) in SEC buffer (50 mM Tris–HCl pH 7.4, 150 mM NaCl, 5% [v/v] glycerol, 5 mM DTT) for 2 hr at RT, followed by overnight at 4°C, and another 2 hr with fresh SEC buffer to remove GSH. The purified protein was recovered from the cassettes and immediately used for streptavidin pull-downs.

For His6-tagged PTP constructs, clarified cell lysates were incubated with washed Ni-NTA agarose beads (QIAGEN) for 1 hr at 4°C. Beads were packed into a gravity column and washed with 10 ml of Ni-NTA wash buffer (50 mM Tris–HCl, pH 7.4, 500 mM NaCl, 5% [v/v] glycerol, 0.5 mM TCEP) with 10 mM imidazole, followed by 10 ml of Ni-NTA wash buffer with 20 mM imidazole. Protein was eluted in 6 × 0.5 ml fractions of elution buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 5% [v/v] glycerol, 0.5 mM TCEP, 200 mM imidazole). For His6-tag-only constructs used in ITC experiments, proteins were further purified by anion-exchange chromatography (AEX). For PTPRK-D2, eluted Ni-NTA fractions were pooled and then diluted to low-salt/imidazole concentration by 1:10 dilution with dilution buffer (50 mM Tris–HCl, pH 7.4, 5% [v/v] glycerol, 5 mM DTT). The diluted sample was applied to a MonoQ 5/50 GL AEX column (Cytiva), equilibrated in low-salt buffer (50 mM Tris–HCl, pH 7.4, 10 mM NaCl, 5% [v/v] glycerol, 5 mM DTT) and bound protein eluted using a linear gradient with high-salt buffer (50 mM Tris–HCl, pH 7.4, 1 M NaCl, 5% [v/v] glycerol, 5 mM DTT). For PTPRK-ICD, AEX purification was performed as above, but with all buffers adjusted to pH 7.0.

For biotinylated Avi-tag constructs for use in pull-down assays, eluted Ni-NTA fractions were purified by SEC on a HiLoad Superdex 75 (D1 and D2 domains) or 200 (ICDs) pg 16/600 column equilibrated in 50 mM Tris, pH 7.4, 150 mM NaCl, 5% (v/v) glycerol, 5 mM DTT. Routinely, purity of all constructs was assessed by SDS-PAGE followed by staining with Coomassie.

Assessment of recombinant protein biotinylation

To assess biotinylation, 10 µg of in vivo biotinylated protein was solubilized in an appropriate volume of 5× SDS loading buffer and heated to 95°C for 5 min. Samples were allowed to cool to RT prior to addition of a threefold molar excess of streptavidin (IBA Lifesciences). Samples were incubated for 10 min at RT, then immediately analyzed by SDS-PAGE followed by Coomassie staining. Approximate levels of biotinylation were calculated by 2D densitometry of protein band depletion upon addition of streptavidin using Fiji (Schindelin et al., 2012).

pNPP phosphatase activity assay

Recombinant PTP domains were made up to 500 µl in assay buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 5% [v/v] glycerol, 5 mM DTT) at 0.6 µM. PTP domains and 20 mM pNPP substrate (in assay buffer) were equilibrated to 30°C for 15 min in an orbital shaking heat block at 500 RPM. To initiate reactions, 500 µl of 20 mM pNPP substrate was added to PTP containing tubes (0.3 µM PTP and 10 mM pNPP final concentrations). Reactions were carried out for 30 min at 30°C in an orbital shaking heat block at 500 RPM. At each time point (0, 2, 5, 10, 20, and 30 min), 100 µl of the total reaction was transferred to a 96-well microplate well containing 50 µl 0.58 M NaOH, terminating the reaction. After the final time point, absorbance of each sample was measured at 405 nm in a Spectramax M5 plate reader (Molecular Devices). Product formation was calculated by interpolation of absorbance values using a 4-nitrophenol standard curve of known concentration.

Protein pull-downs using wheat germ cell-free expression

Wheat germ reactions were performed using the TnT SP6 High-Yield Wheat Germ Protein Expression System (Promega). On ice, 3 µg of the relevant WG expression plasmid in a total volume of 20 µl ddH2O was combined with 30 µl of freshly thawed wheat germ mastermix. Reactions were carried out for 2 hr at 25°C in a thermocycler.

For wheat germ pull-down assays, 0.5 nmol recombinant in vivo biotinylated PTP domain (36 or 18 µg for ICD and D1/D2 domains, respectively) was bound to 125 µl streptavidin-coated magnetic beads in a total volume of 1 ml of purification buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 5% [v/v] glycerol) for 1 hr at 4°C with rotation. Beads were collected on a magnetic stand and washed twice with purification buffer, followed by two washes in pull-down buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 10% [v/v] glycerol, 1% [v/v] Triton X-100). Wheat germ reactions were mixed thoroughly and diluted to a total volume of 200 µl in pull-down buffer. Pull-downs were carried out with 100 µl diluted wheat germ reaction combined with protein-conjugated streptavidin beads in a total volume of 200 µl (approximately 2.5 µM final bait concentration) pull-down buffer in 96-well microplate wells, for 1 hr at RT on a high-speed orbital shaking platform, 700 rpm. Magnetic beads were collected on a 96-well magnetic stand and supernatants removed. To wash, beads were resuspended in 200 µl of pull-down buffer containing 250 mM NaCl followed by shaking for 1 min at 700 rpm. This step was repeated for a total of four washes. After the final wash, samples were transferred to 1.5 ml tubes on a vertical magnetic stand. Beads were rinsed once with 1 ml TBS, then resuspended in 80 µl TBS with 20 µl 5× SDS sample buffer supplemented with 7 mM biotin. Pull-downs were eluted by incubation at 95°C for 10 min, beads separated on a magnetic stand and supernatants collected. Routinely, 50 µl of the final 100 µl sample were analyzed by SDS-PAGE followed by immunoblotting.

Recombinant protein pull-downs

For PTPRK pull-downs from GST-fusion preys, recombinant in vivo biotinylated PTP domains were bound to streptavidin magnetic beads as described for wheat germ-based pull-downs (see above). Pull-downs were performed using 0.5 nmol PTPRK bait (1 µM) combined with 1 nmol (2 µM) Afadin GST-fusions in a total volume of 500 µl of pull-down buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 10% [v/v] glycerol, 1% [v/v] Triton X-100), incubated for 1 hr with end-over-end rotation at 4°C. Beads were collected on a magnetic stand and supernatants removed. Pull-downs were washed by thorough resuspension in 500 µl pull-down buffer for a total of four washes. Beads were rinsed once with 1 ml TBS, followed by resuspension in 80 µl TBS with 20 µl 5× SDS sample buffer supplemented with 7 mM biotin. Pull-downs were eluted by incubation at 95°C for 10 min, beads separated on a magnetic stand and supernatants collected. Routinely, 50 µl of the final 100 µl sample were analyzed and visualized by Coomassie staining.

