Structural Basis of the Avian Influenza NS1 Protein Interactions with the Cell Polarity Regulator Scribble

Scribble is a highly conserved regulator of cell polarity, a process that enables the generation of asymmetry at the cellular and tissue level in higher organisms. Scribble acts in concert with Disc-large (Dlg) and Lethal-2-giant larvae (Lgl) to form the Scribble polarity complex, and its functional dysregulation is associated with poor prognosis during viral infections. Viruses have been shown to interfere with Scribble by targeting Scribble PDZ domains to subvert the network of interactions that enable normal control of cell polarity via Scribble, as well as the localisation of the Scribble module within the cell. The influenza A virus NS1 protein was shown to bind to human Scribble (SCRIB) via its C-terminal PDZ binding motif (PBM). It was reported that the PBM sequence ESEV is a virulence determinant for influenza A virus H5N1 whilst other sequences, such as ESKV, KSEV and RSKV, demonstrated no affinity towards Scribble. We now show, using isothermal titration calorimetry (ITC), that ESKV and KSEV bind to SCRIB PDZ domains and that ESEV unexpectedly displayed an affinity towards all four PDZs and not just a selected few. We then define the structural basis for the interactions of SCRIB PDZ1 domain with ESEV and ESKV PBM motifs, as well as SCRIB PDZ3 with the ESKV PBM motif. These findings will serve as a platform for understanding the role of Scribble PDZ domains and their interactions with different NS1 PBMs and the mechanisms that mediate cell polarity within the context of the pathogenesis of influenza A virus.


Introduction
Cell polarity refers to the differential distribution of macromolecules within a cell that allows it to orientate itself in a specific direction, which is an essential property for correct tissue development, organisation and function [1]. Examples of cell polarity include the axonal-dendritic directionality of neurons, the migration and asymmetric division of mesenchymal cells, and the apical-basal orientation or planar polarity of epithelial cells [2][3][4][5][6]. These phenomena can be categorised into the four main types of cell polarity as follows: (1) asymmetric cell division (ACD), (2) planar cell polarity (PCP), (3) apical-basal cell polarity (ABCP) and (4) front-rear cell polarity (FRCP) [2][3][4][5][6]. The proper establishment of cell polarity is crucial for normal cell homeostasis. With the loss of cell polarity, tissue becomes disorganised and excessive cell proliferation can present itself as a hallmark of cancer or viral replication [5,7]. For apical-basal, migrating and asymmetrical dividing polarity cell types three key protein complexes are involved in an antagonistic relationship to control cell polarity: the PAR module, the Crumbs module and the Scribble module [8]. These polarity regulatory modules play important roles in determining cell architecture and are highly conserved in higher organisms. sequence variation has been seen at the NS1 C-terminus, including premature termination, which yields NS1 proteins that lack PBM sequences. For avian influenza A NS1,~90% of isolates contain an ESKV or ESEV PBM motif at the C-terminus [25], whereas in human influenza A the RSKV PBM motif is more widely found [26]. Studies using pull-down assays have shown that ESEV PBM of the avian influenza A virus H5N1 NS1 protein binds to the PDZ domains of Scribble, Dlg1 and various other PDZ domain proteins (MAGI-1, MAGI-2 and MAGI-3); however, it has not been clarified which specific PDZ domain(s) are involved in the interaction [17]. The disruption of Scribble, and other PDZ domains containing polarity proteins, also disables tight junctions (TJ) formation and supports viral replication, by enhancing dissemination to host cells, and viral spread [17]. The severity of the H5N1 influenza A virus can range from diarrhoea, encephalitis, cytokine storm, oedema within the lungs to organ failure and death [27]. Understanding the mechanism that underlies the avian influenza A virus H5N1 NS1 interactions and disruption of PDZ domain mediated signalling will enable a better understanding of the pathogenesis of H5N1.
The transformed E. coli BL21 (DE3) pLysS cells were grown in 2YT media (1.6% w/v Tryptone, 1.0% w/v Yeast Extract and 0.5% w/v sodium chloride (NaCl)), supplemented with 100 µg/mL ampicillin; cells were incubated at 37 • C and shaken at 160 rpm for approximately 20 h. Expression of all PDZ constructs was induced using an autoinduction protein expression protocol by growing the small-scale bacterial culture in 2YT media with 100 mM NaCl, 10 mM Tris pH 7.6, 1 mM MgSO 4 , 5052 solution (25% glycerol, 10% lactose and 2.5% glucose) and 100 µg/mL ampicillin, which was used for a large-scale cell culture by incubated shaking at 37 • C 160 rpm for approximately 4 h and then at 16 • C for at least 20 h [31]. For His-MBP-tagged affinity purification, proteins were purified via immobilised metal affinity chromatography (IMAC), using HisTrap HP columns (GE Healthcare, Chicago, IL, USA). GST-tagged proteins were purified using glutathione Sepharose 4B resins (GE Healthcare) equilibrated with GST purification buffer. All proteins were purified as previously described [28][29][30].

