key: cord-0834549-qdk7en3m
authors: Zhu, Yanlei; Alvarez, Flavio; Wolff, Nicolas; Mechaly, Ariel; Brûlé, Sébastien; Neitthoffer, Benoit; Etienne-Manneville, Sandrine; Haouz, Ahmed; Boëda, Batiste; Caillet-Saguy, Célia
title: Interactions of SARS-CoV-2 protein E with cell junctions and polarity PDZ-containing proteins
date: 2021-12-06
journal: bioRxiv
DOI: 10.1101/2021.12.04.471219
sha: 0feb3ebda8f3e0b495416f828246f33707b0e113
doc_id: 834549
cord_uid: qdk7en3m

The C-terminus of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) protein E contains a PBM (PDZ binding motif) targeting PDZ (PSD-95/Dlg/ZO-1) domains identical to the PBM of SARS-CoV. The latter is involved in the pathogenicity of the virus. Recently, we identified ten human PDZ-containing proteins showing significant interactions with SARS-CoV-2 protein E PBM. We selected several of them involved in cellular junctions and cell polarity (TJP1, PARD3, MLLT4, LNX2) and MPP5/Pals1 previously shown to interact with SARS-CoV E PBM. Targeting cellular junctions and polarity components is a common strategy by viruses to hijack cell machinery to their advantage. In this study, we showed that these host PDZ domains TJP1, PARD3, MLLT4, LNX2 and MPP5/PALS1 interact in a PBM-dependent manner in vitro and colocalize with the full-length E protein in cellulo, sequestrating the PDZ domains to the Golgi compartment. We solved three crystal structures of complexes between human LNX2, MLLT4 and MPP5 PDZs and SARS-CoV-2 E PBM highlighting its binding preferences for several cellular targets. Finally, we showed different affinities for the PDZ domains with the original SARS-CoV-2 C-terminal sequence containing the PBM and the one of the beta variant that contains a mutation close to the PBM. The acquired mutations in E protein localized near the PBM might have important effects both on the structure and the ion-channel activity of the E protein and on the host machinery targeted by the variants during the infection.

The C-terminus of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) protein E 20 contains a PBM (PDZ binding motif) targeting PDZ (PSD-95/Dlg/ZO-1) domains identical to the PBM 21 of SARS-CoV. The latter is involved in the pathogenicity of the virus. Recently, we identified ten 22

human PDZ-containing proteins showing significant interactions with SARS-CoV-2 protein E PBM. 23 We selected several of them involved in cellular junctions and cell polarity (TJP1, PARD3, MLLT4, 24 LNX2) and MPP5/Pals1 previously shown to interact with SARS-CoV E PBM. Targeting cellular  25 junctions and polarity components is a common strategy by viruses to hijack cell machinery to their 26 advantage. In this study, we showed that these host PDZ domains TJP1, PARD3, MLLT4, LNX2 and 27 MPP5/PALS1 interact in a PBM-dependent manner in vitro and colocalize with the full-length E 28 protein in cellulo, sequestrating the PDZ domains to the Golgi compartment. We solved three crystal 29 structures of complexes between human LNX2, MLLT4 and MPP5 PDZs and SARS-CoV-2 E PBM 30 highlighting its binding preferences for several cellular targets. Finally, we showed different affinities 31

for the PDZ domains with the original SARS-CoV-2 C-terminal sequence containing the PBM and the 32 one of the beta variant that contains a mutation close to the PBM. The acquired mutations in E protein 33 localized near the PBM might have important effects both on the structure and the ion-channel activity 34

of the E protein and on the host machinery targeted by the variants during the infection. proteome (Luck et al., 2012) . PBMs are mainly located at the C-terminus of target proteins and interact 50 directly with PDZ domains. They are classified into three types: type I PBM (-X-S/T-X-ϕCOOH), type 51 II PBM (-X-ϕ-X-ϕCOOH) and type III PBM (-X-D/E-X-ϕCOOH), with ϕ signifying a hydrophobic residue. 52

The C-terminal PBM sequence of SARS-CoV E protein is of type II (-DLLVCOOH) and has been 53 identified as a virulence factor (Jimenez-Guardeño et al., 2014). It is likely that the abilities to target 54 PDZ proteins also make significant contributions to the pathogenesis of the E protein of SARS-CoV-55 2. Indeed, the E protein is highly conserved with 94.7% identity between SARS-CoV and SARS-CoV-56 2 and their PBMs are strictly conserved. 57

