Article

Cooperative binding of the tandem WW domains of PLEKHA7 to PDZD11 promotes conformation-dependent interaction with tetraspanin 33

ROUAUD, Florian, et al.

Abstract

Pleckstrin homology domain-containing A7 (PLEKHA7) is a cytoplasmic at adherens junctions that has been implicated in hypertension, glaucoma, and responses to Staphylococcus aureus α-toxin. Complex formation between PLEKHA7, PDZ domain-containing 11 (PDZD11), tetraspanin 33, and the α-toxin receptor ADAM metallopeptidase domain 10 (ADAM10) promotes junctional clustering of ADAM10 and α-toxin-mediated pore formation. However, how the N-terminal region of PDZD11 interacts with the N-terminal tandem WW domains of PLEKHA7 and how this interaction promotes tetraspanin 33 binding to the WW1 domain is unclear. Here, we used site-directed mutagenesis, glutathione S-transferase pulldown experiments, immunofluorescence, molecular modeling, and docking experiments to characterize the mechanisms driving these interactions. We found that Asp-30 of WW1 and His-75 of WW2 interact through a hydrogen bond and, together with Thr-35 of WW1, form a binding pocket that accommodates a polyproline stretch within the N-terminal PDZD11 region. By strengthening the interactions of the ternary complex, the WW2 domain stabilized the WW1 [...]

Reference

ROUAUD, Florian, et al. Cooperative binding of the tandem WW domains of PLEKHA7 to PDZD11 promotes conformation-dependent interaction with tetraspanin 33. Journal of Biological Chemistry, 2020, vol. 295, no. 28, p. 9299-9312

DOI : 10.1074/jbc.RA120.012987 PMID : 32371390

Available at: http://archive-ouverte.unige.ch/unige:144944

Disclaimer: layout of this document may differ from the published version.

1 / 1 ARTICLE

Author’s Choice Cooperative binding of the tandem WW domains of PLEKHA7 to PDZD11 promotes conformation-dependent interaction with tetraspanin 33 Received for publication, February 11, 2020, and in revised form, April 30, 2020 Published, Papers in Press, May 5, 2020, DOI 10.1074/jbc.RA120.012987 Florian Rouaud1,2, Francesca Tessaro3 , Laura Aimaretti3, Leonardo Scapozza3, and Sandra Citi1,2,* From the 1Department of Cell Biology, Faculty of Sciences and the 2Institute for Genetics and Genomics in Geneva (iGE3), University of Geneva, Geneva, Switzerland, and the 3Pharmaceutical Biochemistry Group, School of Pharmaceutical Sciences, and Institute of Pharmaceutical Sciences of Western Switzerland University of Geneva, Geneva, Switzerland Edited by Phyllis I. Hanson

Pleckstrin homology domain–containing A7 (PLEKHA7) is At the tissue and organism level, PLEKHA7 has been impli- a cytoplasmic protein at adherens junctions that has been cated in and disease. Genome-wide associa- implicated in hypertension, glaucoma, and responses to tion studies show that single nucleotide polymorphism in the Downloaded from Staphylococcus aureus a-toxin. Complex formation between PLEKHA7 are associated with human hypertension and PLEKHA7, PDZ domain–containing 11 (PDZD11), tetraspanin high systolic pressure (5), and with primary angle closure glau- 33, and the a-toxin receptor ADAM metallopeptidase domain coma (6). In agreement, rats KO for PLEKHA7 show an atte- 10 (ADAM10) promotes junctional clustering of ADAM10 and nuated increase in blood pressure and kidney damage induced a-toxin–mediated pore formation. However, how the N-termi- by a high salt diet, and increased intracellular calcium and ni- http://www.jbc.org/ nal region of PDZD11 interacts with the N-terminal tandem trous oxide signaling in endothelial cells (7). In a WW domains of PLEKHA7 and how this interaction promotes model, the PLEKHA7 homolog Hadp1 is required for cardiac tetraspanin 33 binding to the WW1 domain is unclear. Here, morphogenesis and contractility (8). PLEKHA7 has also been S we used site-directed mutagenesis, glutathione -transferase involved in the control of microRNA processing (9, 10) and pulldown experiments, immunofluorescence, molecular mod- Rho GTPase activity (11). In cultured cells and mice, PLEKHA7 at Université de Genève on October 14, 2020 eling, and docking experiments to characterize the mechanisms is required to make cells more susceptible to the cytotoxic driving these interactions. We found that Asp-30 of WW1 and effects of Staphylococcus aureus a-toxin, through a mechanism His-75 of WW2 interact through a hydrogen bond and, to- gether with Thr-35 of WW1, form a binding pocket that that involves junctional clustering of ADAM10, mediated by tet- accommodates a polyproline stretch within the N-terminal raspanin 33 (Tspan33) (12, 13). PDZD11 region. By strengthening the interactions of the ter- Considering the important roles of PLEKHA7 in physiology nary complex, the WW2 domain stabilized the WW1 domain and disease, it is crucial to understand the molecular, cellular, and cooperatively promoted the interaction with PDZD11. and structural basis for its functions. PLEKHA7 comprises N- Modeling results indicated that, in turn, PDZD11 binding terminal tandem WW, pleckstrin homology, and coiled-coil induces a conformational rearrangement, which strengthens domains, and interacts with different junctional (1, 3, the ternary complex, and contributes to enlarging a “hydropho- 14) and membrane phospholipids (8). We identified PDZD11 bic hot spot” region on the WW1 domain. The last two lipo- as the highest affinity interactor of PLEKHA7 using 2-hybrid philic residues of tetraspanin 33, Trp-283 and Tyr-282, were screens and proteomic approaches, and showed that the inter- required for its interaction with PLEKHA7. Docking of the tetra- action of the N-terminal region of PDZD11 with the WW1 do- spanin 33 C terminus revealed that it fits into the hydrophobic main of PLEKHA7 is essential to cluster the Ig-like adhesion hot spot region of the accessible surface of WW1. We conclude molecules nectins at adherens junctions, and promote efficient that communication between thetwotandemWWdomainsof junction assembly (15). The same interaction is also required to – PLEKHA7 and the PLEKHA7 PDZD11 interaction modulate the dock the Staphylococcus aureus a-toxin receptor ADAM10 at ligand-binding properties of PLEKHA7. junctions, through binding of the WW1 domain of PLEKHA7 to the ADAM10 chaperone Tspan33 (13). Thus, the interaction with Cell-cell junctions are implicated in developmental, physio- PDZD11 is critical in mediating the activity of PLEKHA7 as a scaf- logical, and pathological processes. PLEKHA7, a protein local- fold for transmembrane proteins, such as nectins and Tspan33. ized at cadherin-based adherens junctions, was originally dis- However, nothing is known about the structural basis for the inter- covered as an interactor of -catenin (1) and paracingulin action of PLEKHA7 with PDZD11, and the mechanisms through (2, 3), and its interaction with , through the which PDZD11 promotes PLEKHA7 binding to Tspan33. minus-end binding protein nezha/CAMSAP3, stabilizes cad- WW domains are domains of 35-40 amino acids, character- herin-based junctions and the barrier function of epithelia (1, 4). ized by two highly conserved Trp residues, separated by 20 to 23 amino acids. WW domains fold into a very stable structure This article contains supporting information. with three antiparallel b-sheets, and can accommodate a high Author's Choice—Final version open access under the terms of the Creative Commons CC-BY license. degree of sequence variability (16, 17). They typically bind ei- * For correspondence: Sandra Citi, [email protected]. ther to proline-rich sequences or phosphorylated Ser or Thr

J. Biol. Chem. (2020) 295(28) 9299–9312 9299 © 2020 Rouaud et al. Published by The American Society for Biochemistry and Molecular Biology, Inc. Cooperative PLEKHA7-PDZD11-Tspan33 interaction residues, and are found in many human proteins, which play interacted with PDZD11 as efficiently as WT, and none with important roles in nuclear signaling, protein stabilization, the CFP-HA (Fig. 1D). In addition, Ponceau S staining of GST assembly of multiprotein networks, and diseases (18–20). fusion proteins showed no degradation, because all fusion pro- Because of the critical importance of the PDZD11 interaction teins migrated as one major polypeptide (Fig. 1D, Ponceau, red in PLEKHA7 function, we sought to gain further insights into arrows). These results suggest that the presence of the WW2 the structural aspects of the interaction between the WW domain stabilizes the WW1 domain. Next, to obtain separate domains of PLEKHA7 and PDZD11, using site-directed muta- evidence that these results reflect physiologically relevant genesis, GST (glutathione-S-transferase) pulldown assays, and mechanisms and interactions, we examined the effect of WW1 immunofluorescence. Our data allows to generate a model for mutations on the interaction between full-length PLEKHA7 the structure of the complex using molecular modeling and and PDZD11 in cells. PLEKHA7 recruits PDZD11 to junctions docking studies, which provides a rational mechanism through (15), and we used as a readout the rescue of the junctional local- which intramolecular WW1–WW2 interactions promote ization of PDZD11, which is abolished in PLEKHA7-KO cells, binding to PDZD11, and this in turn promotes interaction with by different WW domain mutants of PLEKHA7. In PLEKHA7- Tspan33. KO cells transfected with GFP alone, no PDZD11 labeling was detected at junctions (negative control, Fig. 1E, 1GFP, arrow- Results heads). In contrast, endogenous PDZD11 was rescued to junc-

Specific mutations within the WW1 domain of PLEKHA7 tions when PLEKHA7-KO cells where transfected with GFP- Downloaded from reduce binding to PDZD11, but the WW2 domain rescues the tagged full-length WT PLEKHA7 (Fig. 1E, WT, arrows)(15). effects of these mutations Consistent with the GST pulldown results carried out for the WW(1 1 2) domains, all the full-length PLEKHA7 constructs The N-terminal region of PDZD11 interacts with the WW1, harboring WW1 domain mutations rescued the junctional but not the WW2 domain of PLEKHA7 (Fig. 1A)(15). How-