For GST-Afadin pull-downs from PTPRK preys, 1 nmol of GST-Afadin-CC was bound to 30 µl glutathione sepharose beads for 1 hr at 4°C with rotation. Bead-bound GST-Afadin-CC was combined with 0.5 nmol PTPRK proteins in a total volume of 500 µl of pull-down buffer (2 µM GST-Afadin, 1 µM PTPRK final concentration) and pull-downs incubated for 1 hr at 4°C with end-over-end rotation. Beads were collected by brief centrifugation and pull-downs washed by thorough resuspension in 500 µl of pull-down buffer. After a total of four washes, pull-downs were eluted by boiling at 95°C in 1× SDS sample buffer in TBS.

AlphaFold2-Multimer structure predictions

All AF2 models were generated using default parameters and run via a locally installed version of AF2-Multimer (version 2.1.0; Evans et al., 2022; Jumper et al., 2021). All models and associated statistics have been deposited in the University of Cambridge Data Repository (https://doi.org/10.17863/CAM.82741). Graphical figures were rendered in PyMOL (Schrödinger LLC).

Size-exclusion chromatography coupled to multi-angle light scattering (SEC-MALS)

For SEC-MALS analysis, the GST-tag was removed from Afadin-CC by PreScission (3C) protease cleavage, using 30 µg GST-3C per 1 mg GST-Afadin-CC in 50 mM Tris, pH 7.4, 150 mM NaCl, 5% (v/v) glycerol, 5 mM DTT, 0.5 mM EDTA. Cleavage reactions were performed for 16 hr at 4°C, followed by removal of GST-3C and cleaved GST by reabsorption onto glutathione sepharose beads for 1 hr at 4°C. Cleavage reactions were then passed through a gravity column to separate sepharose beads from eluate containing the cleaved Afadin-CC. Afadin-CC was immediately loaded onto a HiLoad Superdex 75 pg 16/600 column (Cytiva) equilibrated in SEC buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 5% [v/v] glycerol, 5 mM DTT) and peak fractions concentrated in 3K MWCO centrifugal filter unit (Merck Millipore) to 1 mg/ml (120 µM).

A 100 µl sample at 1 mg/ml was injected onto a Superdex 75 Increase 10/300 GL column (Cytiva) equilibrated in SEC buffer at a flow rate of 0.5 ml/min. Data for static light scattering and differential refractive index were measured in-line using DAWN 8+ and Optilab T-rEX detectors, respectively (both Wyatt Technology). The absolute molar masses of the elution peaks were calculated in ASTRA 6 (Wyatt Technology) using a protein dn/dc value of 0.185 ml/g.

Isothermal titration calorimetry (ITC)

For ITC experiments, the GST-tag was removed from Afadin-CC as described for SEC-MALS experiments (see above), with final SEC purification into ITC buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 0.5 mM TCEP). PTP domains were exchanged into ITC buffer by SEC using a HiLoad Superdex 200 pg 16/600 (PTPRK-ICD) or Superdex 75 Increase 10/300 (PTPRK-D2) column. ITC experiments were performed on an automated MicroCal PEAQ-ITC (Malvern Panalytical) at 25°C. Syringe titrant (Afadin-CC) at 260 µM was titrated into 30 µM cell titrate (PTPRK-ICD and D2) as 13 × 3 µl injections. ITC data were analyzed and fit to a one-site binding model using the MicroCal PEAQ-ITC analysis software (Malvern Panalytical).

Differential scanning fluorimetry (DSF)

DSF was performed using Sypro Orange dye (Invitrogen) as per manufacturer’s protocol in a Bio-Rad Real-Time PCR system. Reaction mixes consisting of 5 μg recombinant protein and 1× Sypro Orange dye were made up to a total volume of 20 μl in 1× PBS. Samples were then heated on a 1°C per 10 s gradient from 25 to 95°C and protein unfolding at each temperature monitored by measurement of fluorescence using the FRET channel. Fluorescent signal vs. temperature was fitted to a nonlinear Boltzmann-sigmoidal regression in GraphPad Prism, with the Tm calculated from the inflection point of the fitted curve.

Cells and cell culture

MCF10A cells were obtained from the American Type Culture Collection (ATCC, CRL-10317). HEK293T cells were obtained from D Ron. Cells were maintained in a 37°C humidified 5% CO2 ventilator and passaged every 2–4 days depending on the cell line. MCF10A cells were grown in MCF10A culture medium (1:1 DMEM:Ham’s F12, supplemented with 5% [v/v] horse serum, 20 ng/ml EGF, 0.5 mg/ml hydrocortisone, 100 ng/ml cholera toxin, and 10 mg/ml insulin) (Debnath et al., 2003). PTPRK-KO MCF10A cells and doxycycline-inducible stable cell lines were generated in a previous study (Fearnley et al., 2019) and cultured as for WT cells. HEK293T cells were cultured in DMEM containing 10% (v/v) FBS (Sigma-Aldrich), 2 mM l-glutamine (Sigma-Aldrich). For doxycycline induction of MCF10A cell lines, 2 × 106 were seeded per 10 cm dish and cultured for 6 days, with media changes on days 2 and 4. Doxycycline (Sigma-Aldrich) was included on day 4, 48 hr prior to lysis. Cell lines that were not obtained from commercial sources were subjected to Mycoplasma testing using MycoAlert PLUS (Lonza) or MycoProbe (R&D Systems) Mycoplasma Detection Kits.

Lipid-based transfection of Afadin constructs for overexpression

6 × 105 HEK293T cells were reverse transfected with mScarlet or mScarlet-Afadin constructs using GeneJuice (Merck Millipore #70967-3). For each well of a 6-well plate, 6 µl GeneJuice transfection reagent was incubated with 200 µl serum/antibiotic-free OptiMEM (Thermo Fisher Scientific) for 5 min at RT, then combined with 2 µg of plasmid DNA for a further 15 min at RT prior to addition to cells in full growth media. After 4 hr, cells underwent a media change.

Calf intestinal alkaline phosphatase treatment

HEK293T cells were transiently transfected with indicated mScarlet constructs, pervanadate treated and lysed in RIPA buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1% [v/v] NP-40, 0.5% [w/v] sodium deoxycholate, 1 mM EDTA, 0.2% [w/v] SDS), cOmplete Protease Inhibitor Cocktail (Roche) and phosphatase inhibitor phosSTOP (Roche). Lysates were cleared by centrifugation at 13,000 × g for 15 min at 4°C. Supernatant protein concentration was determined by BCA assay. 50 µg protein was resuspended with 50 U calf intestinal alkaline phosphatase (CIP) and 1× dephosphorylation buffer (Invivogen #18009-019). Samples were incubated for 40 min at 37°C with gentle agitation. The temperature was increased to 65°C for 10 min to inactivate CIP. 1× sample buffer was added to samples prior to SDS-PAGE and immunoblotting.

