key: cord-0264419-e0cvtgnd authors: Maseko, Sibusiso B.; Van Molle, Inge; Blibek, Karim; Gorgulla, Christoph; Olivet, Julien; Blavier, Jeremy; Vandermeulen, Charlotte; Skupiewski, Stéphanie; Saha, Deeya; Ntombela, Thandokuhle; Lim, Julianne; Lembo, Frederique; Beauvois, Aurelie; Hamaidia, Malik; Borg, Jean-Paul; Zimmermann, Pascale; Delvigne, Frank; Willems, Luc; Van Weyenbergh, Johan; Kim, Dae-Kyum; Dequiedt, Franck; Arthanari, Haribabu; Volkov, Alexander N.; Twizere, Jean-Claude title: Interactome and structural basis for targeting the human T-cell leukemia virus Tax oncoprotein date: 2021-08-25 journal: bioRxiv DOI: 10.1101/2021.08.25.457680 sha: c739620e1f4f857cfd878d061d0392c692071544 doc_id: 264419 cord_uid: e0cvtgnd Human T-cell leukemia virus type-1 (HTLV-1) is the first pathogenic retrovirus discovered in human. Although HTLV-1-induced diseases are well characterized and linked to the encoded Tax-1 protein, there is currently no strategy to target Tax-1 functions with small molecules. Here, we report a comprehensive interaction map between Tax-1 and human PDZ domain-containing proteins (hPDZome), and we show that Tax-1 interacts with one-third of them. This includes proteins involved in cell cycle, cell-cell junction and cytoskeleton organization, as well as in membrane complexes assembly. Using nuclear magnetic resonance (NMR) spectroscopy, we have determined the structural basis of the interaction between the C-terminal PDZ binding motif (PBM) of Tax-1, and the PDZ domains of DLG1 and syntenin-1. Finally, we have used molecular modeling and mammalian cell-based assays to demonstrate that Tax-1/PDZ-domain interactions are amenable to small-molecule inhibition. Thus, our work provides a framework for the design of targeted therapies for HTLV-1-induced diseases. Highlights comprehensive interactome map of HTLV-1 Tax / human PDZ proteins basis of Tax-1 PBM binding to human DLG1 and syntenin-1 PDZ domains”. significance of inhibiting Tax-1 functions of the Tax-1 / PDZ interface Graphical abstract Human T-cell lymphotropic viruses (HTLV-1 to -4) are members of Deltaretrovirus genus of the Retroviridae family. The first human retrovirus to be isolated 40 years ago (Poiesz et al., 1981) , HTLV-1 is the etiological agent of adult T-cell leukemia/lymphoma (ATL), an aggressive neoplasm, as well as HTLV-1-associated myelopathy (HAM). HAM is also termed tropical spastic paraparesis (TSP) (HAM/TSP), an immune degenerative neurologic syndrome. HTLV-2 was discovered shortly after HTLV-1 (Kalyanaraman et al., 1982) , but has not been convincingly linked to lymphoproliferative disorders (Feuer and Green, 2005) . HTLV-3 and 4 were the latest identified (Calattini et al., 2005; Wolfe et al., 2005) , and their possible association with human diseases needs further investigation. Currently about 10-20 million people worldwide are infected with HTLV-1, and 1 to 10% will develop severe ATL or HAM/TSP diseases (Gessain et al., 1985; Hinuma et al., 1981; Pasquier et al., 2018) . For the last 40 years, ATL patients have been treated using chemotherapy-based approaches, but with very limited benefit the median survival rate being 8-10 months (Bazarbachi et al., 2011; Cook and Phillips, 2021; Utsunomiya et al., 2015) . Improved survival is achieved by antiviral therapies combining zidovudine and interferonalpha (Nasr et al., 2017) , allogenic hematopoietic stem cell transplantation (Fuji et al., 2016) , or the use of monoclonal antibodies targeting CC chemokine receptor 4 (CCR4), which is frequently expressed in ATL patient samples (Ishida et al., 2015; Yoshie, 2005; Yoshie et al., 2002) . However, the above therapies are not fully effective, mainly due to clinical disease heterogeneity, discrepancies in treatment between countries, and lack of specific and universally targeted drugs (Cook and Phillips, 2021) . In contrast to ATL, no treatment is available for HAM/TSP patients. Both HTLV-1-induced diseases are associated with high proviral loads. Targeting viral replication, including re-positioning existing anti-human immunodeficiency virus (HIV) drugs, was investigated to decrease HTLV-1 cell-to-cell transmission (Pasquier et al., 2018) and may lead to effective anti-HTLV-1 therapies. In addition to the essential retroviral genes gag, pol and env, HTLV-1 encodes regulatory proteins including Tax-1, Rex, p12I, p13II, p30II and HBZ. The Tax-1 protein is a major determinant in HTLV pathogenesis and persistence. The transforming activity of HTLV-1 has been essentially attributed to the Tax-1 gene, which is able to induce leukemialymphoma in transgenic mice (Hasegawa et al., 2006; Ohsugi et al., 2007) . The Tax-1 protein is indeed a potent transcriptional activator of viral and cellular genes through association with transcription modulators including CREB (Ohsugi et al., 2007) ,SRF (FuJiI et al., 1988; Winter and Marriott, 2007) , AP-1 (Yoshida, 2001) , NF-kB (Ballard et al., 1988; Béraud et al., 1994; Leung and Nabel, 1988; Xiao et al., 2001) , CBP/p300 (Harrod et al., 1998; Kwok et al., 1996) , PCAF (Jiang et al., 1999) , and P-TEFb (Cho et al., 2010) . Thus, Tax-1 deregulates an array of cellular genes directly involved in T-cell proliferation (Albrecht et al., 1992; Crenon et al., 1993; Himes et al., 1993; Maruyama et al., 1987; Twizere et al., 2003) , migration (Twizere et al., 2007) , apoptosis (Taylor and Nicot, 2008) , and cell-cycle (Neuveut and Jeang, 2000; . Tax-1 is a multidomain protein harboring intrinsically disordered linker regions, encompassing