localization of endogenous PDZD11 (Fig. 1E). Yet, deletion of http://www.jbc.org/ ever, the interaction is stronger when both WW1 and WW2 the WW1 domain prevented the rescue of the junctional local- domains are present (13, 15). The sequences of the WW1 (Fig. ization of PDZD11, confirming the critical role of the WW1 do- 1B) and WW2 (Fig. S1A) domains are highly conserved, main in PLEKHA7 interaction with PDZD11 (Fig. S1B). In because .90% of residues are identical between different spe- summary, these results identify residues in the WW1 domain cies (26/32 in WW1, 24/26 in WW2). To identify the residues that are required for the interaction with PDZD11 by GST pull- involved in the WW1-PDZD11 interaction and the mechanism at Université de Genève on October 14, 2020 down, suggesting that they are either directly implicated in the through which the WW2 domain promotes binding of PDZD11 to PLEKHA7 we used site-directed mutagenesis. We interaction, or they are required to maintain the correct folding generated alanine mutants within the WW1 domain at amino of the WW1 domain. Moreover, because the presence of the acid positions 15, 17, 19, 22, 25, 27, 29, 30, 32, and 35 (Fig. 1C). WW2 domain offsets the effects of specific mutations on the The WT and mutant WW1 sequences were fused downstream interaction between PLEKHA7-PDZD11 both by GST pull- of GST to generate baits for pulldown assays, where the preys down and in cells, we conclude that the WW2 domain protects were either PDZD11-HA or CFP-HA (negative control) (Fig. the WW1 domain from the de-stabilizing effects of the WW1 1C). Immunoblot analysis showed that the WT sequence and mutations. Therefore, we next wondered whether mutations the mutants Y17A, D22A, F27A, and L32A interacted with within the WW2 domain might affect the ability of the WW1 PDZD11, whereas the W15A, V19A, V25A, N29A, D30A, and domain to bind to PDZD11. Thus, we generated mutations T35A mutants failed to detectably interact with PDZD11 (Fig. within the WW2 domain, at positions 59, 60, 61, 65, 68, 71, 1C, top). None of the baits interacted with negative control 73, 75, 76, 84, and 85 (Fig. S1C, top). None of the mutations 1 HA-tagged CFP (Fig. 1C, bottom). Ponceau S staining of the affected the ability of the WW1 WW2 domain to interact blots revealed that all the WW1 mutant constructs migrated as with PDZD11, either by GST pulldown (Fig. S1C, bottom)or multiple bands, typically three polypeptides, with different rela- in cells (Fig. S1D). These results indicate that the structure of tive intensities (Fig. 1C, Ponceau), suggesting that the GST the WW1 domain and its ability to interact with PDZD11 is fusion proteins are unstable, and subjected to partial proteo- resistant to several mutations within the conserved residues lytic degradation. Among mutants that failed to interact with of the WW2 domain. PDZD11, T35A showed the least degree of degradation and migrated similar to WT, as indicated by the prevalence of the His-75 of the WW2 domain cooperates with Asp-30 and/or higher molecular size polypeptide (Fig. 1C, red arrows). Next, Thr-35 of the WW1 domain of PLEKHA7 to promote we asked whether mutations of the WW1 domain that resulted interaction with PDZD11 in decreased interaction with PDZD11 were also effective in Next, we sought to clarify the mechanism of WW1 stabiliza- the context of constructs comprising both WW1 and WW2 tion by WW2, by attempting to find combinations of mutations domains. Thus, we generated fusion proteins comprising the within the two domains that would affect the interaction with first 160 residues of PLEKHA7 (e.g. both WW1 and WW2 PDZD11. We selected two mutations within the WW1 domain: domains and a short downstream sequence, but not the pleck- the T35A mutation, which showed the least degradation of the strin homology domain, Fig. 1A), either with WT sequence, or recombinant bait protein (Fig. 1C), or the nearby D30A muta- with the single amino acid mutations within the WW1 domain tion, which also showed strongly decreased interaction with (Fig. 1D). Immunoblot analysis showed that all mutant baits PDZD11 (Fig. 1C). When we combined either the T35A or

9300 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction Downloaded from http://www.jbc.org/ at Université de Genève on October 14, 2020

J. Biol. Chem. (2020) 295(28) 9299–9312 9301 Cooperative PLEKHA7-PDZD11-Tspan33 interaction

D30A mutations in the WW1 domain with different mutations thus generating a “binding pocket” suitable for accommodating within the WW2 domain, only T35A 1 H75A and D30A 1 PDZD11 at the interface between the two domains (Fig. 2D, red H75A combinations resulted in the complete loss of binding to rectangle). The specific role for His-75 was highlighted by the the PDZD11 prey by GST pulldown assay (Fig. 2A and Fig. S2, observation that mutations in neighboring amino acids A and C), and in a failure to rescue the junctional localization of (D74A, N76A) had no effects on the interaction with PDZD11, PDZD11 within cells (Fig. 2B). Other combinations of either alone or in combination with either D30A or T35A (Fig. mutations could rescue junctional PDZD11 (Fig. S2B), demon- S2, C–F). In agreement, structural inspection of the models strating that the results of GST pulldowns reflect physiologi- reveals that Asp-74 and Asn-76 face away from WW1 domain, cally relevant interactions and are not artifacts. Finally, when suggesting that they are not involved in the interaction between combining together either the D30A or T35A mutations with the two domains. Because in our model Arg-33 is localized in mutations of the residues flanking His-75 within the WW2 do- the binding pocket close to Asp-30 (Fig. 2C), we tested its rele- main, e.g. either D74A or N76A, the interaction with PDZD11 vance by combining the mutation R33A with mutations of was maintained (Fig. S2, C–E). These results were confirmed D74A, H75A, or N76A. These double mutants interacted well when we switched bait and prey, e.g. we used GST-PDZD11 as with PDZD11 by GST pulldown and in cells (Fig. S2, E and F), a bait, and either WT or mutant constructs of the WW1 1 indicating that His-75 of WW2 is the main interactor of Asp-30 WW2 domains of PLEKHA7 as preys. The single mutations of WW1 (Fig. 2C). –

D30A, T35A, and H75A did not affect PDZD11 interaction The WW1 WW2 complex resulted to be stable in time dur- Downloaded from with the WW1 1 WW2 domains (Fig. S2G), but the double ing the 120 ns of molecular dynamic (MD) simulations, as mutants, either D30A 1 H75A, or T35A1H75A resulted in shown by the low values of root mean square deviation (RMSD) loss of interaction with PDZD11 (Fig. S2H). (Table S1), specifically in the last 60 ns of the trajectory (Fig. Next, we combined the information derived from these mu- S3E, WW(1 1 2)-WT). The maintenance of a close distance tagenesis studies with structural modeling to gain further between the two domains is shown also by the clustered struc- http://www.jbc.org/ understanding on the structural basis of the PLEKHA7- tures from MD simulation, showing the representative confor- PDZD11 interaction. Based on the homology of the WW1 and mations assumed by the WW(1 1 2)-WT complex along the WW2 domains with templates of the ubiquitin ligase Nedd4 simulating time (Fig. S3G). In agreement, hydrogen bond fre- (48% sequence identity with WW1) and the Syntaxin-binding quencies revealed that the main interacting residues were local- protein (58% sequence identity with WW2), we modeled a ized at the interface between the two domains, which con- three antiparallel b-sheet structures of the WW1 and WW2 curred at the maintenance of the complex stability (Table S2). at Université de Genève on October 14, 2020 domains (Fig. S3, A and B)(22). As a validation of the correct Importantly, molecular dynamic simulations also showed that geometry of the models, the amino acid residues were located the WW1–WW2 system carrying the double point mutations in the “allowed” regions of the Ramachandran plot (Fig. S3, C T35A1H75A and D30A1H75A appear more unstable in and D). Based on the models and on the experimental results terms of RMSD values compared either to the WT system or to highlighting the importance of the His-75 of the WW2 domain single mutants D30A and T35A (Fig. S3E and Table S1). As was together with Asp-30 and Thr-35 of the WW1 domain, we car- expected by analyzing the overall complex, we noticed that the ried out a series of molecular docking analyses, which allowed residues located at the N-terminal, C-terminal, and loop to model the interaction between the WW1 and WW2 regions are more prone to movements, compared with the resi- domains (Fig. 2, C and D). According to this model, His-75 of dues forming the b-sheets (Fig. S3F). To test whether a specific WW2 is accommodated in a small pocket of the WW1 domain geometry of arrangement between WW1 and WW2 is required formed by residues Asp-30, Arg-33, Tyr-17, and Ser-16 (Fig. to generate the binding pocket, we asked whether the isolated 2D, see below). Although the contact surface between the two WW domains can directly bind to each other or to the domains is small (549 Å2) compared to the overall accessible WW11WW2 domain. GST pulldown showed that neither surface area of the complex (5546.75 Å2), the electrostatic WW1, WW2, nor the WW1 1 2 domains interact with either potential surface at the interface of the two domains shows the isolated WW2 domain or the WW1 1 2 domain (Fig. S4A). good complementarity (Fig. 2D). Residues Asn-76, Glu-77, and Furthermore, the WW2 domain did not, when isolated, pro- Glu-78, at the interface of WW2 domain, electrostatically con- mote the interaction of the WW1 domain with PDZD11 (Fig. tribute to strengthening the interaction (Fig. 2D, see below). In S4, B and C). Together, these data indicate that the tandem addition, His-75 forms a hydrogen bond with Asp-30 of WW1, WW domains of PLEKHA7 do not dimerize, but interact

Figure 1. The WW2 domain of PLEKHA7 stabilizes the WW1 domain and rescues the interaction of WW1 mutants with PDZD11. A, schematic diagrams of PDZD11 (top) and PLEKHA7 (bottom), with the indicated structural domains. The N-terminal sequence of PDZD11 (P7-ID: PLEKHA7-interaction domain) interacts with the WW1 domain of PLEKHA7 (arrow)(15). B, top: sequence of the WW1 domain of PLEKHA7. Bottom: Weblogo diagram of residue conservation. C, top: sequence of WW1 domain, with mutations of highly conserved residues (to Ala) highlighted in yellow.Thenumbers above each residue indicate residue number in the sequence. Bottom: immunoblot analysis of GST pulldowns using GST (G) fused to WT and mutant WW1 domains as bait, and either PDZD11-HA or CFP-HA as preys. GST and GST fusion baits are indicated in red, and preys are indicated in green. Ponceau S-stained blots below immunoblots show baits. Red arrows indicate baits showing no or little proteolytic degradation. D, top: sequence of WW1 (continuous line, residues 1–53) and WW2 (dotted line, residues 43–98) domains, with highlighted WW1 mutations. The sequence linking the WW1 to the WW2 domain is in a dotted box. Bottom, immunoblot analysis of GST pulldowns using GST fused to WT and mutant WW1 1 WW2 domains as bait, and either PDZD11-HA or CFP-HA as preys. Ponceau S-stained blots below the im- munoblots show baits. E, immunofluorescence localization of endogenous PDZD11 in PLEKHA7-KO mCCD cells rescued either with GFP, or with either WT or WW1 point mutants of GFP-tagged full-length PLEKHA7. Merged images show nuclei in blue (DAPI). Arrows indicate junctional labeling. Arrowheads indicate decreased/undetected junctional labeling. Bar =20mM.