Protein E is a small transmembrane protein of 75 residues involved in several phases of the virus life 58 cycle, such as assembly, budding, envelope formation, and pathogenesis (Schoeman and WT), GFP-tagged full-length E protein with the PBM mutated with glycines (GFP-E-GGGG) and 120 GFP-tagged cytoplasmic tail of protein E (last 12 residues) (GFP-E last 12 aa) designated 1, 2, 3 and 121 4 in figure 2 ( Fig. 2A) . The GFP-tagged constructs expression was assessed by examining the 122 fluorescence emitted by the GFP by microscopy and by Western-Blot on the cell lysates using anti-123 GFP antibody (Fig. 2B) . GST alone was used as a negative control. We confirmed using Ponceau S 124 staining that equal amount of GST-tagged protein constructs was bound to the GST resin (Fig. 2C ). 125

GST alone and its fusion with PDZ ZO1, PDZ LNX2, PDZ MLLT4, PDZ MPP5 and PDZ PARD3, 126 are used as baits and were immobilized on glutathione beads and tested for their ability to pull down 127 GFP alone, GFP-E-WT, GFP-E-GGGG and GFP-E last 12 aa by Western-blot using the anti-GFP 128 antibody. After washing, identical amounts of beads were analysed for the presence of GFP-tagged 129

proteins. 130 131

The GFP-E last 12 aa was detected in all interactions with GST-PDZ domains but not detected with 132

GST alone confirming the interactions identified in our high-throughput holdup assay (Caillet-Saguy 133 et al., 2021) such validating the interactions within the context of lysates of HEK293 cells 134

overexpressing SARS-CoV-2 protein E constructs (Fig. 2C ). 135 136

GFP-E-WT was also detected in all interactions with GST-PDZ domains but with GST-PDZ PARD3 137 that showed no clear band (Fig. 2C ). In all cases, the bands have a weaker intensity than GFP-E last 12 138 aa in agreement with a significant lower expression in cells, as shown in the inputs for GFP-E-WT 139 compared to GFP-E last 12 aa (Fig. 2B ). We failed to obtain a clear detection of the band for GST-140 PDZ PARD3 ( Fig 2C) . Conversely, GFP-E-GGGG was not detected in all interactions with GST-PDZ 141 domains except with GST-PDZ PARD3 that showed a weak band (Fig. 2C 

terminal of protein E 148

We deciphered the molecular basis of recognition of the C-terminal sequence of the protein E of SARS-149

CoV-2 encompassing the PBM by the PDZ domains of the proteins engaged in cellular junction and 150 polarity. To this aim, we solved the crystal structures of the complex formed by the C-terminal peptide 151

of protein E encompassing the PBM ( Fig 1C) The crystal structure of MLLT4-PDZ in complex with SARS-CoV-2-E-PBM peptide was solved at a 157 resolution of 2.28 Å. The final refined model contains two PDZ domains per asymmetric unit that 158 adopt a swapped dimer conformation with the PDZ folding comprising five β strands and two α helices. 159

Two peptides are bound to each PDZ dimer (Fig. 3A ). An electron density is observed for the last five 160

and last three residues of the peptide containing the PBM indicating a well-defined conformation of 161 these last C-terminal residues. The last three residues of the SARS-CoV-2-E-PBM peptide bind to the 162 α2/β2-groove in each PDZ unit (Fig. 3D ). 163 164

The swapped dimer comprises an intermolecular interaction between the two PDZ domains through 165 the β1/β6 pair (Fig. 3A) , and several pairs of inter-domain H-bonds between the fragments Lys 1014-166

Gly 1017 of each chain (Fig. 3A ). 167 168

The E PBM binds to the PDZ in a conventional manner as an antiparallel extension of the β2 strand by 169

inserting into a binding groove formed by the β2 strand, the α2 helix and the "GLGF" motif. The last 170

four residues of the SARS-CoV-2-E-PBM peptide were in contact with the PDZ domain, whereas only 171 the last three residues were in contact and the upstream residues were distant from the PDZ domain 172 surface in the previously reported complex between MLLT4-PDZ with the nectin-3 or the Bcr PBM 173

peptides (Chen et al., 2007; Fujiwara et al., 2015) . 174

The residues of the two peptides (Fig. 3F) . In both cases, the last three residues of the SARS-CoV-2-E-211

PBM peptide bind to the canonical α2 helix /β2 strand groove in each PDZ unit (Fig. 3C ). 212