Circular Dichroism (CD) Spectroscopy
All proteins were diluted to 0.150 mg/mL in phosphate buffer (50 mM phosphate) and subjected to CD spectroscopy using the AVIV model 420 CD spectrometer at a wavelength from 190 to 260 nm with sample entry every 1 nm and an averaging time of 4 s at 25 • C. Data were processed using the AVIV Biomedical software and plotted using Microsoft Excel.

Isothermal Titration Calorimetry (ITC)
Binding of the NS1 peptides (Mimotopes Pty Ltd., Melbourne, Australia) to SCRIB PDZ constructs was determined by ITC with a range of influenza A virus strain PBM peptide sequences (Table 1) [17,32]. The EMAGTIRSEV peptide was from the A/Puerto Rico/8/1943/H1N1 strain (Uniprot accession code P03496, residues 221-230) (Mimotopes). As a negative control, the KMERTIEPEV sequence, which is not a PBM sequence and was obtained from the H5N1 A/Hong Kong/156/97 strain (Uniprot accession code O56264, residues 221-230) (Mimotopes), was used [17]. The NS1 peptide sequences are distinguished based on the last four C-terminal residues of their PBM. The superpeptide was originally identified via the screening of a phage-displayed peptide library and was shown to harbour affinity towards a wide range of different PDZ domains [33] including those from polarity regulatory proteins [28,34]. These specific sequences, similarities and chemical composition of the peptides used throughout the study can be seen in Figure 1. ). An * (asterisk) indicates positions which have a single, fully conserved residue; . (period) indicates conservation between groups of weakly similar properties and : (colon) means that conserved substitutions are observed. Residues coloured in red are physiochemically small and hydrophobic (i.e., AVFPMILW). Blue is acidic (i.e., DE), Magenta is basic (i.e., RK). Green is an amine (i.e., STYHCNGQ).

NS1 Peptide Origin
Peptide Sequence  [17,32]. The EMAGTIRSEV peptide was from the A/Puerto Rico/8/1943/H1N1 strain (Uniprot accession code P03496, residues 221-230) (Mimotopes). As a negative control, the KMERTIEPEV sequence, which is not a PBM sequence and was obtained from the H5N1 A/Hong Kong/156/97 strain (Uniprot accession code O56264, residues 221-230) (Mimotopes), was used [17]. The NS1 peptide sequences are distinguished based on the last four C-terminal residues of their PBM. The superpeptide was originally identified via the screening of a phage-displayed peptide library and was shown to harbour affinity towards a wide range of different PDZ domains [33] including those from polarity regulatory proteins [28,34]. These specific sequences, similarities and chemical composition of the peptides used throughout the study can be seen in Figure 1.  ). An * (asterisk) indicates positions which have a single, fully conserved residue; . (period) indicates conservation between groups of weakly similar properties and : (colon) means that conserved substitutions are observed. Residues coloured in red are physiochemically small and hydrophobic (i.e., AVFPMILW). Blue is acidic (i.e., DE), Magenta is basic (i.e., RK). Green is an amine (i.e., STYHCNGQ).

NS1 Peptide Origin
Peptide Sequence ITC was conducted using a MicroCal iTC200 System (GE Healthcare) at 25 • C as previously described [28]. Peptides and PDZ domain proteins were diluted in their respective buffer (either Tris Buffer or HEPES Buffer) to concentrations of 900 µM and~75 µM respectively. Data were processed with the Origin 7.0 software (OriginLab Corporation, Northampton, MA, USA) using a "one-site binding model" to compute the binding affinities and stoichiometries.