9302 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction Downloaded from http://www.jbc.org/ at Université de Genève on October 14, 2020

J. Biol. Chem. (2020) 295(28) 9299–9312 9303 Cooperative PLEKHA7-PDZD11-Tspan33 interaction weakly, when in a specific spatial geometry, through a small The ability of these fusion proteins to interact with the complementary surface, and Asp-30, Thr-35, and His-75 drive PLEKHA7 prey was examined by GST pulldown (Fig. 3, C and the creation of a pocket that allows high-affinity binding to D). Immunoblot analysis showed that the E19A,N20A, and PDZD11 (Fig. 2, C and D). Interestingly, the residue His-75 and P26A 1 P27A mutants interacted with PLEKHA7 similarly to even more so the Thr-35 are well conserved within tandem WT (Fig. 3, C and D). The P16A and P21A,P22A mutants WW domains (Fig. S5, A–C), suggesting that they may play a showed reduced interaction with PLEKHA7 only when in the similar role in stabilizing potential interactions between tan- context of the short (1–30) construct (Fig. 3C), but not in the dem domains of additional proteins. context of full-length PDZD11 (Fig. 3D). In contrast, the muta- tion Y18A and the combined mutations of either P16,21,22A, The N-terminal polyproline sequences of PDZD11 are or P21,22,26,27A, or P16,26,27A or P16,21,22,26,27A resulted required for interaction with PLEKHA7 in either decreased or undetectable interaction (polyproline Next, we sought to identify the sequences within PDZD11, mutations) with PLEKHA7 both in the context of the short N- which are required for its interaction with PLEKHA7. Previ- terminal construct, and in the context of full-length PDZD11 ously, we showed that the isolated N-terminal domain of (Fig. 3, C and D). These results indicate that the residues that PDZD11 (P7-ID: P7-interaction domain, residues 1-44) is suffi- affect interaction with PLEKHA7 are either required to main- cient for the interaction with the WW1 domain of PLEKHA7 tain the correct folding of the N-terminal domain of PDZD11

(15). Thus, we generated a series of deletion mutants of the P7- or are directly implicated in the interaction with PLEKHA7. Downloaded from ID, within full-length PDZD11 and analyzed their ability to Ponceau S staining of the blots showed that within the short N- interact with full-length PLEKHA7 by GST pulldown (Fig. 3A). terminal bait, the proline mutants but not the Y18A mutant Immunoblot analysis showed that deletion of up to 14 N-termi- migrated as two main polypeptides (Fig. 3C), whereas within nal residues (D4, D9, D14) of full-length PDZD11 did not the full-length molecule, they all migrated as a single band (Fig.

impact the binding to PLEKHA7, when compared with WT 3D). This observation suggests that proline mutations destabi- http://www.jbc.org/ PDZD11, whereas deletion of 19, 24, or 29 residues (D19, D24, lize the conformation of the short peptide, rendering it suscep- and D29 constructs) resulted in either decreased or no interac- tible to proteolytic degradation, whereas when the mutations tion above background (Fig. 3A). Furthermore, a bait construct are in the context of the full-length molecule, the structure is consisting of the first N-terminal 30 residues of PDZD11 was stabilized. sufficient to interact with the PLEKHA7 prey (Fig. 3A, N-term Next, we asked whether the PDZD11 mutations influenced at Université de Genève on October 14, 2020 1-30). This analysis suggested that residues 16-30 of PDZD11 the junctional localization of exogenous PDZD11 in WT cells are essential for its interaction with PLEKHA7. (Fig. 3E). In cells transiently transfected with GFP alone, GFP To identify more precisely the residues of PDZD11 involved labeling was detected in cytoplasm (Fig. 3E, 1GFP). In cells in interaction with PLEKHA7 we performed mutagenesis of transfected with either GFP-tagged PDZD11 WT, or with the residues within the 16-30 PDZD11 sequence. Analysis of the mutants P16A, E19A, N20A, P21,22A and P26,27A exogenous sequence conservation of the N terminus of PDZD11 in 100 PDZD11, labeled by GFP, was colocalized with endogenous vertebrate species revealed that 27/31 residues are identical, PLEKHA7 at junctions (Fig. 3E, arrows). The GFP-tagged con- including 5 proline residues, 4 of which (Pro-21, -22, -26, and struct of PDZD11 with the mutations at Y18A was localized -27) in the form of two consecutive prolines ( Fig. 3B, red box). both at junctions and in the cytoplasm (Fig. 3E, arrows and Because polyproline sequences are key ligands for WW asterisks), whereas GFP-tagged constructs of PDZD11 with domains (22), we generated either single or multiple alanine mutations at P16,21,22A, or P16,26,27A, or P21,22,26,27A, or mutants of these prolines, at positions 16, 21, 22, 26, and 27 P16,21,22,26,27A, were localized predominantly in the cyto- (Fig. 3B, yellow highlight). In addition, we also generated ala- plasm (Fig. 3E, asterisks). Together, these results identify the nine mutations of the highly conserved Tyr-18, Glu-19, and polyproline stretches 16, 21, 22, and 26, 27, together with Tyr- Asn-20 residues ( Fig. 3B, yellow highlight). GST fusion proteins 18, as the key residues of PDZD11 that are implicated in were generated either in the context of a short construct of the recruitment by PLEKHA7 at the junction in cells. first 30 residues of PDZD11 (Fig. 3C), or in the context of full- Finally, based on these experimental data, we docked the N- length PDZD11 (Fig. 3D), this latter to provide the binding terminal heptapeptide of PDZD11 (residues 16 to 26) into the region with the stabilizing effect of the downstream sequence. binding pocket of the WW1–WW2 model of PLEKHA7 (Fig.

Figure 2. Binding of PDZD11 to the cleft between WW1 and WW2 domains requires either Asp-30 or Thr-35 of WW1 and His-75 of WW2. A, top: sequence of WW1 1 WW2 domains, with mutations in residues within WW1 and WW2 highlighted in yellow.Thenumbers above and below each residue indi- cate residue number in the sequence. Bottom: immunoblot analysis of GST pulldowns using GST (G) fused to WT and mutant WW1 1 WW2 domains as bait, and either PDZD11-HA or CFP-HA as preys. B, immunofluorescence localization of endogenous PDZD11 in PLEKHA7-KO mCCD cells rescued with either GFP, or with either WT or WW1 point mutants or WW2 point mutants, or WW1 1 WW2 point mutants of GFP-tagged full-length PLEKHA7. Merged images show nuclei in blue (DAPI). Arrows indicate junctional labeling and asterisks indicate reduced/undetectable labeling. Bar =20mM. C, model of interaction between the WW1 domain (gray, with surface representation) and WW2 domain (orange). The dotted black line shows the hydrogen bond occurring between His-75 and Asp-30. D, electrostatic potential surface (EPS) of the WW1 and WW2 domains. The color ramp is set with a minimum value of 20.2 (red) and a maximum of 0.2 (blue). The WW2 domain presents in transparency an orange molecular surface for the EPS = 0 instead of the white used for WW1 to help in distinguish of the two domains. The probe radius to define the accessible surface area was set to 1.4 Å. The image shows the molecule from different space orientations turning along the y and x axes. In the upper part of the panel, a violet rectangle indicates the pocket that accommodate PDZD11. In the panel below the interact- ing surface between the two domains is shown from the below (1 90°) and upper (290°) views, respectively. The WW2 and WW1 domains are not displayed to allow the visualization of the contact surface indicated by the black circle. Images were obtained using the Molecular Surface tool from Maestro (Schrö- dinger release 2016-4) program.

9304 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction Downloaded from http://www.jbc.org/ at Université de Genève on October 14, 2020

Figure 3. Polyproline stretches and tyrosine 18 in the N-terminal region of PDZD11 are required for its interaction with PLEKHA7. A, top: schematic domain organization of PDZD11, and schemes of truncated constructs used in GST pulldown assays, and their interaction (Bind P7-His) with full- length PLEKHA7 (indicated by 1, 6, 2). Bottom: immunoblot analysis of GST pulldowns using either GST (G), or GST fused to either full-length PDZD11 or the truncated constructs (shown on top) as baits, and full-length PLEKHA7-His as prey. Ponceau S-stained blots below the immunoblots show baits. B, sequence of the P7-ID domain (15) of PDZD11, and Weblogo diagram of residue conservation within this region. C, immunoblot analysis of GST pulldowns using GST (G) fused to either WT or mutated P7-ID N-terminal domain of PDZD11 as bait and PLEKHA7-His as prey. D, immunoblot analysis of GST pulldowns using either GST, or GST fused to either WT or mutated full-length PDZD11 as bait, and PLEKHA7-His as prey. Ponceau S-stained blots below immunoblots show baits. E, im- munofluorescence localization of endogenous PLEKHA7 in WT mCCD cells, transiently transfected with either GFP, or WT or mutants GFP-tagged full-length PDZD11. Merged images show nuclei in blue (DAPI). Arrows indicate junctional labeling, and asterisks indicate decreased/undetectable junctional labeling. Bar 5 20 mM. F, binding mode of the PDZD11 dodecapeptide (PAYENPPAWIPP) in the pocket formed by WW1–WW2 domains. PDZD11 is shown in magenta as licorice style, whereas the WW1 and WW2 domains are represented, respectively, in gray and orange as a new cartoon and transparent surface style. The cyan dotted line indicates the cation-p interaction of Tyr-18 with Arg-21 in the WW1 domain. Hydrophobic contacts in the WW2 domain are represented with or- ange curved lines (near Phe-64), which indicate the pocket accommodating the tandem prolines 26-27. The image was generated with VMD software.

3F). The experimental constraints used for docking are the tion similar to what is known for other proline-rich WW do- identified interacting residues on WW1 and WW2. In the main ligand (22) placing the prolines 21-22 in proximity of the model, the peptide displays an N-terminal/C-terminal orienta- pocket formed by Arg-33, Asp-30, and His-75. Furthermore,

J. Biol. Chem. (2020) 295(28) 9299–9312 9305 Cooperative PLEKHA7-PDZD11-Tspan33 interaction we observed that Tyr-18 can drive the interaction with the WW2 domains of PLEKHA7 by GST pulldown (Fig. 4C). Im- WW1 domain in two possible orientations. In the first, Tyr-18 munoblot analysis showed that deletion of the last 2 residues of forms an interaction with Arg-21 either through a cation-p Tspan33 (Trp-Tyr) was sufficient to abolish the interaction interaction with the charged residue’s side chain (Fig. 3F)or between Tspan33 and PLEKHA7 (Fig. 4C). Interestingly, the C- with a hydrogen bond (Fig. S5D, violet model). In the second terminal region of Tspan33 presents an excellent complemen- mode, the peptide shifts toward the right, and the Tyr-18 can tarity in terms of lipophilic surface with respect to the hydro- form a hydrogen bond with the backbone of Arg-33 (Fig. S5D, phobic hot spot region of the WW1 domain (Fig. 4D left). The pink model). Our experimental data do not test the existence or docking of the C-term Tspan33 decapeptide perfectly places relative populations of these modes. The PDZD11 peptide also the last two residues, Trp-282 and Tyr-283, in the hydrophobic interacts with WW2, reinforcing the overall complex stability. region of the WW1 domain (Fig. 4D, right). Indeed, Trp-282 The tandem prolines 26-27 and Trp-24 are accommodated in and Tyr-283 are the main contributors to the lipophilic surface the hydrophobic and solvent-exposed face (formed by Phe-64 of the Tpsan33 C-term, suggesting that the recognition of and Phe-72) of the WW2 domain (Fig. 3F). Finally, the stability these macromolecules (WW1–WW2-PDZD11 and Tspan33) of the complex was monitored measuring the central mass dis- is mediated especially by hydrophobic/hydrophilic forces. To tance between the PDZD11 peptide (residues 16-22) and, confirm the role of these residues, we studied mutants of PLE- respectively, WW1 and WW2 domains during 120 ns of MD KHA7, and examined how double and single mutations within

simulation (Fig. S5E). The PDZD11 peptide strongly maintains the WW1 and WW2 domains affect interaction with the C ter- Downloaded from the interaction with the WW1 domain, but not the one with minus of Tspan33 by GST pulldown. Although the GST fusion WW2, which is restored only in the final steps of the simulation of the last 28 residues of Tspan33 (G-T33-Cterm, Fig. 4E) inter- (Fig. S5E). Although the short MD simulation does not permit acted well with the WW1 1 WW2 prey, either the double full evaluation of large scale motions and populations of side mutations D30A1H75A or T35A1H75A (Fig. 4E) or the single