As Here we used the recently developed ALFA tag, that is small (14 residues) and electroneutral (Götzke 238 et al., 2019) to tag the SARS-CoV-2 E protein at the N-terminal. HeLa cells transiently transfected 239

with SARS-CoV-2 ALFA-E encoding plasmids displayed strong Golgi expression of the viral protein 240 as revealed by the co-staining with Golgi marker GM130 (Supplementary Figure S1A) . 241 We then investigated the ability of the viral E protein to recruit GFP-tagged PDZ domains to the Golgi 242 compartment. When transfected alone, GFP-tagged PDZ domains from ZO1, MLLT4, MPP5/PALS1, 243 LNX2 and PARD3 do not accumulate in the Golgi apparatus (supplementary Figure S1B ). 244

Interestingly, all these GFP-tagged PDZ domains relocalize to the Golgi compartment when they are 245 co-transfected with the ALFA-E construct as shown by the GM130 co-staining for ZO1 (Fig. 4A ) and 246 the other PDZ domains (Supplementary Figure S1C ). ZO-1 PDZ2/E protein colocalization is strictly 247 PBM-dependent as no colocalization is observed between ZO1-PDZ2 and the ALFA E PBM mutant 248

for which the PBM sequence DVLL was substituted by four glycines (Fig. 4B) . Likewise, this 249 interaction was specific as SCRIB-PDZ1 domain, an unrelated PDZ domain that was not identified in 250 our screen, showed no tropism for the Golgi compartment when it was cotransfected with the ALFA E 251 construct (Fig. 4C) . These results indicate that the viral E protein can bind to ZO1, MLLT4, MPP5, 252

LNX2 and PARD3 PDZ domains in cellulo in a PBM specific manner and recruit them to the Golgi 253 apparatus. 254

Acquired mutations in SARS-CoV-2 protein E localized close to the PBM were reported and the most 257 common non-synonymous mutations were S68F and P71L (Hassan et al., 2020) . We focused on the 258 P71L mutation found in the beta variant. The VOC beta was first identified in South Africa in 259

September 2020 and has been reported in more than 130 countries. The mutations on the beta variant 260 make it more transmissible (Tegally et al., 2021) Figure S3 ). The 341

MLLT4-PDZ is found mainly monomeric in solution with a sedimentation coefficient of 1.4S, 342

suggesting that the swapped dimer is probably an artefact of crystallization. Interestingly, MLLT4-343 PDZ is able to recognize type II PBM but also type I PBM due to an unexpected glutamine in the α2 344 helix as previously reported (Zhou et al., 2005) . 345

Remarkably, ZO-1-PDZ2 forms a very stable swapped dimer with an extended antiparallel inter-346 domain β sheet. The β1 and β2 strands of one domain is swapped with those from the second domain 347

allowing the formation of the binding groove (Fanning et al., 2007; Chen et al., 2008) . We also 348

performed AUC experiments on ZO-1-PDZ2 to verify its oligomeric state in solution (Table 4; beta variant E C-terminal peptide (-NLNSSRVLDLLVCOOH) containing the P71L mutation close to 358 the PBM (Fig. 1B) . This position -4 is not well defined within an electronic density in the crystal 359 structures and when a density is observed, it is the last upstream residue modeled for the C-terminal of 360 the protein E. In the three complexes we studied, the residue in position -4 is not involved in the 361 interaction network with the PDZ domains. Surprisingly, we determined that P71L mutation markedly 362 altered affinities with host PDZ domains. While it noticeably reduces the affinities for the PDZ domains 363 of ZO-1, MLLT4 and PARD3, it enhances its affinity for LNX2 compared to SARS-CoV-2 WT. 364

Influences of PBM upstream residues on the affinity for PDZ domains has already been reported 365

previously ( linker-E sequence was subsequently cloned into a pCMV backbone vector using ClaI and XhoI. The 396 E PBM C-terminal mutation (DVLL>GGGG) was performed using the Q5 site-directed mutagenesis 397 kit (NEB). The sequences corresponding to the PDZ domains were cloned into the pCMV GFP vector 398

using EcoRI and XhoI sites. 399

The vectors were used to transform E. coli BL21 Star (DE3) (Invitrogen) strain. Bacteria were grown 401 in LB medium supplemented with ampicillin (100 mg / L) at 37°C. Protein expression was induced at 402