Crystallisation and Structure Determination
Complexes of SCRIB PDZ domains (Table 2) were reconstituted as described (see also [35]). Briefly, proteins and 8-mer peptides were mixed at a 1:4 molar ratio and then concentrated using 3 kDa cut-off Ultra-0.5 centrifugal filter units (Millipore, Burlington, MA, USA). Concentrated protein complexes were subjected to high-throughput crystallisation screening using a Gryphon LCP (Art Robbins Instruments, Sunnyvale, CA, USA) in-house at 20 • C and prepared as drops of 0.2 µL protein and 0.2 µL reservoir solution in 96-well sitting drop crystallisation plates (Swissci, High Wycombe, UK) with commercial sparse matrix screens; JCSG-plus: HT-96 sparse matrix screen (Molecular Dimensions, Altamonte Springs, FL, USA), ShotGun sparse matrix screen (Molecular Dimensions), PACT Premier HT-96 sparse matrix screen (Molecular Dimensions), Structure Screen 1 + 2 sparse matrix screen (Molecular Dimensions) and Salt Rx sparse matrix screen (Hampton Research, Aliso Viejo, CA, USA). Crystals were mounted on nylon and copper loops (MiTGen, Ithaca, NY, USA). All data were collected at the Australian Synchrotron using the MX2 beamline equipped with the Eiger 16M detector (Dectris, Baden, Switzerland) with an oscillation range 0.1 • per frame with a wavelength of 0.9537, integrated using XDS [36] and scaled using AIMLESS [37]. Table 2. Interactions of SCRIB PDZ domains with influenza A virus_NS1 peptides. All affinities were measured using isothermal titration calorimetry, with K D values given in µM as a mean of three independent experiments with SD. NB stands for 'no binding' and N is the average stoichiometry. NB = no binding. $ = negative control. Mutants were not measured for PBM peptides that did not bind to WT SCRIB PDZ domains.   [38] with the previously solved structure of SCRIB PDZ1:b-PIX (PDB ID: 5VWK) [28] as a search model. SCRIB PDZ1:ESEV crystals contained 2 molecules each of SCRIB PDZ1 and ESEV peptide in the asymmetric unit, with 26.81% solvent content. The final TFZ and LLG values after molecular replacement were 22.7 and 919.2 respectively. The final model was built manually over several cycles using Coot [39] and refined using PHENIX with a final R work /R free of 0.25/0.28, with 96.83% of residues in the favoured region of the Ramachandran plot and 0% of rotamer outliers.

Biochemical Characterisation of SCRIB PDZ Domains and NS1 Peptide Interactions
NS1 interactions with SCRIB PDZ domains have been previously described in the literature [17,22,32] using immunofluorescence analysis, co-immunoprecipitation and GST pull-down assays. Although these experiments provided information regarding the relative binding of some of the PBM sequences to Scribble PDZ domains, no affinity measurements had been performed. To investigate these interactions and define their affinities, SCRIB PDZ domains were systematically evaluated using isothermal titration calorimetry (ITC) for binding to a range of NS1 PBM sequences ( Table 1). As expected the EPEV peptide did not bind any of the SCRIB PDZ domains since it lacks a canonical PBM, similar to the RSKV peptide that also showed no binding, in agreement with previously published results [22]. Our ITC analysis showed the ESEV PBM from the highly pathogenic strain peptide is able to interact with all four individual SCRIB PDZ domains with affinities ranging from 11.8 ± 5.5 to 21.4 ± 1.7 µM ( Figure 2, Table 2). Interestingly, these interactions contrast with previously observed interactions from GST pull-down studies that suggested the ESEV PBM sequence in the context of recombinant full-length NS1 was only able to bind to a tandem of PDZ1 and PDZ2 [22]. Examination of SCRIB PDZ domain interactions with the PBM peptide from the ESKV PBM revealed with SCRIB PDZ1, 2 and 3 domains, but not with PDZ4 (Table 2). KSEV, RSKV and RSEV showed no binding to the individual PDZ domains ( Table 2). The negative control EPEV also showed no binding as expected ( Table 2) [17].

Structural Analysis of SCRIB PDZ-NS1 Interactions
To gain structural insight into the interactions between the SCRIB PDZ domains and NS1 PBM motifs, we systematically attempted to crystallise complexes of individual SCRIB PDZ domains bound to identified interacting NS1 PBM peptides from our ITC analysis.