chain rotamers, these results suggest that the binding of mutations D30A, T35A, and H75A (Fig. 4F) abolished binding. http://www.jbc.org/ PDZD11 to the tandem WW domains can cause a conforma- This suggested that the lipophilic surface of WW1 that interacts tional rearrangement of the tandem WW1–WW2 domains, with Tspan33 is exquisitely sensitive to the conformation of the which can favor the successive binding of other effector pro- tandem domain structure, and may require PDZD11 for stabili- teins, as Tspan33. zation. To test this, we compared the binding of the WW1 1 WW2 prey to the Tspan33 C-terminal bait in the absence or at Université de Genève on October 14, 2020 PDZD11 promotes the interaction of Tspan33 with PLEKHA7 presence of either PDZD11 or CFP, as a negative control (Fig. 4, by increasing a hydrophobic surface on the WW1 domain G and H). Immunoblot analysis showed that PDZD11 increases By GST pulldown analysis, the N-terminal region of the binding of both WT and single mutant preys to the Tspan33 PLEKHA7 binds weakly to the C-terminal region of Tspan33, bait, but cannot rescue the loss of binding to the double mutant but this interaction is greatly enhanced in the presence of prey (Fig. 4, G and H), consistent with the observation that it PDZD11 (13). In agreement, PDZD11 is required in cells for does not bind to the double mutant (Fig. 2A). the interaction of PLEKHA7 with Tspan33, and clustering of the Tspan33-ADAM10 complex at junctions (13). To investi- Discussion gate the mechanism through which PDZD11 promotes binding The precise molecular mechanisms through which PLE- of Tspan33 to the WW domains of PLEKHA7 using the inter- KHA7 is implicated in disease (5, 6), morphogenesis (8, 23), action model developed above, we stripped off the PDZD11 and diverse cellular functions (9–11) are poorly understood. ligand to investigate the lipophilic molecular surface properties Here we address the structure-function relationships of the N- of the WW1–WW2 complex either in the absence (Fig. 4A)or terminal tandem WW domains of PLEKHA7, a critically im- presence (Fig. 4B) of PDZD11. PDZD11 (Fig. 4B, purple) binds portant region for the ability of PLEKHA7 to scaffold and clus- to the pocket generated at the interface between WW1 and ter transmembrane proteins (13, 15). We propose a structural WW2 domains (see also Fig. 3F). We noticed that the WW1 model (Movie S1), which provides a rational explanation for domain displays a highly lipophilic region at the solvent-acces- the ability of the WW2 domain of PLEKHA7 to promote bind- sible surface, which resulted in a “hydrophobic hot spot” for ing of the WW1 domain to PDZD11, and for the ability of protein recognition (Fig. 4A, right, orange arrow). The lipo- PDZD11 to promote Tspan33 binding to the WW1 domain. philic cavity is formed mainly by the conserved residues Leu- Our model assumes a single rigid conformational state, which 10, Trp-15, Tyr-17, Phe-27, Leu-32, and Leu-38, which are not as noted above requires additional experimental testing. directly involved in the interaction with PDZD11. Strikingly, Proteins that contain multiple or tandem WW domains the presence of PDZD11 significantly increased (by 73.26 Å2) show different structural configurations and reciprocal func- the size of this lipophilic surface and thus, extending the hydro- tional interactions between these domains (24, 25). In some phobic hot spot region (Fig. 4B, right, red dotted rectangle). Pre- cases tandem WW domains are separated by a relatively rigid viously, we showed that the isolated C-terminal domain of helical linker, and their aromatic polyproline-binding surfaces Tspan33 (residues 256-283) is sufficient for the interaction show opposite orientations, allowing interaction with distinct with the WW1 domain of PLEKHA7 (13). Here we generated ligands (26). In other cases there is a greater degree of freedom three deletion mutants (D2, D5, D10) of the C-terminal tail of between the two WW domains, allowing different spatial orien- Tspan33 and analyzed their ability to interact with the WW1 1 tation of ligands and substrates (27). An additive effect can

9306 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction Downloaded from http://www.jbc.org/ at Université de Genève on October 14, 2020

J. Biol. Chem. (2020) 295(28) 9299–9312 9307 Cooperative PLEKHA7-PDZD11-Tspan33 interaction occur when both WW domains can interact with the same PDZD11 does not precisely match any of these sequences. ligand, and the binding affinity to one is increased by the pres- However, group II and group III WW domains not only recog- ence of the second domain (24, 25). The second of the tandem nize PPLP and PR-containing peptides, but also polyproline WW domains can act as a chaperone, to facilitate ligand bind- stretches containing glycine, methionine, or arginine (20). In ing to the first domain, whereas not interacting with the ligand, addition to the polyproline sequence of the N-terminal region or can change either the conformation or ligand binding speci- of PDZD11, the Tyr-18 residue of PDZD11 also appears crucial ficity of the second WW domain. Tandem WW domains can in the binding to the WW domains, probably guiding the orien- also form a cleft at their junction, which allows ligand binding tation of the peptide and interacting with either Arg-21 or Arg- to both domains (28–34). Our results indicate proteolytic deg- 33. Moreover, the tandem prolines Pro-26–Pro-27 of PDZD11 radation of the isolated bacterially expressed WW1 domain appear difficult to pinpoint in the interaction with the tandem constructs, whereas the same constructs are not degraded domains, as shown by GST pulldowns. Nevertheless, their when in the context of the WW1 1 WW2 sequence, suggesting accommodation in the solvent-exposed face of WW2 domain that the presence of the WW2 domain stabilizes the structure might contribute to strengthening the overall stability of the of the WW1 domain. The WW2 domain may also act by modi- complex. fying the conformation of the WW1 domain ligand-binding The PDZD11-dependent clustering of transmembrane pro- groove, because none of the WW1 mutations that prevented teins by PLEKHA7 appears to occur by at least two different the interaction of the isolated WW1 domain with PDZD11 was mechanisms. In the case of nectins, their C terminus interacts Downloaded from effective when in the context of the tandem WW1 1 WW2 do- with the PDZ domain of PDZD11 through a canonical PDZ- main. Another, nonexclusive possibility, is that binding of the binding motif (15), indicating that PLEKHA7 anchors nectins ligand to the WW2 domain occurs only when in the context of indirectly, through PDZD11. However, in the case of Tspan33, the WW1 1 WW2 domains, and contributes to increasing the the C-terminal region of Tspan33 interacts directly with the binding affinity to the WW1 domain for either the same ligand, WW1 domain of PLEKHA7, but not with PDZD11 (13). The http://www.jbc.org/ or for a different ligand. Conversely, binding of PDZD11 to the C-terminal sequence of Tspan33 contains a proline residue WW1 domain may synergistically enhance the PDZD11 bind- (Asp-Pro-Trp-Tyr) but does not fall into any of the previously ing potential of the WW2 domain, as has been suggested for established ligand groups for recognition by WW domains, other tandem WW proteins (24). The relatively high sequence making it unlikely that it is a canonical binding mode. In agree-

identity between the WW1 and WW2 domains of PLEKHA7 ment, based on the lipophilic surface properties, the C-terminal at Université de Genève on October 14, 2020 suggests that they could both bind to the same ligand. By Tspan33 sequence can recognize a highly hydrophobic and sol- exploring how combined mutations in both WW1 and WW2 vent-exposed region of WW1 that is distinct from the one that domains affect the interaction with PDZD11, and modeling the interacts with PDZD11. The binding of PDZD11 to the cleft structure of the PLEKHA7 WW domains, we propose that between the WW1 and WW2 domains not only stabilizes the Asp-30 of the WW1 domain interacts with His-75 of WW2. In WW1 1 WW2 tandem domains, but also contributes to our model, which needs to be further validated by either NMR increasing the surface of the lipophilic region, which is a perfect spectroscopy or other computational and biophysical methods, “hot spot” for proteins recognition driven by hydrophobic the contact surface between the two WW domains is comple- forces (35). Molecular docking shows that the Trp-Tyr dipep- mentary, but relatively small (549 Å2), and these domains form tide at the extreme C terminus of Tspan33 perfectly fits into a binding pocket that interacts with the N-terminal, proline- the extended lipophilic surface. rich peptide of PDZD11. Three residues in the WW1 domain, Our molecular studies provide a rationale explaining how Arg-21, Asp-30, and Thr-35, and one residue in the WW2 do- PDZD11 promotes the binding of the C terminus of Tspan33 main, His-75, are critically involved in promoting the interac- to PLEKHA7. Previously we showed that clustering of tion of PLEKHA7 with PDZD11. ADAM10 at junctions occurs through sequential and hierarch- WW domains are classified based on the sequences of the ical multimolecular interactions. ADAM10 must first be interacting ligands in 4 groups: PPXY (I), PPLP (II), PR (III), or p docked to junctions through the interaction of its chaperone (S/T)P (IV). The sequence of the WW-interacting region of Tspan33 with the PLEKHA7-PDZD11 complex, before it can

Figure 4. Binding of PDZ11 to the WW1 1 WW2 domains induces an increase in the size of a hydrophobic hot spot region on WW1, which docks the tetraspanin 33 C-terminal peptide. A and B, lipophilic potential of the surface of the WW1–WW2 domains in the absence (A) or presence (B) of PDZD11. PDZD11 (16-27 peptide) is represented in magenta, with a wide frames surface. The image shows the molecule from different space orientations turning along the y (190°) and x (245°) axis. An orange double arrow indicates the hydrophobic hot spot region of the WW1 domain. In the presence of PDZD11 the surface of the lipophilic surface is increased, thus potentially enhancing the binding of hydrophobic ligands. The image was generated using the Molcad tool from SybylX-2.1.1. °C. Top, schemes of truncated constructs of Tspan33 used in GST pulldown assays. Bottom, immunoblot analysis of GST pulldowns using GST (G) fused to either C-terminal Tspan33 (G-T33-Cterm) or the truncated constructs (shown on left) as baits, and either PLEKHA7-WW(1 1 2)-Myc or GFP-Myc as preys. D, binding of the C-term dodecapeptide Tspan-33 (NQQHRADFWY) on the lipophilic surface of the WW1 domain. The WW1–WW2-PDZD11 complex is represented as surface with the WW1–WW2 domains as lipophilic potential and PDZD11 as magenta wide frames. The Tspan33 peptide is represented by gray capped sticks and a transparent lipophilic surface, where the Tyr and Trp residues constitute a highly hydrophobic part of the peptide. On the right, molecular docking shows how the Tspan33 peptide perfectly fits the lipophilic surface. E and F, single and double mutants of the WW1 domain fail to bind to Tspan33. Immunoblot analysis of GST pulldowns using either GST or GST (G) fused to Tspan33 C terminus (G-T33-Cterm) as baits, and either PLEKHA7-WW(1 1 2)-Myc WT and mutant or GFP-Myc as preys. G and H, PDZD11 promotes binding of the single but not double mutants of WW1 1 WW2 to Tspan33. Immunoblot anal- ysis of GST pulldowns using either GST or GST-Tspan33 baits (indicated in red), with either WT or single or double mutant WW1 1 WW2 preys (green) Ponceau S-stained blots show the amounts of recombinant proteins used as bait. The third protein (either PDZD11-HA or CFP-HA, negative control) for tri-molecular pulldowns is shown in blue (normalization shown in H).