Lipid-based transfection of siRNA pools

6 × 105 HEK293T cells were reverse transfected with nontargeting or AFADIN siRNA duplexes (ON-TARGETplus SMARTpool, Dharmacon Horizon Discovery) using lipofectamine RNAiMAX as per the manufacturer’s instructions (Thermo Fisher Scientific # 13778030). Briefly, for each well of a 6-well plate, 6 µl RNAiMAX was used to transfect 10 nM siRNA duplexes. After 24 hr, media was replaced with complete growth medium and left to recover for another 24 hr prior to cell treatments and processing for analysis.

Generation of pervanadate-treated lysates

Pervanadate was generated based on Huyer et al., 1997. To avoid decomposition, it was prepared immediately prior to use. In a 1.5 ml tube, 100 µl of 100 mM sodium orthovanadate (Alfa Aesar) was combined with 103 µl 0.49 M H2O2 in 20 mM HEPES, pH 7.3, mixed by gentle inversion and incubated at RT for 5 min. Excess H2O2 was quenched by addition of 23 µl of 0.5 mg/ml catalase in 50 mM potassium phosphate, followed by mixing by gentle inversion. This yields a 44.2 mM pervanadate stock solution.

To generate pervanadate-treated lysates, MCF10A cells were seeded at 2 × 106 in a 10 cm dish and cultured until confluent. Culture medium was then aspirated and cells treated with 8 ml of fresh MCF10A growth medium (see ‘Cells and cell culture’) supplemented with 100 µM pervanadate. Cells were treated for 30 min in a 37°C, 5% CO2 incubator, then moved to ice and washed twice with ice-cold PBS. Each dish was lysed in 600 µl lysis buffer (50  mM Tris–HCl, pH 7.4, 150  mM NaCl, 10% [v/v] glycerol, 1% [v/v] Triton X-100, 1 mM EDTA, 5 mM iodoacetamide [IAA], 1 mM sodium orthovanadate, 10 mM NaF, 1× EDTA-free protease inhibitor) on ice, with periodic agitation, in the dark at 4°C for 30 min. Lysates were scraped, collected into 1.5 ml tubes, and treated with 10 mM DTT on ice for 15 min. Lysates were cleared by centrifugation (14,000 × g, 15 min, 4°C). Supernatants were then collected, snap frozen in liquid nitrogen, and stored at –80°C until use.

SDS-PAGE and immunoblotting

For Coomassie-stained gels, proteins were resolved on 4–12% Bis-Tris NuPAGE, and for Western blotting samples were run on 8 or 10% SDS-PAGE. Protein was transferred to 0.2 µm nitrocellulose membranes by wet transfer at 70 V for 2 hr at 4°C. Membranes were blocked in 5% (w/v) skimmed-milk in TBS-T (20 mM Tris–HCl, pH 7.4, 137 mM NaCl, 0.2% [v/v] Tween-20), prior to incubation with primary antibody overnight at 4°C followed by incubation with HRP-conjugated secondary antibody for 1 hr at RT. Blots were developed using EZ-ECL solution (Geneflow) and imaged using a Bio-Rad ChemiDoc MP imaging system.

In lysate dephosphorylation assays

Recombinant PTP domains (300 nM final concentration) were added to 200 µg of pervanadate-treated MCF10A lysate in a total volume of 400 µl wash buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 10% [v/v] glycerol, 1% [v/v] Triton X-100), supplemented with or without 5 mM DTT. Reactions were incubated with end-over-end rotation for indicated times (0.75–1.5 hr) at 4°C, then either directly analyzed by immunoblotting or subjected to pTyr immunoprecipitation.

For pTyr IP experiments, 400 µl dephosphorylation assays were made up to 0.4% (w/v) SDS, vortexed, and incubated on ice for 5 min to terminate reactions. Reactions were then made up to a total volume of 800 µl (0.2% SDS) in wash buffer and 6 µl of rabbit anti-pTyr antibody added to each sample. Samples were then incubated with end-over-end rotation at 4°C for 3 hr, followed by addition of 40 µl washed protein A agarose. IPs were carried out overnight (>16 hr) with end-over-end rotation at 4°C. Agarose beads were collected by centrifugation (15,000 × g, 30 s) and washed by resuspension in wash buffer. This wash step was repeated for a total of four washes. To elute IPs, beads were resuspended in 80 µl of 2.5× SDS loading buffer (diluted in wash buffer) and boiled for 10 min at 95°C. Supernatants were then collected and immediately analyzed by SDS-PAGE followed by immunoblot.

Protein multiple-sequence alignment (MSA)

MSAs were generated using Clustal Omega (Sievers et al., 2011) and edited using Jalview (Waterhouse et al., 2009).