Structural Analysis of SCRIB PDZ-NS1 Interactions
To gain structural insight into the interactions between the SCRIB PDZ domains and NS1 PBM motifs, we systematically attempted to crystallise complexes of individual SCRIB PDZ domains bound to identified interacting NS1 PBM peptides from our ITC analysis.
To validate our crystal structures, we performed mutagenesis to probe SCRIB PDZ domain interactions with NS1 PBM in ITC. The PDZ1 mutant H793A abolished binding with both the ESEV and ESKV PBM (Table 1, Figure 5). SCRIB PDZ2 H928A bound to ESEV with a 5-fold tighter affinity (3.6 ± 0.7 µM) compared to SCRIB PDZ2 WT (18.7 ± 2.4 µM), and, interestingly, did not bind to ESKV (Table 1, Figure 5). SCRIB PDZ3 H1071A did not bind to either ESEV or ESKV, whereas PDZ3 K1040A bound to both (ESEV: 22.2 ± 1.1 µM) (ESKV: 13.8 ± 0.8 µM) (Table 1, Figure 5). None of our SCRIB PDZ4 mutants (PDZ4 R1110D, PDZ4 R1116D and PDZ4 H1170A) interacted with NS1 ESEV. We confirmed correct folding of all mutants using CD spectroscopy ( Figure 6).    Circular dichroism spectra recorded for wild-type and mutant SCRIB PDZ1, 2, 3 and 4 domains indicated that there were no major spectral differences between the proteins, with both wild-type and mutant spectra showing the characteristic features indicative of a fold containing mixed alpha/beta secondary structure elements, suggesting that they were similarly folded, with mutations not leading to unfolding of the PDZ domains.