9308 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction be “locked” by direct interaction with (13). Here we PDZD11-HA (human) full-length, and constructs of pFAST- show a new mechanism, whereby the conformational change BAC containing PLEKHA7-His and constructs of pGEX-4T1 driven by PDZD11 binding to a specific binding pocket allos- containing GST alone, GST-PLEKHA7 (human) WW1 domain terically modifies the conformation of the WW1 domain, to (residues 1-56), WW1 1 WW2 domain (1-162), WW2 domain promote interaction with hydrophobic amino acids of a trans- (50-162), GST-PDZD11 full-length (residues 1-140), and GST- membrane ligand. It is interesting that PDZD11 binds to the C Tetraspanin 33 C-terminal (residues 256-283) were described terminus of the sodium-dependent multivitamin transporter, previously (13, 15). Mutations into alanine were performed and this latter is stabilized by Tetraspanin-1 (Tspan-1) (36). using Q5 Site-directed Mutagenesis Kit, following the manu- Future studies should investigate whether PLEKHA7 is facturer’s guidelines (New England BioLabs). Mutant involved in anchoring of the multivitamin transporter or other constructs of PLEKHA7 and PDZD11 were generated by PDZD11-binding transmembrane proteins (37, 38) to junc- PCR amplification on inserts of either GFP-PLEKHA7 full- tions, through mechanisms similar to the one described here. length in pTre2Hyg vector or GFP-PDZD11 full-length in In summary, we characterize the structural basis of the inter- pcDNA3.1 vector (15), using the appropriate oligonucleotides. action between the tandem PLEKHA7 domains and their inter- The sequence of all mutated constructs was verified, to rule out actions with both PDZD11 and Tspan33. Our model illustrates off-target mutations. For bacterial expression, mutated sequen- the mechanism through which the PLEKHA7-PDZD11 com- ces were subsequently amplified and cloned either into the plex forms a scaffold for a transmembrane ligand with hydro- BamHI-NotI sites (PLEKHA7) or EcoRI-NotI sites (PDZD11) Downloaded from phobic C-terminal sequences. In this model, cooperative inter- of pGEX-4T1 vector. Deletion constructs of PDZD11 and Tet- action between WW1 and WW2 promotes binding of WW1 to raspanin 33 C-terminal were generated by PCR amplification PDZD11, and this in turn promotes Tspan33 binding to a dif- with the appropriate oligonucleotides cloned either into the ferent surface of WW1. This mechanism may be relevant to the EcoRI-NotI sites (PDZD11) or BamHI-NotI sites (Tetraspanin

functions of additional proteins with tandem WW domains. 33 C-terminal) of pGEX-4T1 vector. http://www.jbc.org/

Experimental procedures Recombinant protein expression and GST pulldowns Cell culture and transfection For the production and purification of GST fusion protein baits, Escherichia coli (BL21-DE3) were induced with 0.1 mM Mouse cortical collecting duct cells (mCCD clone N64-Tet- isopropyl 1-thio-b-D-galactopyranoside (2 h at 37 °C). Bacterial off, a kind gift from Prof. Eric Feraille, University of Geneva, at Université de Genève on October 14, 2020 pellets after centrifugation were lysed in PBS containing 1% Switzerland) were grown in Dulbecco’s modified Eagle’s me- Triton X-100, 5 mg/ml of antipain, 5 mg/ml of leupeptin, 5 mg/ dium (Gibco) supplemented with 20% heat-inactivated fetal bo- ml of pepstatin, 1 mM phenylmethylsulfonyl fluoride, and cell vine serum (FBS; PAN Biotech), 13 minimal essential medium debris were removed by centrifugation at 13,000 rpm (15 min nonessential amino acids (NEAA; Gibco). PLEKHA7-KO at 4 °C). The supernatants containing GST-tagged proteins mCCD cells were obtained by CRISPR/Cas9 genomic editing were normalized for protein content by SDS-PAGE. For (13). HEK cells were cultured in the same medium, supple- expression of prey proteins in mammalian cells, 2 3 106 HEK mented with 10% FBS. Transfections were performed using cells were plated in a 100-mm2 dish and transfected with 20 mg Lipofectamine 2000, following the manufacturer’s guidelines of the HA-tagged construct, lysed 2 days after transfection (Thermo Fisher Scientific). using co-IP buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 1 mM EDTA). Expression of PLEKHA7 in Sf9 Antibodies insect cells (Sf9) was as described in Ref. 15. Prey proteins were Primary antibodies were: polyclonal guinea pig anti- normalized by immunoblotting with antibodies against tags PLEKHA7 (in-house gp2737, 1:500 IF); monoclonal mouse (HA, His, and Myc). For GST pulldowns 5 mg of bait GST anti-GFP (Roche Applied Science, 11814460001, 1:100 IF), pol- fusion protein was coupled for 1 h at room temperature to 10 yclonal rabbit anti-PDZD11 (in-house r29958 28, 1:50 IF), ml of GSH-Sepharose beads. Following incubation and 33 mouse anti-HA (Zymed Laboratories Inc., 32-6700, 1:1000 IB), washing with PBS containing 2% BSA and 1% Nonidet P-40, and mouse anti-His (Invitrogen, catalog number 37-2900, the beads were incubated for 1 h at 4 °C with normalized lysates 1:1500 IB). Secondary antibodies were anti-mouse Alexa Fluor of either HEK or insect cells. Proteins bound to the beads were 488, anti-rabbit Alexa Fluor 488, anti-rabbit Cy3, anti-mouse eluted with 20 ml of SDS sample buffer at 95 °C for 5 min, and Cy3, and anti-guinea pig Cy3 (Jackson ImmunoResearch 10 ml of eluate was loaded on SDS gels. Europe, 1:250 IF), and HRP-conjugated anti-mouse (Promega, 1:20000, IB). Immunofluorescence Cells were seeded at density of 40,000 cells/well onto 12-mm Plasmids and mutagenesis glass coverslips in 24-well-plates (FalconTM Polystyrene Micro- Constructs of pTre2-Hyg containing GFP-Myc, GFP- plates). 24 h after seeding cells were transfected, and 48 h later PLEKHA7 (human) full-length (residues 1-1121), PLEKHA7- they were fixed with methanol (6 min at 220 °C), rehydrated in WW(1 1 2)-Myc (residues 1-162), PLEKHA7-WW(2)-Myc PBS (33), permeabilized with 0.3% Triton X-100 in PBS for 5 (residues 50-162), GFP-PDZD11 (human) full-length (residues min, and incubated with 0.03% Triton X-100, 1% BSA, 0.2% gel- 1-140), and constructs of pcDNA3.1 containing CFP-HA, atin in PBS for 20 min. Cells were then incubated with primary

J. Biol. Chem. (2020) 295(28) 9299–9312 9309 Cooperative PLEKHA7-PDZD11-Tspan33 interaction antibodies (2 h at room temperature), washed with PBS, and NY) was used to perform the molecular docking. The calcula- incubated with 0.03% Triton X-100, 1% BSA, 0.2% gelatin in tion considers a predefined target grid-box, and in the case of PBS for 15 min and incubated with secondary antibodies (1 h, PDZD11 peptide docking we centered the box at the level of at room temperature), washed 23 with 0.03% Triton X-100, residues Asp-30, Arg-33, and Thr-35 for the WW1 domain, 13 with cold PBS, and 13 with water. Coverslips were and His-75 for the WW2 domain. In the case of Tspan33 dock- mounted with Vectashield containing DAPI (Reactolab). ing, the grid was centered in the lipophilic region of the WW1 domain considering the residues (Leu-11, Trp-15, Tyr-17, Phe- SDS-PAGE and immunoblotting 27, Leu-32, and Ile-38). Finally, the standard-precision mode For preparation of cell lysates, cells were washed with cold was considered in the case of docking single amino acid resi- PBS, lysed with co-IP buffer with Roche inhibitor (13)at4°C, dues and the SP peptide-mode was chosen for the peptides and incubated for 15 min with gentle agitation. Lysates were (Schrödinger Release 2016-4: Glide, Schrödinger, LLC, New sonicated 5 s at 66% power (3 bursts), centrifuged 20 min at York, NY). 13,000 rpm, and supernatants were recovered. Proteins were mixed with SDS sample buffer and incubated at 95 °C for 5 min. Molecular dynamic simulations Inputs of prey protein lysates and baits in bacterial lysates were normalized by analysis on SDS gels (8–12% acrylamide, 100 V). The Amber16 suite programs (21) was used to perform mo- For immunoblots, gels were transferred onto nitrocellulose lecular dynamic simulations, respectively, 160 ns for the Downloaded from – – (0.45 mM) (100 V for 80 min at 4 °C), and blots were incubated WW1 WW2 complex and 120 ns for the WW1 WW2- with primary antibody (anti-HA, anti-Myc, or anti-His), fol- PDZD11 complex, WW1–WW2 single mutants (T35A, D30A) lowed by secondary HRP-labeled antibody (1:20,000), and and WW1–WW2 double mutant (T35A,H75A and D30A, visualized with ECL (Amersham Biosciences). H75A) complex. The WW1–WW2 complex representing the

putative interacting mode of the two domains was chosen as a http://www.jbc.org/ Sequence analysis starting structure for the dynamic calculation. The Amber To analyze protein sequence conservation, the sequences of ff14SB force field was used to parametrize the two structures ’ the WW domain of PLEKHA7 or PDZD11 from all available using the Leap program from Amber s package. The systems vertebrate species (n 5 100) were retrieved with blastp, were then solvated with TIP3P water into a cubic box of 12 Å using either the human PLEKHA7 or human PDZD11 sequen- and neutralized with sodium as counter ion. The Build panel of at Université de Genève on October 14, 2020 ces as a query. Sequences were aligned using Weblogo (RRID: the Maestro interface (Schrödinger Release 2016-4: Maestro, SCR_010236). Schrödinger, LLC, New York, NY) was used to mutate the cor- responding residue. All complexes then underwent minimiza- Homology modeling tion, heating, and equilibration protocols: (i) a first 5000-step The Prime program from Schrödinger package (Prime, minimization step, where restrains were applied to the amino Schrödinger, LLC, New York, NY, 2016) was used to generate acid residues (steepest descendent and conjugated gradient the homology models of WW1 and WW2 domains. The selec- methods were applied), followed by a second 5000-step mini- tion of the corresponding template for building the models was mization where the entire system was allowed to move, using based on the sequence similarity study performed with BLAST the Sander program. (ii) During the heating phase the tempera- (basic local alignment search tool). The solution structure of ture increases from 0 to 298 K using the Langevin thermostat, -1 -2 Nedd4 WW domain (39) was chosen for building the WW1 do- and with 100 kcal mol Å positional restrains for protein and main, whereas the WW domain of the human syntaxin-binding ions over the course of the first 80 ps at constant pressure (1 protein 4 (40) was used for the WW2 model. Ramachandran atm). The restrained forces were gradually turned off to zero in plots were then used to assess the correct geometry of the the subsequent 100 ps, allowing movement of all atoms. (iii) A amino acids residues as confirmation of the quality of the final heating and equilibrating step was performed at 300 K at model (RRID:SCR_017590). constant pressure over 820 ps. Finally, the molecular dynamic production was performed under NTP ensemble (constant pres- Molecular docking sure, 1 atm; and temperature, 300 K) using the PMEMD Cuda. The homology models were submitted to the Protein Prepa- The SHAKE algorithm was used to restrain the lengths of all ration Wizard of Schrödinger tools package, where bond orders bonds involving hydrogen atoms, allowing an increase of the and atom types were assigned and also terminals were capped. time step up to 2 fs with a cut-off space of 9 Å. The default parti- The pKa values were predicted using PROPKA at neutral pH cle-mesh Ewald settings (which correspond to a grid spacing of 26 and finally an all-atoms restrain minimization with a maximum ;1 Å and a direct space tolerance of 10 ) were used to deter- RMSD of 0.3 Å was performed. The amino acids and the pep- mine long-range charge interactions. The CPPTRAJ of Amber16 tide sequences used for docking were prepared with the was used to process and analyze the molecular dynamic results. LigPrep tool that generates different energetically favored con- ThesurfaceareaattheinterfaceoftheWW1andWW2 formations, considering also the possible protonation states of domains was determined on the basis of the difference between the ligands. Glide program from Schrödinger package (Schrö- the solvent accessible surface areas of the two domains sepa- dinger Release 2016-4: Glide, Schrödinger, LLC, New York, rately (WW1 1 WW2) or in a complex (WW1–WW2).