Appendix 1

Appendix 1—key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (human)PTPRKENSEMBL: ENSG00000152894
Gene (human)AFDNENSEMBL: ENSG00000130396
Cell line (human)MCF10AATCCCRL-10317
Cell line (human)HEK293TD. RonN/A
Transfected
construct (human)
MCF10A PTPRK
KO pooled.tGFP
Fearnley et al., 2019N/ALentivirally transduced stable cell line
Transfected
construct (human)
MCF10A PTPRK KO pooled.tGFP.
P2A.PTPRK
Fearnley et al., 2019N/ALentivirally transduced stable cell line
Transfected
construct (human)
MCF10A PTPRK KO
pooled.tGFP.P2A.
PTPRK.C1089S
Fearnley et al., 2019N/ALentivirally transduced stable cell line
Transfected
construct (human)
MCF10A tGFPFearnley et al., 2019N/ALentivirally transduced stable cell line
AntibodyAnti-PTPRK
(rabbit monoclonal)
Fearnley et al., 20192.H4Western blot: 1:1000
AntibodyAnti-Afadin-pY1230
(sheep polyclonal)
This studyN/ACharacterized in Figure 1—figure supplement 2
Western blot: 1 µg/ml
(with 10 µg/ml
non-phosphopeptide)
Available on request from Sharpe lab, Babraham Institute
AntibodyAnti-Afadin
(mouse monoclonal)
BD Transduction LaboratoriesCat#610732Western blot: 1:1000
AntibodyAnti-p120 catenin
(mouse monoclonal)
BD Transduction LaboratoriesCat#610133Western blot: 1:1000
AntibodyAnti-RFP
(mouse monoclonal)
Thermo Fisher ScientificCat#MA5-15257Western blot: 1:1000
AntibodyAnti-Turbo-GFP
(mouse monoclonal)
OriGeneTA150041Western blot: 1:1000
AntibodyAnti-Afadin
(rabbit monoclonal)
Cell Signaling TechnologyCat#13531Western blot: 1:1000
AntibodyAnti-His (mouse
monoclonal)
Cell Signaling TechnologyCat#2366Western blot: 1:1000
AntibodyAnti-phospho-tyrosine
(P-Tyr-1000) (rabbit
monoclonal)
Cell Signaling TechnologyCat#8954Western blot: 1:2000
AntibodyAnti-paxillin (rabbit monoclonal)Cell Signaling TechnologyCat#12065 (D9G12)Western blot: 1:1000
AntibodyAnti-phospho-p120
catenin (Y904)
(rabbit polyclonal)
Cell Signaling TechnologyCat#2910Western blot: 1:1000
AntibodyAnti-phospho-p120
catenin (Y228)
(rabbit polyclonal)
Cell Signaling TechnologyCat#2911Western blot: 1:1000
AntibodyAnti-phospho-
paxillin (Y118)
Cell Signaling TechnologyCat#2541Western blot: 1:1000
AntibodyAnti-Tubulin (alpha)
(mouse monoclonal)
SigmaCat#T6199Western blot: 1:1000
AntibodyHRP-conjugated-
donkey anti-goat IgG
Jackson ImmunoResearchCat#705-035-147Western blot: 1:5000
AntibodyHRP-conjugated-
donkey anti-rabbit IgG
Jackson ImmunoResearchCat#711-035-152Western blot: 1:5000
AntibodyHRP-conjugated-
donkey anti-mouse IgG
Jackson ImmunoResearchCat#711-035-152Western blot: 1:5000
AntibodyHRP-conjugated-mouse anti-
rabbit IgG (conformation specific)
Cell Signaling TechnologyCat#5127SWestern blot: 1:2000
Recombinant
DNA reagent
pET15bJ. DeaneN/A
Recombinant
DNA reagent
pET15b.His.TEV.AviFearnley et al., 2019N/A
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.ICD
Fearnley et al., 2019UniProt: Q15262-3
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.ICD.D1057A
Fearnley et al., 2019UniProt: Q15262-3
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.ICD.C1089S
Fearnley et al., 2019UniProt: Q15262-3
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D1
Fearnley et al., 2019UniProt: Q15262-3
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D2
Fearnley et al., 2019UniProt: Q15262-3
Recombinant
DNA reagent
PET15b.His.TEV.Avi.
PTPRK.D2.G1273H
This studyUniProt: Q15262-3Mutations: G1273H
See Figure 4
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D2.L1335R
This studyUniProt: Q15262-3Mutations: L1335R

See Figure 4D
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.ICD.DM
This studyUniProt: Q15262-3Mutations: G1273H L1335R
See Figure 4E
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D2.DM
This studyUniProt: Q15262-3Mutations: G1273H L1335R
See Figure 4E
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D2.F1225A
This studyUniProt: Q15262-3Mutation: F1225A
See Figure 4H
Recombinant
DNA reagent
pET15b.His.TEV.Avi.
PTPRK.D2.M-loop
This studyUniProt: Q15262-3Mutations: C1372Y E1373N
E1374G
See Figure 4H
Recombinant
DNA reagent
pGEX-6P-1J. DeaneN/A
Recombinant
DNA reagent
pGEX-6P-Afadin-1098-C*This studyUniProt: P55196-4See Figure 1E
Recombinant
DNA reagent
pGEX-6P-Afadin-1098-1507This studyUniProt: P55196-4See Figure 1E
Recombinant
DNA reagent
pGEX-6P-Afadin-1098-1407This studyUniProt: P55196-4See Figure 1
Recombinant DNA reagentpGEX-6P-Afadin-1514-C*This studyUniProt: P55196-4See Figure 1E
Recombinant DNA reagentpGEX-6P-Afadin-1393-C*This studyUniProt: P55196-4See Figure 1E
Recombinant DNA reagentpGEX-6P-Afadin-1393-1507This studyUniProt: P55196-4See Figure 1E
Recombinant DNA reagentpGEX-6P-Afadin-CCThis studyUniProt: P55196-4Encoding amino acids: 1393–1455