Discussion
To understand the ability of the influenza A virus protein NS1 to influence SCRIB signalling as an exogenous interaction partner of SCRIB, we systematically examined the interactions of individual SCRIB PDZ domains with the PBM peptides of several influenza A virus strains in vitro using ITC. The ESEV sequence showed an affinity towards all four SCRIB PDZ domains, with SCRIB PDZ4 showing the highest affinity interaction (K D = 11.8 ± 5.5 µM), followed by SCRIB PDZ3 (K D = 13.1 ± 1.6 µM), SCRIB PDZ2 (K D = 18.7 ± 2.4 µM) and lastly SCRIB PDZ1 (K D = 21.4 ± 1.7 µM). Our measurements were in contrast to previously reported GST pull-down assays [22], which suggested that NS1 ESEV was unable to bind to any of the single SCRIB PDZ domains and only interacted with SCRIB PDZ domains arranged as PDZ1 and PDZ2 tandem. However, we note that our ITC experiments were performed using 10-mer PBM peptides, whereas the GST pull-down study used full-length NS1 proteins, which may affect the accessibility of the PBM to the PDZ constructs [22].
The NS1 PBM sequence ESKV from the Hubei strain was originally identified via a phylogenetic analysis that identified a mutation in the NS1 protein from the H5N1 virus; the avian virus-type PBM sequence ESEV (which affects virulence) was replaced with ESKV [32]. Whilst NS1, featuring the ESKV sequence, supported respiratory infection in mice, the substitution of Lys to Glu resulted in a slight decrease in virulence [32]. Our ITC analysis of the ESKV PBM revealed an interaction with SCRIB PDZ2 (K D = 9.5 ± 0.7 µM) as the tightest interactor, followed by SCRIB PDZ3 (K D = 12.1 ± 0.9 µM) and SCRIB PDZ1 (K D = 30.0 ± 0.9 µM) as the weakest interactor. Furthermore, our data also differ from other biochemical studies reported, where co-immunoprecipitation experiments showed only negligible binding to SCRIB [32]. Interestingly, our ITC measurements showed that a single amino acid difference in the PBM sequence (Glu acid to Lys) in position −1 abolishes the interaction with SCRIB PDZ4. To date there is no available structure of the individual SCRIB PDZ4 domain bound to an interactor; however, several studies have previously reported interactions of SCRIB PDZ4 with other ligands [14,40]. Notably, glutamic acid is negatively charged and the change in charge at position −1 raises the possibility that it is interacting with a positively charged residue within the SCRIB PDZ4 domain. With this possibility in mind, Arg1116 PDZ4 was a target for mutagenesis, since β2-strand is a typical location for the −1 position of a PBM to interact. By mutating its basic residue Arg1116 into an acidic residue like Asp, this mutation was shown to ablate the interaction of the ESEV peptide ( Figure 5). This suggests the possibility that specific ionic interactions play an important role in NS1 binding to SCRIB PDZ4.
The majority of the SCRIB PDZ mutants displayed abolished binding (PDZ1 H793A, PDZ3 H1071A, PDZ4 R1110D, PDZ4 R1116D, PDZ4 H1170A) with ESEV and ESKV (Table 1, Figure 5). It appears that the conserved His in each domain found on the α2 helix, PDZ1 H793, PDZ3 H1071 and PDZ4 H1170A is an important position even when this was not obvious from the SCRIB PDZ1:ESEV structure, whereas in the SCRIB PDZ1:ESKV and SCRIB PDZ3:ESKV complexes a clear role for this His could be identified (Figure 3). In the case of the SCRIB PDZ1:ESEV complex, this is possibly due to the crystallographic packing (resulting in a different rotamer for the Ser 223 position in ESEV compared to ESKV), which prevents the Ser hydroxyl group from reaching SCRIB PDZ1 H793 for a hydrogen bond in the SCRIB PDZ1:ESEV complex ( Figure 3). However, this interaction may be available in solution, since substitution of the His by an Ala significantly impacts the affinity of the interaction. Notably, SCRIB PDZ2 H928A bound to ESEV with a 5-fold tighter affinity (3.6 ± 0.7 µM) compared to PDZ2 WT (18.7 ± 2.4 µM), and did not bind to ESKV (Table 1, Figure 5). In the absence of a structure for a SCRIB PDZ2:NS1 PBM complex, we have no structural insight into how this amino acid change in SCRIB PDZ2 leads to such a difference in interactions between the ESEV and ESKV peptides, and why SCRIB PDZ2 H928A has an increased affinity for ESEV.
Considering that H5N1 originated in birds, we examined the protein sequence conservation of interacting human and avian Scribble PDZ domains (Figure 7). Examination of the main NS1 interaction, mediating secondary structure elements from Scrib PDZ domains, reveals that all residues involved in hydrogen bonds are conserved, as well as all main residues involved in van der Waals contacts. This suggests that the avian influenza A virus NS1 may be as effective at manipulating Scribble-mediated polarity signalling in avian species as in humans.
have no structural insight into how this amino acid change in SCRIB PDZ2 leads to such a difference in interactions between the ESEV and ESKV peptides, and why SCRIB PDZ2 H928A has an increased affinity for ESEV.