9310 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative PLEKHA7-PDZD11-Tspan33 interaction

Data availability S. A., Cheng, C. Y., Xu, L., Jia, H., et al. (2012) Genome-wide association analyses identify three new susceptibility loci for primary angle closure Data and materials are available upon request from S.C. glaucoma. Nat. Genet. 44, 1142–1146 CrossRef ([email protected]), except for molecular docking and 7. Endres, B. T., Priestley, J. R., Palygin, O., Flister, M. J., Hoffman, M. J., molecular dynamic simulation datasets, SRR9822082 which Weinberg, B. D., Grzybowski, M., Lombard, J. H., Staruschenko, A., Mor- are available upon request from L.S. (leonardo.scapozza@ eno, C., Jacob, H. J., and Geurts, A. M. (2014) Mutation of Plekha7 attenu- unige.ch). ates salt-sensitive hypertension in the rat. Proc. Natl. Acad. Sci. U.S.A. 111, 12817–12822 CrossRef 8. Wythe, J. D., Jurynec, M. J., Urness, L. D., Jones, C. A., Sabeh, M. K., Wer- dich, A. A., Sato, M., Yost, H. J., Grunwald, D. J., Macrae, C. A., and Li, — Acknowledgments We thank all the colleagues cited in the text for D. Y. (2011) Hadp1, a newly identified pleckstrin homology domain pro- their kind gift of cell lines and reagents. tein, is required for cardiac contractility in zebrafish. Dis. Model Mech. 4, 607–621 CrossRef Author contributions—F.R., F.T., L.S., and S.C. conceptualization; 9. Kourtidis, A., Ngok, S. P., Pulimeno, P., Feathers, R. W., Carpio, L. R., F.R., F.T., L.A., L.S., and S.C. resources; F.R., F.T., L.A., and L.S. Baker, T. R., Carr, J. M., Yan, I. K., Borges, S., Perez, E. A., Storz, P., Cop- data curation; F.R., F.T., and L.S. formal analysis; F.R., L.S., and S.C. land, J. A., Patel, T., Thompson, E. A., Citi, S., et al. (2015) Distinct E-cad- supervision; F.R. and L.S. validation; F.R., F.T., L.A., L.S., and S.C. herin-based complexes regulate cell behaviour through miRNA processing or Src and p120 catenin activity. Nat. Cell Biol. 17, 1145–1157 investigation; F.R. and L.S. visualization; F.R. and S.C. writing-origi- CrossRef nal draft; F.R., L.S., and S.C. writing-review and editing; F.T., L.A., Downloaded from 10. Kourtidis, A., and Anastasiadis, P. Z. (2016) PLEKHA7 defines an apical and L.S. software; F.T. and L.S. methodology; L.S. and S.C. funding junctional complex with cytoskeletal associations and miRNA-mediated acquisition; S.C. project administration. growth implications. Cell Cycle 15, 498–505 CrossRef 11. Lee, M. C., Shei, W., Chan, A. S., Chua, B. T., Goh, S. R., Chong, Y. F., Funding and additional information—This work was supported by Hilmy, M. H., Nongpiur, M. E., Baskaran, M., Khor, C. C., Aung, T., Hun- ziker, W., and Vithana, E. N. (2017) Primary angle closure glaucoma

Swiss National Fund Grant 31003A_172809 (to S. C.). http://www.jbc.org/ (PACG) susceptibility gene PLEKHA7 encodes a novel Rac1/Cdc42 GAP — that modulates cell migration and blood-aqueous barrier function. Hum. Conflict of interest The authors declare that they have no conflict Mol. Genet. 26, 4011–4027 CrossRef of interest with the contents of this article. 12. Popov, L. M., Marceau, C. D., Starkl, P. M., Lumb, J. H., Shah, J., Guerrera, D., Cooper, R. L., Merakou, C., Bouley, D. M., Meng, W., Kiyonari, H., — Abbreviations The abbreviations used are: KO, knock-out; Takeichi, M., Galli, S. J., Bagnoli, F., Citi, S., et al. (2015) The adherens ADAM10, A Disintegrin and metalloproteinase domain-containing junctions control susceptibility to Staphylococcus aureus a-toxin. Proc. at Université de Genève on October 14, 2020 protein 10; CFP, cyan fluorescent protein; DAPI, 4’,6-diamidino-2-phe- Natl. Acad. Sci. U.S.A. 112, 14337–14342 CrossRef nylindole; FBS, fetal bovine serum; GST, glutathione S-transferase; HA, 13. Shah, J., Rouaud, F., Guerrera, D., Vasileva, E., Popov, L. M., Kelley, W. L., hemagglutinin; HRP, horseradish peroxidase; IB, immunoblot; IF, im- Rubinstein, E., Carette, J. E., Amieva, M. R., and Citi, S. (2018) A dock- munofluorescence; mCCD, mouse kidney collecting duct; Nedd4, neu- and-lock mechanism clusters ADAM10 at cell-cell junctions to promote ral precursor cell-expressed developmentally downregulated gene 4; a-toxin cytotoxicity. Cell Rep. 25, 2132–2147 e2137 CrossRef PDZ, Psd95-Dlg1-ZO-1 domain; PDZD11, PDZ domain-containing 14. Kurita, S., Yamada, T., Rikitsu, E., Ikeda, W., and Takai, Y. (2013) Binding protein 11; PLEKHA7, pleckstrin homology domain-containing family between the junctional proteins afadin and PLEKHA7 and implication in 288, A member 7; Sf, Spodoptera frugiferda; Tspan33, Tetraspanin 33; WW, the formation of adherens junction in epithelial cells. J. Biol. Chem. 29356–29368 CrossRef tryptophan-tryptophan; YFP, yellow fluorescent protein; MD, molecular 15. Guerrera, D., Shah, J., Vasileva, E., Sluysmans, S., Mean, I., Jond, L., Poser, dynamics; RMSD, root mean square deviation; HEK, human embryonic I., Mann, M., Hyman, A. A., and Citi, S. (2016) PLEKHA7 recruits kidney; ZO-1, zonula occludens-1. PDZD11 to adherens junctions to stabilize nectins. J. Biol. Chem. 291, 11016–11029 CrossRef 16. Sudol, M., Chen, H. I., Bougeret, C., Einbond, A., and Bork, P. (1995) Char- References acterization of a novel protein-binding module–the WW domain. FEBS – 1. Meng, W., Mushika, Y., Ichii, T., and Takeichi, M. (2008) Anchorage of Lett. 369, 67 71 CrossRef minus ends to adherens junctions regulates epithelial cell-cell 17. Macias, M. J., Gervais, V., Civera, C., and Oschkinat, H. (2000) Structural contacts. Cell 135, 948–959 CrossRef analysis of WW domains and design of a WW prototype. Nat. Struct. Biol. – 2. Pulimeno, P., Bauer, C., Stutz, J., and Citi, S. (2010) PLEKHA7 is an adhe- 7, 375 379 CrossRef rens junction protein with a tissue distribution and subcellular localization 18. Chang, H. T., Liu, C. C., Chen, S. T., Yap, Y. V., Chang, N. S., and Sze, C. I. distinct from ZO-1 and E-cadherin. PLoS ONE 5, e12207 CrossRef (2014) WW domain-containing oxidoreductase in neuronal injury and – 3. Pulimeno, P., Paschoud, S., and Citi, S. (2011) A role for ZO-1 and PLE- neurological diseases. Oncotarget 5, 11792 11799 CrossRef KHA7 in recruiting paracingulin to tight and adherens junctions of epithe- 19. Grusche, F. A., Richardson, H. E., and Harvey, K. F. (2010) Upstream regu- – lial cells. J. Biol. Chem. 286, 16743–16750 CrossRef lation of the hippo size control pathway. Curr. Biol. 20, R574 R582 4. Paschoud, S., Jond, L., Guerrera, D., and Citi, S. (2014) PLEKHA7 modu- CrossRef lates epithelial tight junction barrier function. Tissue Barriers 2, e28755 20. Ingham, R. J., Colwill, K., Howard, C., Dettwiler, S., Lim, C. S., Yu, J., Hersi, CrossRef K., Raaijmakers, J., Gish, G., Mbamalu, G., Taylor, L., Yeung, B., Vassilov- 5. Levy, D., Ehret, G. B., Rice, K., Verwoert, G. C., Launer, L. J., Dehghan, A., ski, G., Amin, M., Chen, F., et al. (2005) WW domains provide a platform Glazer, N. L., Morrison, A. C., Johnson, A. D., Aspelund, T., Aulchenko, for the assembly of multiprotein networks. Mol. Cell Biol. 25, 7092–7106 Y., Lumley, T., Köttgen, A., Vasan, R. S., Rivadeneira, F., et al. (2009) Ge- CrossRef nome-wide association study of blood pressure and hypertension. Nat. 21. Case, D. A., Betz, R. M., Cerutti, D. S., Cheatham, T. E., Darden, I. T. A., Genet. 41, 677–687 CrossRef Duke, R. E., Giese, T., Gohlke, J. H., Goetz, A., Homeyer, W., Izadi, N., 6. Vithana, E. N., Khor, C. C., Qiao, C., Nongpiur, M. E., George, R., Chen, Janowski, S., Kaus, P., Kovalenko, J. A., Lee, T. S., et al. (2016) AMBER L. J., Do, T., Abu-Amero, K., Huang, C. K., Low, S., Tajudin, L. S., Perera, 2016. University of California, San Francisco