See Figure 1F
Recombinant DNA reagentpGEX-6P-Afadin-CC-WY>AAThis studyUniProt: P55196-4Mutations: W1418A Y1419A
See Figure 3D
Recombinant DNA reagentpGEX-6P-Afadin-CC-ER>AAThis studyUniProt: P55196-4Mutations: E1429A R1430A
See Figure 3D
Recombinant DNA reagentpGEX-6P-Afadin-CC-RK>AAThis studyUniProt: P55196-4Mutations: R1432A K1433A
See Figure 3D
Recombinant DNA reagentpGEX-6P-Afadin-CC-Q>AThis studyUniProt: P55196-4Mutations: Q1443A
See Figure 3D
Recombinant DNA reagentpGEX-6P-Afadin-CC-T>AThis studyUniProt: P55196-4Mutations: T1446A
See Figure 3D
Recombinant DNA reagentpGEX-6P-Afadin-CC-QT>AAThis studyUniProt: P55196-4Mutations: Q1443A T1446A
See Figure 3D
Recombinant DNA reagentpmScarlet-C1Z. KadlecovaN/A
Recombinant DNA reagentpmScarlet-AfadinThis studyUniProt: P55196-4See Figure 1—figure supplement 2D
Recombinant DNA reagentpmScarlet-Afadin-Y1226FThis studyUniProt: P55196-4Mutation: Y1226F
See Figure 1—figure supplement 2D
Recombinant DNA reagentpmScarlet-Afadin-Y1230FThis studyUniProt: P55196-4Mutation: Y1230F
See Figure 1—figure supplement 2D
Recombinant DNA reagentpmScarlet-Afadin-YF YFThis studyUniProt: P55196-4Mutations: Y1226F Y1230F
See Figure 1—figure supplement 2D
Sequence-based reagentON-TARGETplus
Human AFDN siRNA:
Dharmacon, GE HealthcareL-020075-02-0005
Sequence-based reagentON-TARGETplus
Non-targeting pool siRNA:
Dharmacon, GE HealthcareCat#D-001810-10-05
Peptide, recombinant proteinCatalaseSigmaCat#C134514
Peptide, recombinant proteinCholera toxinSigmaCat#C-8052
Peptide, recombinant proteinInsulinSigmaCat#I-1882
Peptide, recombinant proteinEpidermal growth factorPeproTechCat#AF-100-15-1MG
Commercial assay or kitQ5 High-Fidelity
DNA Polymerase
New England BiolabsCat#M0491S
Commercial assay or kitPhusion Hot Start II
DNA polymerase
Thermo Fisher ScientificCat#F549L
Commercial assay or kitEZ-ECL substrateGeneflowCat#K1-0170
Commercial assay or kitInstantBlueExpedeonCat#ISB1L
Commercial assay or kitPhosphatase inhibitor cocktailRocheCat#04906845001
Commercial assay or kitTaqMan Universal Master Mix IIApplied BiosystemsCat#4440040
Chemical compound, drugHydrogen peroxideThermo Fisher ScientificCat#H/1750/15
Chemical compound, drugSodium orthovanadateAlfa AesarCat#J60191
Chemical compound, drug250 kDa-FITC-dextranSigmaCat#FD250S-100MG
Chemical compound, drugPara-Nitrophenol-
phosphate (pNPP)
New England BiolabsCat#P0757
Chemical compound, drugIPTGGeneronCat#GEN-S-02122
Chemical compound, drugD-biotinSigmaCat#B4639
Chemical compound, drugl-GlutamineSigmaCat#G7513
Chemical compound, drugHydrocortisoneSigmaCat#H-0888
Chemical compound, drugNH4OHAcros OrganicsCat#460801000
Chemical compound, drugMethanol-free 16%
(w/v) paraformaldehyde (PFA)
Thermo Fisher ScientificCat#28906
Software, algorithmFIJI/ImageJLaboratory for Optical and
Computational Instrumentation
University of Wisconsin-Madison
Software, algorithmGraphPadPrism
Software, algorithmChimeraUCSF
Strain, strain
background
(Escherichia coli)
STABLE competent E. coliNEBCat#C3040I
Strain, strain
background (E. coli)
DH5alpha competent E. coliInvitrogenCat#18265017
Strain, strain
background (E. coli)
BL21 DE3
Rosetta E. coli
J. DeaneN/A
OtherDMEMThermo Fisher ScientificCat#41965-039Component of cell culture media
OtherHam's F-12SigmaCat#N4888Component of cell culture media
OtherHorse serumThermo Fisher ScientificCat#16050-122Component of cell culture media
OtherHRP-conjugated streptavidinThermo Fisher ScientificCat#434323For detection of biotinylated proteins
OtherFetal bovine serumSigmaCat#F7524-500mlComponent of cell culture media
OtherTrypsin-EDTA solutionSigmaCat#T3924Reagent used to lift cells
from culture vessel
OtherGeneJuiceMerck MilliporeCat#70967-3Transfection reagent
OtherEDTA-free protease inhibitorsRocheCat#11836170001Component of cell lysis buffer
OtherLipofectamine
RNAiMAX
InvitrogenCat#13778075Transfection reagent
OtherOptiMEMThermo Fisher ScientificCat#31985070Component of cell culture media
OtherProtein G agarose
beads
Merck MilliporeCat#16-266Affinity reagent for
immunoprecipitations
OtherNi-NTA agaroseQIAGENCat#1018244Affinity reagent for His-tag purification
OtherStreptavidin-coated
magnetic beads
New England BiolabsCat#S1420SAffinity reagent for Avi-tag pull-downs
OtherStreptavidin agaroseThermo Fisher ScientificCat#20357Affinity reagent for Avi-tag pull-downs
OtherSuperdex 200 16/600 columnGE HealthcareCat#28-9893-35Chromatography column
OtherSuperdex 75 16/600 columnGE HealthcareCat#28-9893-33Chromatography column
OtherUltracel-3K regenerated
cellulose centrifugal filter
Merck MilliporeCat#UFC900324Chromatography column
OtherUltracel-10K regenerated
cellulose centrifugal filter
Merck MilliporeCat#UFC901024Used for protein concentration
OtherUltracel-30K regenerated
cellulose centrifugal filter
Merck MilliporeCat#UFC903024Used for protein concentration
OtherNuPAGE 4–12% Bis-Tris gelThermo Fisher ScientificCat#NP0321BOXSDS PAGE electrophoresis gel
OtherSlide-A-Lyzer 20K MWCOThermo Fisher ScientificCat#66003Dialysis cassette
OtherSYPRO Orange dyeThermo Fisher ScientificCat#S6650Used for thermal shift assays
OtherMycoAlert PLUS
Mycoplasma Detection Kit
Lonza#LT07-705Used for testing cell lines
OtherMycoProbe Mycoplasma
Detection Kit
R&D Systems#CUL001BUsed for testing cell lines

Data availability

All structural models predicted using AlphaFold have been deposited in the University of Cambridge Data Repository: https://doi.org/10.17863/CAM.82741.

The following data sets were generated

References

Decision letter

  1. Tony Hunter
    Reviewing Editor; Salk Institute for Biological Studies, United States
  2. Jonathan A Cooper
    Senior Editor; Fred Hutchinson Cancer Research Center, United States
  3. Tony Hunter
    Reviewer; Salk Institute for Biological Studies, United States
  4. Stefan Knapp
    Reviewer; Goethe University Frankfurt, Germany

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Molecular mechanism of Afadin substrate recruitment to the receptor phosphatase PTPRK via its pseudophosphatase domain" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Tony Hunter as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Jonathan Cooper as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Stefan Knapp (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

The reviewers agree that your new insights into how Afadin interacts with the D2 pseudophosphatase domain of PTPRK and how this may serve as a recruitment mechanism through which the D1 active phosphatase domain of PTPRK can dephosphorylate Afadin are of significant interest, particularly because they could be the harbinger of a general mechanism for twin domain RPTP substrate recruitment. However, in the absence of a genuine Afadin CC/PTPRK D2 structure, some additional validation of the proposed Afadin CC/PTPRK D2 interface is required to bolster your conclusions before publication.

The essential revisions requested are:

1. The use of additional mutations of key residues in the predicted Afadin CC/PTPRK D2 interface to functionally validate the structural significance of the proposed Afadin recruitment interaction.

2. Further characterization of the pY1230 antibodies to demonstrate their specificity for pY1230 Afadin.

Some additional suggestions for improvement can be found in the individual reviews, where one particular concern is that the anti-pY1230 Afadin antibodies have not been sufficiently well characterized. We look forward to receiving a revised version of your paper describing this exciting new mechanism in which the poorly understood pseudophosphatase domain in PTPRK acts to recruit pTyr substrate proteins.

Reviewer #1 (Recommendations for the authors):

1. The AF2 model of the Afadin 1393-1455 region bound to the PTPRK D2 domain is supported by the properties of the acidic pocket DM D2 mutant, but, in the absence of a co-crystal structure, it would be reassuring if some additional mutations in the D2 domain as well as in the Afadin core binding motif were analyzed.

2. The DM mutant data show that the D2 domain interaction is important for Afadin pTyr dephosphorylation in vitro, but one would also like evidence that the DM PTPRK mutant lacks Afadin pY1230 dephosphorylating activity in cells. The authors have the PTPRK KO MCF10A cells they generated in their first eLife paper that could be used to re-express the WT and DM PTPRK and then monitor Afadin dephosphorylation with their new anti-pY1230 antibodies.