Considering that H5N1 originated in birds, we examined the protein sequence conservation of interacting human and avian Scribble PDZ domains (Figure 7). Examination of the main NS1 interaction, mediating secondary structure elements from Scrib PDZ domains, reveals that all residues involved in hydrogen bonds are conserved, as well as all main residues involved in van der Waals contacts. This suggests that the avian influenza A virus NS1 may be as effective at manipulating Scribble-mediated polarity signalling in avian species as in humans. Our affinity measurements for the NS1 PBM interactions with Scribble can be compared to previously measured interactions with other endogenous interactors, such as β-PIX [28,41] β-PIX is an important guanine nucleotide exchange factor for small GTPases, where its membrane-associated localisation is dependent on Scribble interactions, enabling the regulation of cellular processes, such as vesicle trafficking, cytoskeletal organisation, and cell migration [41][42][43][44]. Previous studies have shown that all four of the Scribble PDZ domains can bind to β-PIX through ELISA assays; however, using ITC, we showed that the β-PIX PBM peptide (sequence: PAWDETNL) only bound to SCRIB PDZ1, 2 and 3 [28,41]. The ability of NS1 ESEV to bind to SCRIB PDZ4 may be a significant aspect that allows the influenza A virus NS1 to outcompete endogenous proteins that are unable to engage SCRIB PDZ4. SCRIB PDZ1 displays a higher affinity to β-PIX (3.3 ± 0.3 μM), compared to both NS1 ESEV (21.4 ± 1.7 μM) and ESKV (30.0 ± 0.9 μM) PBMs. In contrast, SCRIB PDZ2 bound tighter to NS1 ESEV (18.7 ± 2.4 μM) and ESKV (9.5 ± 0.7 μM) compared to β-PIX (67.8 ± 7.9 μM). Interestingly, there were no significant differences in the interaction of SCRIB PDZ3 (β-PIX:14.7 ± 2.1 μM, ESEV: 13.1 ± 1.6 μM, ESKV:12.1 ± 0.9 μM). Our affinity measurements indicate that β-PIX has a preference for SCRIB PDZ1, whereas NS1 ESEV and ESKV lean towards SCRIB PDZ4 and PDZ2, respectively, thus providing a mechanism where NS1 can selectively outcompete the endogenous β-PIX for specific Scribble PDZ domains. Our affinity measurements for the NS1 PBM interactions with Scribble can be compared to previously measured interactions with other endogenous interactors, such as β-PIX [28,41] β-PIX is an important guanine nucleotide exchange factor for small GTPases, where its membrane-associated localisation is dependent on Scribble interactions, enabling the regulation of cellular processes, such as vesicle trafficking, cytoskeletal organisation, and cell migration [41][42][43][44]. Previous studies have shown that all four of the Scribble PDZ domains can bind to β-PIX through ELISA assays; however, using ITC, we showed that the β-PIX PBM peptide (sequence: PAWDETNL) only bound to SCRIB PDZ1, 2 and 3 [28,41]. The ability of NS1 ESEV to bind to SCRIB PDZ4 may be a significant aspect that allows the influenza A virus NS1 to outcompete endogenous proteins that are unable to engage SCRIB PDZ4. SCRIB PDZ1 displays a higher affinity to β-PIX (3.3 ± 0.3 µM), compared to both NS1 ESEV (21.4 ± 1.7 µM) and ESKV (30.0 ± 0.9 µM) PBMs. In contrast, SCRIB PDZ2 bound tighter to NS1 ESEV (18.7 ± 2.4 µM) and ESKV (9.5 ± 0.7 µM) compared to β-PIX (67.8 ± 7.9 µM). Interestingly, there were no significant differences in the interaction of SCRIB PDZ3 (β-PIX:14.7 ± 2.1 µM, ESEV: 13.1 ± 1.6 µM, ESKV:12.1 ± 0.9 µM). Our affinity measurements indicate that β-PIX has a preference for SCRIB PDZ1, whereas NS1 ESEV and ESKV lean towards SCRIB PDZ4 and PDZ2, respectively, thus providing a mechanism where NS1 can selectively outcompete the endogenous β-PIX for specific Scribble PDZ domains.
When complexes of Scribble PDZ domains are superimposed, no overt differences in the mode of binding between the endogenous cellular and the viral interactor are obvious. SCRIB PDZ1:ESEV and SCRIB PDZ1:ESKV superimposed with SCRIB PDZ1:β-PIX with RMSD values of 0.77 and 0.75 Å, respectively. SCRIB PDZ1:ESKV Chain A superimposed with SCRIB PDZ3:β-PIX showed an RMSD of 0.57 Å. Consequently, all differences between these complexes are due to specific interactions during engagement with the PDZ domains. The −1 position of β-PIX N645 forms a hydrogen bond with S741 PDZ1 , whereas the NS1 PBM does not interact with SCRIB PDZ1 from the −1 position; instead, the −2 position S223 appears to be the important position, forming a hydrogen bond with I742 PDZ1 . Additionally, S1026 PDZ3 interacts with the β-PIX W641, whereas S1026 PDZ3 is not involved in the SCRIB PDZ3:ESKV complex; instead, a K1040 PDZ3 -E222 ESKV hydrogen bond is formed.
In summary, we show that SCRIB PDZ domains display differential binding affinities for the influenza A virus NS1 and that subtle differences in the detailed interactions drive the differences in affinity observed for PDZ1-4. Our findings provide a structural basis for avian influenza A virus NS1 subversion of SCRIB-mediated polarity signalling, and will form the platform for future structure-guided investigations to understand how the differential ability of individual SCRIB PDZ domains to engage NS1 impacts the control of cell polarity during viral replication.  Data Availability Statement: Data supporting the findings of this manuscript are available from the corresponding authors upon reasonable request. Coordinate files were deposited at the Protein Data Bank (https://www.rcsb.org/) (accessed on 1 July 2021) using accession codes 7QTO, 7QTP and 7QTU for SCRIB PDZ1:ESEV, SCRIB PDZ1:ESKV and SCRIB PDZ3:ESKV, respectively. The raw X-ray diffraction data were deposited at the SBGrid Data Bank [45] (https://data.sbgrid.org/data/ accessed on 20 January 2022) using their PDB accession codes 7QTO, 7QTP and 7QTU for SCRIB PDZ1:ESEV, SCRIB PDZ1:ESKV and SCRIB PDZ3:ESKV, respectively.