J. Biol. Chem. (2020) 295(28) 9299–9312 9311 Cooperative PLEKHA7-PDZD11-Tspan33 interaction

22. Macias, M. J., Wiesner, S., and Sudol, M. (2002) WW and SH3 domains, 32. Chong, P. A., Lin, H., Wrana, J. L., and Forman-Kay, J. D. (2010) Coupling two different scaffolds to recognize proline-rich ligands. FEBS Lett. 513, of tandem Smad ubiquitination regulatory factor (Smurf) WW domains 30–37 CrossRef modulates target specificity. Proc. Natl. Acad. Sci. U.S.A. 107, 18404– 23. Tavano, S., Taverna, E., Kalebic, N., Haffner, C., Namba, T., Dahl, A., 18409 CrossRef Wilsch-Bräuninger, M., Paridaen, J., and Huttner, W. B. (2018) Insm1 33. Ramirez-Espain, X., Ruiz, L., Martin-Malpartida, P., Oschkinat, H., and induces neural progenitor delamination in developing neocortex via Macias, M. J. (2007) Structural characterization of a new binding motif downregulation of the adherens junction belt-specific protein Plekha7. and a novel binding mode in group 2 WW domains. J. Mol. Biol. 373, – Neuron 97, 1299 1314.e8 CrossRef 1255–1268 CrossRef 24. Sudol, M., Recinos, C. C., Abraczinskas, J., Humbert, J., and Farooq, A. 34. Li, Y., Zhou, Z., Alimandi, M., and Chen, C. (2009) WW domain contain- (2005) WW or WoW: the WW domains in a union of bliss. IUBMB Life ing E3 ubiquitin protein ligase 1 targets the full-length ErbB4 for ubiqui- – 57, 773 778 CrossRef tin-mediated degradation in . Oncogene 28, 2948–2958 25. Dodson, E. J., Fishbain-Yoskovitz, V., Rotem-Bamberger, S., and Schueler- CrossRef Furman, O. (2015) Versatile communication strategies among tandem 35. Young, L., Jernigan, R. L., and Covell, D. G. (2008) A role for surface hydro- WW domain repeats. Exp. Biol. Med. (Maywood) 240, 351–360 CrossRef phobicity in protein-protein recognition. Protein Sci. 3, 717–729 CrossRef 26. Fedoroff, O. Y., Townson, S. A., Golovanov, A. P., Baron, M., and Avis, 36. Nabokina, S. M., Senthilkumar, S. R., and Said, H. M. (2011) Tspan-1 inter- J. M. (2004) The structure and dynamics of tandem WW domains in a acts with the thiamine transporter-1 in human intestinal epithelial cells negative regulator of notch signaling, Suppressor of deltex. J. Biol. Chem. and modulates its stability. Am. J. Physiol. Gastrointest. Liver Physiol. 301, 279, 34991–35000 CrossRef G808–G813 CrossRef 27. Ingham, R. J., Gish, G., and Pawson, T. (2004) The Nedd4 family of E3 37. Goellner, G. M., DeMarco, S. J., and Strehler, E. E. (2003) Characterization

ubiquitin ligases: functional diversity within a common modular architec- Downloaded from of PISP, a novel single-PDZ protein that binds to all plasma membrane ture. Oncogene 23, 1972–1984 CrossRef 21 – 28. Zweifel, M. E., Leahy, D. J., and Barrick, D. (2005) Structure and Notch re- Ca -ATPase b-splice variants. Ann. N.Y. Acad. Sci. 986, 461 471 ceptor binding of the tandem WWE domain of Deltex. Structure 13, CrossRef 1599–1611 CrossRef 38. Stephenson, S. E., Dubach, D., Lim, C. M., Mercer, J. F., and La Fontaine, S. 29. Huang, X., Beullens, M., Zhang, J., Zhou, Y., Nicolaescu, E., Lesage, B., Hu, (2005) A single PDZ domain protein interacts with the Menkes copper ATPase, ATP7A: a new protein implicated in copper homeostasis. J. Biol.

Q., Wu, J., Bollen, M., and Shi, Y. (2009) Structure and function of the two http://www.jbc.org/ – tandem WW domains of the pre-mRNA splicing factor FBP21 (formin- Chem. 280, 33270 33279 CrossRef binding protein 21). J. Biol. Chem. 284, 25375–25387 CrossRef 39. Kanelis, V., Rotin, D., and Forman-Kay, J.D. (2001) Domain solution struc- 30. Bhattacharya, S., Zoladek, T., and Haines, D. S. (2008) WW domains 2 and ture of the rNEDD4 WWIII domain-renal BP2 peptide complex. Protein 3 of Rsp5p play overlapping roles in binding to the LPKY motif of Spt23p Data Bank, CrossRef and Mga2p. Int. J. Biochem. Cell Biol. 40, 147–157 CrossRef 40. Ohnishi, S., Tochio, N., Sato, M., Koshiba, S., Harada, T., Watanabe, S., 31. Kanelis, V., Rotin, D., and Forman-Kay, J. D. (2001) Solution structure of a Kigawa, T., Yokoyama, S., RIKEN Structural Genomics/Proteomics Initia- Nedd4 WW domain-ENaC peptide complex. Nat. Struct. Biol. 8, 407–412 tive (RSGI) (2007) Solution structure of the WW domain from the human at Université de Genève on October 14, 2020 CrossRef syntaxin-binding protein 4. , CrossRef

9312 J. Biol. Chem. (2020) 295(28) 9299–9312 Cooperative binding of the tandem WW domains of PLEKHA7 to PDZD11 promotes conformation-dependent interaction with tetraspanin 33 Florian Rouaud, Francesca Tessaro, Laura Aimaretti, Leonardo Scapozza and Sandra Citi J. Biol. Chem. 2020, 295:9299-9312. doi: 10.1074/jbc.RA120.012987 originally published online May 5, 2020

Access the most updated version of this article at doi: 10.1074/jbc.RA120.012987

Alerts:

• When this article is cited Downloaded from • When a correction for this article is posted

Click here to choose from all of JBC's e-mail alerts

This article cites 39 references, 11 of which can be accessed free at http://www.jbc.org/ http://www.jbc.org/content/295/28/9299.full.html#ref-list-1

at Université de Genève on October 14, 2020 A 60 WW2 Domain PLEKHA7 87 D GFP PDZD11 Merged

-WEEGFTEEGASYFIDHNQQTTAFRHP- G59A Usage frequency

B GFP PDZD11 Merged W60A

+GFP E61A

WT T65A *

*

Δ WW1

* G68A GFP-PLEKHA7-FL * * *

Δ WW(1+ 2)

Y71A +GFP-PLEKHA7-FL

C

1 WW1 Domain PLEKHA7 42 I73A MAATVGRDTLPEHWSYGVCRDGRVFFINDQLRCTTWLHPRTGEPV- 54 87 -NSGHMIRSDLPRGWEEGFTEEGASYFIDHNQQTTAFRHPVTQFSPE-

A A A A A A A AA AA H75A 59 60 61 65 68 71 73 7576 8485 WW2 Domain PLEKHA7

GST-PLEKHA7-WW(1+2) N76A

kDa Input G WT G59A W60A E61A T65A G68A Y71A I73A H75A N76A H84A P85A 15

IB:HA H84A PDZD11-HA 25 Ponceau

25 IB:HA P85A CFP-HA

25 Ponceau

Supporting Information Figure 1 SUPPORTING INFORMATION FIGURE 1. Mutations within the WW2 domain of PLEKHA7 do not affect its interaction with PDZD11. A. Top: sequence of the WW2 domain of PLEKHA7, and bottom: Weblogo diagram of residue conservation. B. Requirement for the WW1 domain for PLEKHA7 interaction with PDZD11 in cells. Immunofluorescence localization of endogenous PDZD11 in PLEKHA7-KO mCCD cells rescued either with GFP, or WT GFP-PLEKHA7, or GFP-PLEKHA7 with N-terminal deletion of either WW1 (DWW1) or WW1+WW2 (DWW(1+2) domains). C. Top: sequence of WW1+WW2 domains, with mutations of highly conserved residues of WW2 (to Ala) highlighted in yellow. The numbers below each residue indicate residue number in the sequence. Bottom: immunoblot analysis of GST pulldowns using either GST (G), or GST fused to either WT or mutant WW1+WW2 domains as bait, and either PDZD11-HA or CFP-HA as preys. D. Immunofluorescence localization of endogenous PDZD11 in PLEKHA7-KO mCCD cells rescued with WW2 point mutants of GFP-tagged full-length PLEKHA7. Merged images show nuclei in blue (DAPI) and arrows indicate junctional labeling and asterisk reduced/undetectable labeling. Bar= 20 µm. 35 A A B 1 WW1 Domain PLEKHA7 42 +GFP-PLEKHA7-FL MAATVGRDTLPEH LHPRTGEPV- WSYGVCRDGRVFFINDQLRCTTW T35A+G59A T35A+W60A T35A+E61A T35A+T65A T35A+G68A T35A+Y71A T35A+I73A T35A+H84A T35A+P85A 54 87 -NSGHMIRSDLPRGWEEGFTEEGASYFIDHNQQTTAFRHPVTQFSPE-

A A A A A A A AA AA GFP 59 60 61 65 68 71 73 7576 8485 WW2 Domain PLEKHA7 GST-PLEKHA7-WW(1+2) PDZD11 kDa Input GST WT T35A T35A+G59AT35A+W60AT35A+E61AT35A+T65AT35A+G68AT35A+Y71AT35A+H75AT35A+N76AT35A+H84AT35A+P85A 15 IB:HA Merged PDZD11-HA

25 Ponceau

25 IB:HA CFP-HA

25 Ponceau

30 C A F 1 WW1 Domain PLEKHA7 42 +GFP-PLEKHA7-FL MAATVGRDTLPEH SYGVCRDGRVFFINDQLRCTT LHPRTGEPV- W W R33A D74A N76A D30A+D74A D30A+N76A R33A+D74A R33A+H75A R33A+N76A T35A+D74A T35A+N76A 54 87 -NSGHMIRSDLPRGWEEGFTEEGASYFIDHNQQTTAFRHPVTQFSPE- WW2 Domain PLEKHA7 AAA GFP 747576 GST-PLEKHA7-WW(1+2)

kDa Input GST WT D30A D74A H75A N76A D30A+D74AD30A+H75AD30A+N76A PDZD11 IB:HA 15 PDZD11-HA Ponceau 25 Merged 25 IB:HA CFP-HA

Ponceau G 25

35 33 D E A A WW1 Domain PLEKHA7 1 WW1 Domain PLEKHA7 42 1 42 Input: MAATVGRDTLPEHWSYGVCRDGRVFFINDQLRCTTWLHPRTGEPV- MAATVGRDTLPEHWSYGVCRDGRVFFINDQLRCTTWLHPRTGEPV- kDa P7-WW(1+2)G -MycG-P11 P7-WW(1+2)-MycG G-P11 D30AP7-WW(1+2)G -MycG-P11 T35AP7-WW(1+2)G -MycG-P11 H75A 54 87 54 87 20 IB:Myc -NSGHMIRSDLPRGWEEGFTEEGASYFIDHNQQTTAFRHPVTQFSPE- -NSGHMIRSDLPRGWEEGFTEEGASYFIDHNQQTTAFRHPVTQFSPE- 35 WW2 Domain PLEKHA7 WW2 Domain PLEKHA7 AAA AAA 747576 747576

GST-PLEKHA7-WW(1+2) GST-PLEKHA7-WW(1+2) 25 Ponceau H

kDa Input GST WT T35A D74A H75A N76A T35A+D74AT35A+H75AT35A+N76A kDa Input GST WT R33A D74A H75A N76A R33A+D74AR33A+H75AR33A+N76A IB:HA IB:HA 15 15

Input: PDZD11-HA Ponceau

Ponceau GFP-MycG G-P11 P7-WW(1+2)-MycG G-P11 P7-WW(1+2)G -MycG-P11 D30A+H75AP7-WW(1+2)G -MycG-P11 T35A+H75A PDZD11-HA 25 25 kDa 25 25 IB:HA 25 IB:HA IB:Myc 20