OD600nm 0.8-1 with 0.2 mM IPTG at 18°C overnight. Bacteria were then harvested and PDZ domains 403 of MLLT4, LNX2, PARD3 and MPP5 were resuspended in lysis buffer (Tris 50 mM (pH 7.5), NaCl 404 250 mM, β-mercaptoethanol 2 mM, protease inhibitors, one tablet per 50 mL of buffer (EDTA-free, 405

Roche Diagnostics) and benzonase (E1014-25KU >250 Units/liter of culture)). Then, the cells were 406 broken under pressure by CELLD (1.3 kBar at 4°C ). The debris were pelleted by centrifugation 407 (30000g, 1 h, 4 °C) . The GST tag fused proteins were purified by affinity chromatography column with 408

GSTrap HP (GE Healthcare) and the His or His-MBP tag fused proteins by nickel chelated HiTrap 409

Chelating HP (GE Healthcare), followed by a TEV protease cleavage at 16°C overnight. A final step 410 of size-exclusion chromatography was achieved using a Sephacryl S-100 HR 16/600 (GE Healthcare) 411

with buffer Tris 50 mM (pH 7.5), NaCl 150 mM, TCEP 0.5 mM, protease inhibitors, one tablet per 412 100 mL of buffer (Complete, EDTA 2%, Roche Diagnostics). Regarding the purification of ZO-1-413 PDZ2, a denaturation protocol was applied. The bacterial pellet was resuspended in a denaturing 414 solution containing guanidine hydrochloride (6M) and imidazole 20 mM, and cells were lysed by 415 sonication. The debris were pelleted by centrifugation (30000g, 1 h, 4°C). The His-tagged ZO-1-PDZ2 416

were purified using a HiTrap Chelating HP (GE Healthcare), where a renaturation gradient by 417 exchange of denaturation buffer to a gel filtration buffer (Tris 50 mM (pH 7.5), NaCl 150 mM, TCEP 418 0.5 mM, protease inhibitors, one tablet per 100 mL of buffer (Complete, EDTA 2%, Roche 419

Diagnostics)) was performed. Protein elution was then carried out by a gradient of imidazole from 0 to 420 500 mM in gel filtration buffer. Finally, a size exclusion chromatography step was achieved using a 421

Sephacryl® S-200 HR (GE Healthcare). 422

The acetylated peptides containing the C-terminal PBM sequence of E protein (12 residues long) of 424 SARS-CoV-2 (acNLNSSRVPDLLVCOOH) or of it South African mutant P71L 425

(acNLNSSRVLDLLVCOOH) were synthesized in solid phase using the Fmoc strategy (Proteogenix,  426 Schiltigheim, France). Peptides were resuspended in water to prepare stock solutions. 427

The PDZ -PBM complexes for crystallization were generated by mixing LNX2, MLLT4 or MPP5 429 PDZ domains with SARS-CoV-2 E protein PBM peptide at a ratio of 1:2, ensuring that at least 92% 430 of complexes were formed. Initial screening of crystallization conditions was carried out by the vapor 431 diffusion method using a MosquitoTM nanoliter Models were rebuilt using COOT (Emsley et al., 2010) , and refinement was done with phenix.refine 445 of the PHENIX suite (Adams et al., 2010 We thank the staff of the Crystallography core facility at the Institut Pasteur for carrying out robot-550 driven crystallization screenings, and the staff at the beamlines Proxima 1 and Proxima 2 at the French 551 national synchrotron facility (SOLEIL, St Aubin, France), for help with data collection. We thank the 552 DIM 1HEALTH region Ile-de-France for funding the Centrifection project that has allowed the Optima 553 ultracentrifuge investment. 554

Supplementary Material 555 The Supplementary Material for this article can be found online at: 556 Statistics for the highest-resolution shell are shown in parentheses. 

PHENIX: a comprehensive Python-based system for macromolecular structure solution

Host PDZ-containing proteins targeted by SARS-CoV-2

Characterization of the 563 SARS-COV -2 E Protein: Sequence, Structure, Viroporin, and Inhibitors

Role of Severe Acute Respiratory Syndrome Coronavirus 567

Viroporins E, 3a, and 8a in Replication and Pathogenesis. mBio 9

Structural basis for 569 SARS-CoV-2 envelope protein recognition of human cell junction protein PALS1

Domain-swapped dimerization of ZO-1 572 PDZ2 generates specific and regulatory connexin43-binding sites

Solution structure and backbone dynamics of 575 the AF-6 PDZ domain/Bcr peptide complex

Rare Deleterious PARD3 Variants 577 in the aPKC-Binding Region are Implicated in the Pathogenesis of Human Cranial Neural Tube 578 Defects Via Disrupting Apical Tight Junction Formation