3. From the model structure in Figure 3B, it is not clear whether the Afadin helix binds to the backside or frontside of the D2 domain. In other words, where is the Afadin binding pocket relative to the "active" site of D2, and would oxidation of the active cite Cys be expected to alter Afadin binding? Some further discussion is called for.

4. The blotting data for the Afadin band in Figures 5F and G are not very clean and need quantifying. In the experiment in panel F it is not clear why the authors did not immunoblot an Afadin IP with their new pY1230 antibodies, instead of precipitating with anti-pTyr antibodies and then blotting with anti-Afadin antibodies. This would enable them to demonstrate that the Afadin/D2 interaction specifically promotes dephosphorylation of pY1230 as opposed to dephosphorylation of other pTyr residues in Afadin (admittedly panel G suggests that pY1230 is not dephosphorylated by the DM mutant). With regard to these experiments, the authors should indicate in the text that the pervanadate in the lysate was neutralized by addition of an excess of DTT (presumably before addition of the recombinant PTPRK proteins).

5. The anti-pY1230 Afadin antibodies should be more thoroughly characterized for their specificity. For instance, based on the top panel of Figure S2, it is not clear what band the reader is supposed to focus on – presumably the upper band but the "pY1230" signal in this band did not go down as much as the Afadin protein band in the second panel in the Afadin siRNA sample. The authors need to use a Y1230F mutant Afadin and YopH PTP treatment of the pV lysate as a negative controls, etc., to validate the specificity of these antibodies. If the new pY1230 antibodies can be shown to be specific, it would be interesting to know whether they can be used to detect pY1230 phosphorylated Afadin by IF, and whether it colocalizes with PTPRK.

Reviewer #2 (Recommendations for the authors):

1. Would recommend reporting in the figure legend how many experimental replicates have been collected for each experiment.

2. Related to #1, it is noted that the mean and SEM in Figure 2E are based on only two replicates.

3. The statement that the binding of the anti-afadin antibody is affected by phosphorylation could be better supported by experiments via transfection of tagged afadin followed by blotting with anti-tag vs anti-afadin antibodies.

4. Is the "reverse" mutation of PTPRM residues homolog to PTPRK aa 1273 and 1335 respectively into H and L sufficient to rescue the interaction with afadin and its dephosphorylation by PTPRM?

5. Could the authors check the numbering of residues in Supplementary Figure 9: residue 1273 appears to be an L in that figure.

Reviewer #3 (Recommendations for the authors):

I think this is an interesting study that reports an exciting new mechanism involving the poorly understood pseudophosphatase domain in this PTP. I think however since the foundation of the study heavily relies on in silico predictions, more experimental data should be provided to confirm the docking and interaction hypothesis. Some of the suggested experiments, for instance demonstrating that the CC domain in Afadin is indeed helical would be easy to confirm experimentally. Also, key residues in the substrate should be mutated. In general, including a mutant (Y/P mutant) in Afadin would help to confirm many of the experimental data as pY specific peptide antibodies are notoriously unselective.

https://doi.org/10.7554/eLife.79855.sa1

Author response

The essential revisions requested are:

1. The use of additional mutations of key residues in the predicted Afadin CC/PTPRK D2 interface to functionally validate the structural significance of the proposed Afadin recruitment interaction.

We have now made and tested several additional mutations within both the Afadin-CC and PTPRK-D2 domains to further validate the AF2 predicted model of the complex.

For Afadin-CC we introduced several single and double mutations along the helix including residues predicted to be in the interface and residues distal from the interface. These mutations and the pulldown with PTPRK are described in the text and are included as additional panels to a modified Figure 3. All mutations have the expected effect on the interaction based on the predicted complex structure. To help illustrate the positions of these mutations we have also included a figure of the interface with the residues highlighted.

For the PTPRK-D2 we have also introduced two new mutations, one buried in the interface (F1225A) and one on the edge of the interface encompassing a loop that is different in PTPRM (labelled the M-loop). GST-Afadin WT protein was bound to GSH beads and tested for their ability to pulldown WT and mutated PTPRK. These new mutations (illustrated in the new Figure 4 —figure supplement 2) further support the model prediction. F1225A almost completely abolishes binding as predicted, while the M-loop retains binding. These mutations and their effects are now described in the main text and the pull-down data, including controls and retesting of the original DM mutant, are included as panel H in a newly modified Figure 4 focussed solely on the PTPRK interface.

2. Further characterization of the pY1230 antibodies to demonstrate their specificity for pY1230 Afadin.

We would like this antibody to be a useful and freely accessible tool in the field and have taken on board the request for additional validation. To this end we have significantly expanded Supplementary Figure 2 (now Figure 1 —figure supplement 2) and included a dedicated section of the results as follows:

1. We have now included information about all of the Afadin antibodies used in this study, since Afadin(BD) appears to be sensitive to phosphorylation (Figure 1 —figure supplement 2A).

2. We have demonstrated that the Afadin pY1230 antibody detects an upregulated band in PTPRK KO MCF10A cells, consistent with our previous tyrosine phosphoproteomics (Figure 1 —figure supplement 2B). This indicates that the antibody can be used to detect endogenous Afadin phosphorylation.

3. We have included two new knock down experiments demonstrating the recognition of Afadin by our antibody (Figure 1 —figure supplement 2C). There appear to be two Afadin isoforms recognised in HEK293T cells by both the BD and pY1230 antibody, consistent with previous reports (Umeda et al. MBoC, 2015). We have highlighted these in the figure.

4. We have performed mutagenesis to demonstrate the specificity of the antibody. We tagged Afadin with a fluorescent protein tag, reasoning that it would cause a shift in molecular weight that could be resolved by SDS PAGE, as is the case. We noted that the phosphopeptide used spans an additional tyrosine, Y1226, which has been detected as phosphorylated (although to a much lower extent than Y1230) on Phosphosite plus. The data clearly show that Afadin cannot be phosphorylated when Y1230 is mutated to a phenylalanine (compared to CIP control), indicating that this is the predominant site recognised by the antibody. In addition, the endogenous pervanadate-stimulated signal is completely abolished by CIP treatment (Figure 1 —figure supplement 2D).

5. We have included densitometric quantification of the dephosphorylation assay shown in Figure 1B, which was part of a time course and shows preferential dephosphorylation by the PTPRK ICD compared to the PTPRK D1. The signal stops declining with time, which could indicate antibody background, or an inaccessible pool of Afadin-pY1230 (Figure 1 —figure supplement 2E).