35 CFP-HA CFP-HA Ponceau

Ponceau 25 25 25 Ponceau

Supporting Information Figure 2 SUPPORTING INFORMATION FIGURE 2. (A, C, D, E) Top: sequence of WW(1+2) domains, with summary of each individual mutation within the WW2 domain used in different pulldowns, highlighted in yellow. The numbers below each residue indicate number in the sequence. Bottom: immunoblot analysis of GST pulldowns using either GST (G), or GST fused to WT and to each combination of WW1 and WW2 mutation as baits, and either CFP-HA or PDZD11-HA as preys. B, F. Immunofluorescence localization of endogenous PDZD11 in PLEKHA7-KO mCCD cells rescued either with GFP, or with either WT or the indicated combinations of point mutants within the WW1 and WW2 domains of GFP-tagged full-length PLEKHA7. Merged images show nuclei in blue (DAPI). Arrows indicate junctional labeling. Bar= 20 µm. G-H Single but not double mutants preys of WW1 and WW2 bind to the PDZD11 bait. Immunoblot analysis of GST pulldowns using either GST (G), or GST fused to full-length PDZD11 as baits, and either WT or indicated mutants of PLEKHA7-WW(1+2)-Myc or GFP- Myc (negative controls) as preys. (G) shows single mutant preys and (H) shows double mutant preys. A B

WW1 WW2

PLEKHA7 WW1 domain --LPEHWSYGVCRDGRVFFINDQLRCTTWLHPR- PLEKHA7 WW2 domain --LPRGWEEGFTEEGASYFIDHNQQTTAFRHPVT-

Nedd4 WW domain GPLPPGWEERTHTDGRVFFINHNIKKTQWEDPRM Syntaxin-binding protein 4 SGLPYGWEEAYTAGGIKYFINHVTQTTSWIHPVMS

C D 180 180

Ψ 0 Ψ 0

180 180 180 0 180 180 0 180 Φ Φ General/Pre-Pro/Proline Favoured General/Pre-Pro/Proline Allowed General/Pre-Pro/Proline Favoured General/Pre-Pro/Proline Allowed Glycine Favoured Glycine Allowed Glycine Favoured Glycine Allowed

E F 20 10 WW(1+2)-T35A+H75A WW(1+2)-D30A+H75A WW1 WW2 WW(1+2)-D30A 8 15 WW(1+2)-T35A WW(1+2)-WT 6

10 RMSF (Å)

RMSD (Å) 4

5 2

0 0 0 50 100 0 10 20 30 40 50 60 Time (ns) Residues WSYGVCRDGRVFFINDQLRCTTW WEEGFTEEGASYFIDHNQQTTAF PLEKHA7 WW1 PLEKHA7 WW2

G

D30 D30 WW1 WW1 WW1

R33 T35 T35 D30 R33 R33 T35 H75 H75

H75

WW2 WW2 WW2

Cluster 1 Cluster 2 Cluster 3

Supporting Information Figure 3 SUPPORTING INFORMATION FIGURE 3. Mutations D30A+H75A and T35A+H75A destabilize the triple b-sheet structure of the WW (1+2) domains. A. Top: Homology model of the WW1 domain of PLEKHA7, based ubiquitin ligase nedd4 (PDB code: 1I5H_W) as a template. Bottom: amino acid sequence alignment between the PLEKHA7-WW1 and the template. Identical residues are in red, and position of b-sheet secondary structure is indicated by arrows. B. Top: Homology model of WW2 domain based on syntaxin-binding protein (PDB code: 2YSG) as a template. Bottom: amino acid sequence alignment between the model and the template. C-D. Ramachandran plots of WW1 (C) and WW2 (D) domain models. For WW1, all amino acids are in the favored and allowed regions, respectively 25 and 3 residues, and only Glu at position 3 at the N-terminal is found in the outlier region. For WW2, 25 amino acids are in the favored regions and 4 in the allowed regions. E. Comparison of the RMSD values (120 ns trajectory) between the WT WW1-WW2 complex (black tracing) and the mutants D30A (red tracing), T35A (orange tracing), D30A+H75A (green tracing) and T35A-H75A (blue tracing). The x-axis represent the time-scale (120 ns) whereas in the y-axis there are the deviation values express in Å. F. Root mean square fluctuations (RMSF) of the WW1-WW2 complex, with the x-axis showing residue numbers of the two domains, and the Y-axis fluctuation values expressed in Å. G. Representative structures extracted from the cluster analysis of the 120 ns trajectory of WW1-WW2 complex. Cluster3 is the most populated cluster with an occurrence during the simulating time of 50%, whereas cluster 1 and 2 are presented respectively 18% and 19%. The remaining 13% of the clustered trajectory is composed by smaller clusters appearing with % of occurrence less than 4%. A B

kDa Input G G-P11 G-P7-WW1G-P7-WW2G-P7-WW(1+2) Input PDZD11-HAG G-P7-WW1G G-P7-WW1G G-P7-WW1InputG CFP-HAG-P7-WW1G G-P7-WW1G G-P7-WW1 20 IB:Myc GFP-Myc - - - + + - - - - - + + - - P7-WW2-Myc - - - - - + + - - - - - + + 35 kDa 25 20 IB:HA PLEKHA7- Ponceau WW(1+2)-Myc 25 15 15 IB:Myc Ponceau 25 35 WW2-Myc PLEKHA7-

25 Ponceau C Input: 25 IB:Myc kDa GFP-MycP7-WW2-Myc 25 35 20 IB:Myc GFP-Myc Ponceau 25 15

Supporting Information Figure 4 SUPPORTING INFORMATION FIGURE 4. Lack of interaction between WW1 and WW2 domains of PLEKHA7. (A). Immunoblot analyses of GST pulldowns using either GST (G) or GST fused to either WW1 (G-P7-WW1) or WW2 (G-P7-WW2), or both (G-P7-W(1+2) as baits, and either both WW1 and WW2 domains or only WW2 as preys. Preys were tagged with myc at the C-terminus. (B-C) Immunoblot analyses of GST pulldowns using either GST (G) or GST fused to WW1 (G-P7-WW1) as baits, and either PDZD11-HA or CFP-HA as a prey, either in the presence or in the absence of a third protein (either GFP-myc or the WW2 domain) as a third protein (trimolecular pulldown, third protein shown in blue). All preys were tagged with myc at the C-terminus. (C) shows normalization of third protein.

35 A 75 B Usage all WW frequency

Consensus : GPLPPGWEER+DPDGGRVYYVNHNTRTTTWEDPRL Percent conservation: 60.2% 66%

C Usage frequency tandem WW Consensus: GPLPPGWEER+DPDG+RVYYVDHNTRTTTWEDPRL Percent conservation: 68% 71.5%

D

E 50 Distance PLEKHA7(WW2)-PDZD11 Distance PLEKHA7(WW1)-PDZD11 40

30

20

Center of mass distance (Å) 10

0 0 50 100 Time (ns)

Supporting Information Figure 5 SUPPORTING INFORMATION FIGURE 5. Evolutionary conservation of T35 and H75 PLEKHA7 residues in single and tandem WW domains. A. The WW domains of human PLEKHA7 (in a blue box) (9-42, WW1 and 54-87, WW2) were aligned with 86 WW domains of human proteins. Light/dark grey highlighting indicates identical/similar residues. Conserved threonine residues at position 28 of the first WW in tandem WW domains are in a red box. Conserved histidine residues at position 22 of the second (or third) tandem WW domain are in a green box. B-C. Weblogo consensus sequences and usage frequency of single (B) and tandem (C) WW domains. The percent conservation of Thr at position 28 and His at position 22 is indicated. D. Two hypothetical (either pink or purple in the superposition model) modes of interaction between WW1-WW2 and PDZD11. The WW1 and WW2 domains are shown in grey as new cartoon style and the PDZD11 heptapeptide (PAYENPP) in magenta as licorice style. The image was generated with VMD software. E. The central mass distance calculated between PDZD11 and respectively versus WW1 (grey line) and WW2 (orange line) domains are shown as a measure of binding stability. The x-axis shows the simulated time-scale (120 ns) and the y-axis the distances values, expressed in Å.

W1W2 complex Ave RMSD (Å)

WT 6.2 ± 0.81 D30A 6.5 ± 0.90 T35A 8.3 ± 1.02 D30A + H75A 9.2 ± 1.50 T35A + H75A 11.0 ± 1.92

SUPPORTING INFORMATION TABLE 1. The average values of the root-mean square deviations are summarized in the table for each simulated system.

Supporting Information Table 2

DONOR ACCEPTOR OCCUPANCY (%) ARG33-Side GLU66-Side 7.74% VAL19-Main GLN77-Main 4.16% ARG33-Side PHE64-Main 0.42% GLN78-Side TYR17-Main 4.91% SER16-Side GLN78-Side 0.08% ARG33-Side TYR71-Side 0.17% ARG21-Side THR80-Main 0.67% THR79-Side VAL19-Main 0.08% GLN77-Side TYR17-Side 0.08% CYS20-Side THR79-Side 0.17% GLN77-Side VAL19-Main 0.25% TYR71-Side ARG33-Main 0.08% THR35-Side TYR71-Side 0.17% CYS20-Side THR80-Side 0.50% PHE64-Main ARG33-Main 23.48% THR36-Side GLU66-Side 38.38% THR35-Main PHE64-Main 7.74% THR35-Side PHE64-Main 32.06% ARG33-Side GLU62-Main 13.74% ARG24-Side GLU67-Side 19.40% LEU38-Main GLU66-Side 0.67% TRP37-Side GLU67-Side 0.92% TRP37-Main GLU66-Side 1.92% ARG21-Main THR80-Side 0.08% ARG33-Side GLU61-Side 19.73% TYR71-Side THR35-Side 24.98% ARG33-Side GLU62-Side 3.58% CYS20-Side TYR71-Side 0.75% TYR71-Side CYS20-Side 0.25% THR80-Side VAL19-Main 2.50% THR80-Main VAL19-Main 0.33% ARG21-Main THR80-Main 1.08% ARG11-Side GLN53-Side 0.58% GLN78-Side VAL19-Main 1.50% ARG21-Side GLN77-Main 1.17% ARG21-Side THR79-Side 3.66% CYS20-Side GLN78-Main 0.08% ARG24-Side THR80-Side 2.00%

1 Supporting Information Table 2

ARG21-Main GLN78-Main 4.08% TRP37-Side GLU66-Side 0.17% CYS34-Side THR65-Main 0.08% CYS20-Side GLN78-Side 0.08% GLU66-Main THR35-Main 12.82% ARG21-Side ASN76-Side 0.08% ARG21-Main GLN78-Side 0.08% ARG21-Side ASN76-Main 0.08% ARG21-Side HIP75-Main 0.17% GLN78-Side ASP22-Side 0.33% ARG21-Side GLN78-Side 0.08% ARG21-Side GLN78-Main 0.25% ARG21-Side TYR71-Side 0.08% ASP22-Main GLN78-Side 0.08% THR80-Side ASP22-Side 1.08% ARG21-Side THR65-Side 1.33% THR65-Side THR35-Side 0.08% CYS34-Side GLU66-Side 0.58% ARG21-Side GLU67-Side 30.97% VAL19-Main TYR71-Side 9.66% SER16-Side GLU61-Side 10.32% GLN78-Side GLY18-Main 0.42% CYS20-Main GLN78-Side 0.08% ARG21-Side GLU66-Side 0.08% TYR17-Side GLN78-Side 0.08%

SUPPORTING INFORMATION TABLE 2. The hydrogen bond frequency occurrence (%) calculated from the WW1-WW2 trajectory. The interactions monitored occurred between the residues of the WW1 and the WW2 domain, with a cut-off distance of 4Å. Donors and acceptors residues are summarized in the table where for each interaction an average value of the respective occurrence, calculated along the trajectory, is reported.

2