Identification of a Golgi complex-targeting 581 signal in the cytoplasmic tail of the severe acute respiratory syndrome coronavirus envelope protein

A 584 severe acute respiratory syndrome coronavirus that lacks the E gene is attenuated in vitro and in vivo

SARS-CoV-2 envelope protein topology in eukaryotic membranes

Throughput Production of a New Library of Human Single and Tandem PDZ Domains Allows 591 Quantitative PDZ-Peptide Interaction Screening Through High-Throughput Holdup Assay

Features and development of Coot

Domain swapping within PDZ2 is 596 responsible for dimerization of ZO proteins

Crystal 599 structure of afadin PDZ domain-nectin-3 complex shows the structural plasticity of the ligand-600 binding site

Characteristics of 602 SARS-CoV-2 variants of concern B.1.1.7, B.1.351 or P.1: data from seven EU/EEA countries, weeks 603 38/2020 to 10/2021

The ALFA-tag is a highly versatile tool for nanobody-based bioscience applications

Viral targeting of PDZ polarity proteins 608 in the immune system as a potential evasion mechanism

Molecular insights into 611 receptor binding of recent emerging SARS-CoV-2 variants

Mechanism and role of PDZ domains in signaling complex 614 assembly

SARS-CoV2 envelope protein: non-616 synonymous mutations and its consequences

Afadin: A 619 key molecule essential for structural organization of cell-cell junctions of polarized epithelia during 620 embryogenesis

Structures of the human Pals1 PDZ domain with and without ligand suggest gated access of 623 Crb to the PDZ peptide-binding groove

Viral Interactions with PDZ Domain-Containing Proteins-An 626 Oncogenic Trait? Pathogens 5

Emerging theme: cellular PDZ proteins as common targets of 628 pathogenic viruses

Structural basis of coronavirus E protein 630 interactions with human PALS1 PDZ domain

The PDZ-binding motif of severe acute respiratory 634 syndrome coronavirus envelope protein is a determinant of viral pathogenesis

Highly 637 accurate protein structure prediction with AlphaFold

PDZ domains and their binding partners: structure, specificity, 642 and modification

Structure of a conserved Golgi complex-644 targeting signal in coronavirus envelope proteins

SUBCELLULAR 647 LOCALIZATION OF SARS-CoV STRUCTURAL PROTEINS

The emerging contribution of sequence context to 649 the specificity of protein interactions mediated by PDZ domains

Structure and drug binding of the SARS-CoV-2 envelope protein transmembrane domain in 653 lipid bilayers

Solving structures of protein complexes by molecular replacement with Phaser

Subcellular location and topology of severe acute respiratory syndrome 658 coronavirus envelope protein

Interactions of SARS-CoV-2 envelope protein with amilorides correlate with antiviral activity

Exploitation of the Secretory 663 Pathway by SARS-CoV-2 Envelope

SARS-CoV-665 2 infection induces the dedifferentiation of multiciliated cells and impairs mucociliary clearance

The Maguk 668 protein, Pals1, functions as an adapter, linking mammalian homologues of Crumbs and Discs Lost

Coronavirus envelope protein: current knowledge

Recognition of unique carboxyl-terminal motifs by distinct PDZ domains

Detection of a SARS-CoV-2 variant of concern in South Africa

The 679 SARS coronavirus E protein interacts with PALS1 and alters tight junction formation and epithelial 680 morphogenesis

Interference 682 with the PTEN-MAST2 interaction by a viral protein leads to cellular relocalization of PTEN

Tight Junctions Go Viral! Viruses

Comparing 687 the binding properties of peptides mimicking the Envelope protein of SARS-CoV and SARS-CoV-2 688 to the PDZ domain of the tight junction-associated PALS1 protein

Increased risk of hospitalisation and intensive care admission associated with 692 reported cases of SARS-CoV-2 variants B.1.1.7 and B.1.351 in Norway

Quantifying domain-ligand affinities and specificities by high-throughput holdup assay

Antibody resistance of 698 SARS-CoV-2 variants B.1.351 and B.1.1.7

Numb regulates cell-cell adhesion and 700 polarity in response to tyrosine kinase signalling

Throughput Crystallization Pipeline at the Crystallography Core Facility of the Institut Pasteur

Solution Structure of AF-6 PDZ 706 Domain and Its Interaction with the C-terminal Peptides from Neurexin and Bcr