6. To further demonstrate that this site is modulated by PTPRK in post-confluent cells, we have used doxycycline (dox)-inducible cell lines generated in Fearnley et al., 2019. Upon treatment with 500 ng/ml Dox for 48 hours PTPRK is induced to lower levels than wildtype, however, normalized quantification of the Afadin pY1230 against the Afadin (CST) signal clearly indicates downregulation by PTPRK WT, but not the catalytically inactive mutant (Figure 1 —figure supplement 2F and 2G).

Together these data strengthen our assertion that this antibody recognises endogenously phosphorylated Afadin at site Y1230, which is modulated in vitro and in cells by PTPRK phosphatase activity. For clarity, we have highlighted and annotated the relevant bands in figures. We have also included identifiers for each Afadin total antibody was used in particular experiments.

Some additional suggestions for improvement can be found in the individual reviews, where one particular concern is that the anti-pY1230 Afadin antibodies have not been sufficiently well characterized. We look forward to receiving a revised version of your paper describing this exciting new mechanism in which the poorly understood pseudophosphatase domain in PTPRK acts to recruit pTyr substrate proteins.

Reviewer #1 (Recommendations for the authors):

1. The AF2 model of the Afadin 1393-1455 region bound to the PTPRK D2 domain is supported by the properties of the acidic pocket DM D2 mutant, but, in the absence of a co-crystal structure, it would be reassuring if some additional mutations in the D2 domain as well as in the Afadin core binding motif were analyzed.

This request is encompassed by essential revisions point 1.

2. The DM mutant data show that the D2 domain interaction is important for Afadin pTyr dephosphorylation in vitro, but one would also like evidence that the DM PTPRK mutant lacks Afadin pY1230 dephosphorylating activity in cells. The authors have the PTPRK KO MCF10A cells they generated in their first eLife paper that could be used to re-express the WT and DM PTPRK and then monitor Afadin dephosphorylation with their new anti-pY1230 antibodies.

We agree with the reviewer that this would be an interesting experiment. Although we do have the PTPRK KO line these experiments would require careful control of PTPRK mutant expression levels therefore needing generation of new stable cell lines. This work will require substantial time and resources delaying the publication of this current work. As this has not been requested as an essential revision we respectfully acknowledge that these would be excellent experiments for future work.

3. From the model structure in Figure 3B, it is not clear whether the Afadin helix binds to the backside or frontside of the D2 domain. In other words, where is the Afadin binding pocket relative to the "active" site of D2, and would oxidation of the active cite Cys be expected to alter Afadin binding? Some further discussion is called for.

We have made a series of additional figures (included as a panel in new Figure 4A) to clearly illustrate that the D2 “active” site is on the opposite face from the Afadin binding site. Due to the localisation of the “active” site so far from the binding face we would not expect oxidation of this cysteine to alter binding. All experiments were done in the presence of DTT or TCEP, representing binding under reducing conditions.

4. The blotting data for the Afadin band in Figures 5F and G are not very clean and need quantifying. In the experiment in panel F it is not clear why the authors did not immunoblot an Afadin IP with their new pY1230 antibodies, instead of precipitating with anti-pTyr antibodies and then blotting with anti-Afadin antibodies. This would enable them to demonstrate that the Afadin/D2 interaction specifically promotes dephosphorylation of pY1230 as opposed to dephosphorylation of other pTyr residues in Afadin (admittedly panel G suggests that pY1230 is not dephosphorylated by the DM mutant). With regard to these experiments, the authors should indicate in the text that the pervanadate in the lysate was neutralized by addition of an excess of DTT (presumably before addition of the recombinant PTPRK proteins).

We carried out our dephosphorylation assays on pTyr IPs since this allows us to include multiple controls, such as p120-catenin, which is equally dephosphorylated by WT PTPRK and the DM mutant. If we had IP’d Afadin, we would also have had to IP p120-catenin and paxillin as controls. We agree that the DM mutant does not dephosphorylate pY1230 (now quantified in new Figure 5 —figure supplement 1) when measured directly. The DM mutant clearly impairs the ability of PTPRK to dephosphorylate Afadin.

We have now indicated in the Results section that DTT is used to quench any excess pervanadate.

5. The anti-pY1230 Afadin antibodies should be more thoroughly characterized for their specificity. For instance, based on the top panel of Figure S2, it is not clear what band the reader is supposed to focus on – presumably the upper band but the "pY1230" signal in this band did not go down as much as the Afadin protein band in the second panel in the Afadin siRNA sample. The authors need to use a Y1230F mutant Afadin and YopH PTP treatment of the pV lysate as a negative controls, etc., to validate the specificity of these antibodies. If the new pY1230 antibodies can be shown to be specific, it would be interesting to know whether they can be used to detect pY1230 phosphorylated Afadin by IF, and whether it colocalizes with PTPRK.

We have included additional characterisation of the phosphoantibody addressed in the response to essential revisions point 2.

As suggested, we tested the antibody in immunofluorescence experiments, but did not detect a signal that could be modulated by EGF treatment. This may require further optimisation, or may not be a viable application for the antibody.

Reviewer #2 (Recommendations for the authors):

1. Would recommend reporting in the figure legend how many experimental replicates have been collected for each experiment.

Where appropriate we have added the number of replicates for experiments. Several of the pulldowns were done using different combinations of constructs but all were tested a minimum of three times therefore we have stated that gels shown are “representative of n ≥ 3 independent experiments”. We have now included time course quantifications for dephosphorylation assays. For cell line experiments, replicates are either shown explicitly or reported in figure legends.

2. Related to #1, it is noted that the mean and SEM in Figure 2E are based on only two replicates.

ITC experiments require very large quantities of highly pure protein samples making them extremely resource intensive. Multiple measurements are made throughout the titration meaning that each single experiment involves 12 data points that are fit to a single binding model. We carried out the experiment twice with proteins purified independently for each experiment and obtained excellent agreement for the calculated affinity. We therefore did not consider additional ITC experiments to be necessary, especially in light of their resource-intensive nature.

3. The statement that the binding of the anti-afadin antibody is affected by phosphorylation could be better supported by experiments via transfection of tagged afadin followed by blotting with anti-tag vs anti-afadin antibodies.

Please see response to essential revisions point 2.

4. Is the "reverse" mutation of PTPRM residues homolog to PTPRK aa 1273 and 1335 respectively into H and L sufficient to rescue the interaction with afadin and its dephosphorylation by PTPRM?

Although well-designed single point mutations can disrupt a protein-protein interaction it is very rare to identify a gain-of-function mutation as it is not truly a “reverse” experiment. An interaction interface is the sum of many individual interactions that all contribute to the overall binding strength. Dropping the affinity below a certain threshold is relatively straight forward if the correct mutation is selected. However, introducing a mutation that can, on its own, mediate a new interaction is much less likely. As this experiment was not included as an essential revision we respectfully decline to mutate PTPRM into a more PTPRK-like molecule for this manuscript. To further support that it is the protein-protein interaction, and not the phosphosite, that is key to specificity, we have previously demonstrated that a chimeric recombinant PTPRM-D1-PTPRK-D2 protein is competent to dephosphorylate Afadin (Fearnley et al. 2019).

5. Could the authors check the numbering of residues in Supplementary Figure 9: residue 1273 appears to be an L in that figure.

We thank the reviewer for pointing out this mistake, it is now corrected. Note this figure is now labelled Figure 4 —figure supplement 2.

https://doi.org/10.7554/eLife.79855.sa2

Article and author information

Author details

  1. Iain M Hay

    1. Cambridge Institute for Medical Research, University of Cambridge, Cambridge, United Kingdom
    2. Signalling Programme, Babraham Institute, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5451-1768
  2. Katie E Mulholland

    Signalling Programme, Babraham Institute, Cambridge, United Kingdom
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  3. Tiffany Lai

    Signalling Programme, Babraham Institute, Cambridge, United Kingdom
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2451-0892
  4. Stephen C Graham

    Department of Pathology, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Software, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4547-4034
  5. Hayley J Sharpe

    Signalling Programme, Babraham Institute, Cambridge, United Kingdom
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing – review and editing
    For correspondence
    hayley.sharpe@babraham.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4723-298X
  6. Janet E Deane

    Cambridge Institute for Medical Research, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Conceptualization, Supervision, Methodology, Writing - original draft, Writing – review and editing
    For correspondence
    jed55@cam.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4863-0330

Funding

Wellcome and Royal Society (109407/Z/15/Z)

  • Hayley J Sharpe

Wellcome Trust (219447/Z/19/Z)

  • Janet E Deane

Royal Society (UF100371)

  • Janet E Deane

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

Acknowledgements

We thank Benjamin Butt for assistance with installing AlphaFold and Leah Wynn for assistance with Afadin mutagenesis. This study was supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society awarded to HJS (109407/Z/15/Z), which also supports KEM. JED was supported by a Royal Society University Research Fellowship (UF100371) and a Wellcome Trust Senior Research Fellowship (219447/Z/19/Z). IMH was funded by a CIMR PhD studentship and Babraham Institutional bridging funds. TL was funded by a PhD jointly funded by Trinity College, Cambridge, and the BBSRC Cambridge SBS-DTP. HJS is an EMBO Young Investigator and Lister Institute Prize Fellow. A Titan V graphics card used for this research was donated to SCG by the NVIDIA Corporation. For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) license to any Author Accepted Manuscript version arising from this submission.

Senior Editor

  1. Jonathan A Cooper, Fred Hutchinson Cancer Research Center, United States

Reviewing Editor

  1. Tony Hunter, Salk Institute for Biological Studies, United States

Reviewers

  1. Tony Hunter, Salk Institute for Biological Studies, United States
  2. Stefan Knapp, Goethe University Frankfurt, Germany

Publication history

  1. Preprint posted: April 27, 2022 (view preprint)
  2. Received: April 29, 2022
  3. Accepted: October 19, 2022
  4. Accepted Manuscript published: October 20, 2022 (version 1)
  5. Version of Record published: November 7, 2022 (version 2)

Copyright

© 2022, Hay et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 471
    Page views
  • 92
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Iain M Hay
  2. Katie E Mulholland
  3. Tiffany Lai
  4. Stephen C Graham
  5. Hayley J Sharpe
  6. Janet E Deane
(2022)
Molecular mechanism of Afadin substrate recruitment to the receptor phosphatase PTPRK via its pseudophosphatase domain
eLife 11:e79855.
https://doi.org/10.7554/eLife.79855

Further reading

    1. Biochemistry and Chemical Biology
    Ziyue Wang, Michael Mülleder ... Markus Ralser
    Research Article

    The possibility to record proteomes in high throughput and at high quality has opened new avenues for biomedical research, drug discovery, systems biology, and clinical translation. However, high-throughput proteomic experiments often require high sample amounts and can be less sensitive compared to conventional proteomic experiments. Here, we introduce and benchmark Zeno SWATH MS, a data-independent acquisition technique that employs a linear ion trap pulsing (Zeno trap pulsing) to increase the sensitivity in high-throughput proteomic experiments. We demonstrate that when combined with fast micro- or analytical flow-rate chromatography, Zeno SWATH MS increases protein identification with low sample amounts. For instance, using 20 min micro-flow-rate chromatography, Zeno SWATH MS identified more than 5000 proteins consistently, and with a coefficient of variation of 6%, from a 62.5 ng load of human cell line tryptic digest. Using 5 min analytical flow-rate chromatography (800 µl/min), Zeno SWATH MS identified 4907 proteins from a triplicate injection of 2 µg of a human cell lysate, or more than 3000 proteins from a 250 ng tryptic digest. Zeno SWATH MS hence facilitates sensitive high-throughput proteomic experiments with low sample amounts, mitigating the current bottlenecks of high-throughput proteomics.

    1. Biochemistry and Chemical Biology
    2. Neuroscience
    Jinli Geng, Yingjun Tang ... Xiaodong Liu
    Research Article Updated

    Dynamic Ca2+ signals reflect acute changes in membrane excitability, and also mediate signaling cascades in chronic processes. In both cases, chronic Ca2+ imaging is often desired, but challenged by the cytotoxicity intrinsic to calmodulin (CaM)-based GCaMP, a series of genetically-encoded Ca2+ indicators that have been widely applied. Here, we demonstrate the performance of GCaMP-X in chronic Ca2+ imaging of cortical neurons, where GCaMP-X by design is to eliminate the unwanted interactions between the conventional GCaMP and endogenous (apo)CaM-binding proteins. By expressing in adult mice at high levels over an extended time frame, GCaMP-X showed less damage and improved performance in two-photon imaging of sensory (whisker-deflection) responses or spontaneous Ca2+ fluctuations, in comparison with GCaMP. Chronic Ca2+ imaging of one month or longer was conducted for cultured cortical neurons expressing GCaMP-X, unveiling that spontaneous/local Ca2+ transients progressively developed into autonomous/global Ca2+ oscillations. Along with the morphological indices of neurite length and soma size, the major metrics of oscillatory Ca2+, including rate, amplitude and synchrony were also examined. Dysregulations of both neuritogenesis and Ca2+ oscillations became discernible around 2–3 weeks after virus injection or drug induction to express GCaMP in newborn or mature neurons, which were exacerbated by stronger or prolonged expression of GCaMP. In contrast, neurons expressing GCaMP-X were significantly less damaged or perturbed, altogether highlighting the unique importance of oscillatory Ca2+ to neural development and neuronal health. In summary, GCaMP-X provides a viable solution for Ca2+ imaging applications involving long-time and/or high-level expression of Ca2+ probes.