residues 76 -121, 252 -275, and the C-terminal amino acids 320 -353 (Boxus et al., 2008) . Such modular organization ensures conformational plasticity and could explain dynamic hijacking of crucial cellular functions through interaction with a variety of targets. More than 200 cellular proteins have been reported to interact with Tax (Boxus et al., 2008; Legros et al., 2009; Simonis et al., 2012a) , including numerous signaling molecules such as IκB kinases (Harhaj and Sun, 1999; Jin et al., 1999) , MAPK/ERK kinase 1 (MEKK1) (Yin et al., 1998) , TGF-beta activating kinase 1 (TAK1) (Wu and Sun, 2007) , c-Jun N-terminal kinase 1 (JNK), phosphatidylinositol 3-kinase and its downstream messenger protein kinase B (Liu et al., 2001; Peloponese Jr and Jeang, 2006) , protein phosphatases (Hong et al., 2007) , GTP-binding proteins (Twizere et al., 2007; Wu et al., 2004) , cytoskeleton components (Nejmeddine et al., 2005; Wu et al., 2004) , cell cycle molecules (Haller et al., 2000; Haller et al., 2002; Kehn et al., 2005; Kehn et al., 2004) , nuclear effectors such as histone and chromatin modifier enzymes (Kamoi et al., 2006; Legros et al., 2011) , and mini-chromosome maintenance proteins (Boxus et al., 2012) . The carboxyl terminus of Tax-1 harbors a PDZ-binding motif (PBM), which confers binding to a class of proteins containing a defined structure of ~90 amino acids known as PDZ (PSD-95/Discs Large/ZO-1) domain (Sheng and Sala, 2001; Tonikian et al., 2008b) . PDZ-containing proteins are modular polypeptides implicated in large protein complexes assembly, mediating signaling, cell polarity, and communication (Hung and Sheng, 2002) . Our analysis of the human genome estimates the presence of 256 PDZ domains in 149 distinct proteins, excluding variants and isoforms (Belotti et al., 2013a) . To date, 14 PDZ proteins interacting with HTLV-1 Tax have been identified (Boxus et al., 2008; Yan et al., 2009 ). Tax-1-PDZ protein interactions have important implications in HTLV-1-induced leukemogenesis process. In particular, it was shown that the Tax-1 PBM motif, which is not present in the HTLV-2 Tax counterpart, promotes transformation of rat fibroblasts and mouse lymphocytes in vitro (Higuchi et al., 2007; Hirata et al., 2004a; Tsubata et al., 2005) , as well as persistence of the HTLV-1 virus in vivo (Xie et al., 2006) . In this study, we hypothesized that targeting Tax-1-PDZ interactions would provide novel anti-HTLV-1 transmission strategies, which could potentially be used to combat ATL and TSP/HAM diseases. We therefore comprehensively mapped the Tax-1 interactome with human PDZ-containing proteins and found that about one third of the human PDZome is capable of specific interactions with Tax-1. Using nuclear magnetic resonance (NMR) spectroscopy, we structurally characterized interactions between Tax-1 PBM and PDZ domains from two highly relevant host proteins, the human homolog of the drosophila discs large tumor suppressor (DLG1/SAP97) and syntenin-1. Finally, we demonstrate that FJ9, a small molecule able to disrupt Tax-1-syntenin-1 interaction, inhibits several Tax-1 functions, including cell immortalization and HTLV-1 cell-to-cell transmission. The Tax-1 viral oncogene is a hallmark of initiation and maintenance of HTLV-1-induced diseases. However, the encoded Tax-1 protein is still viewed as "undruggable", essentially because of its pleiotropic effects on cellular systems and low expression levels in most HTLV-1-infected cells. We reasoned that a detailed understanding of the Tax-1 interactome could reveal druggable Tax-1 interactome modules, amenable to small molecule inhibition. We have previously highlighted the function of Tax as a "hub" molecule, capable of interacting with several host cell complexes modules at the transcription, posttranscription, and translation levels (Simonis et al., 2012b; Vandermeulen et al., 2021) . To date, high throughput and small-scale focused studies identified 258 human proteins interacting with Tax-1 (Vandermeulen et al., 2021) . To comprehensively estimate the size of the Tax-1 interactome, we identified Tax targets based on motif-domain and domain-domain interactions and predicted 2401 human proteins, including 161 experimentally validated partners, as potential targets for Tax ( Figure 1A ). This suggests that about 10% of the full human proteome could be affected by this viral oncoprotein through a welldefined set of protein modules (Table S1 ). In order to understand the functional plasticity of the Tax interactome, we hypothesized that transient associations controlled by multi-domain scaffolding proteins are able to dynamically relocate Tax complexomes within the cell, as recently demonstrated for binary interactome yeast models (Lambourne et al., 2021) . In the context of Tax Our map of the Tax-PDZ interactome ( Figure 1B ) contains 56 human proteins, representing 37% of the human PDZome (Belotti et al., 2013b) , and a substantial increase from the 14 already known human PDZ proteins directly interacting with Tax ( Figure S1B ). To further determine the relative implication of PDZ-containing proteins in the human interactome, we used our recently released human binary reference interactome map (HuRI) (Luck et al., 2020) , and ranked PDZ proteins according to the number of their direct interactors (" first degree") or indirect associations ("second degree") ( Table S2 ). The average first and second degrees of Tax-1 partners were 24 and 536 PPIs, respectively, further emphasizing the scaffolding role of PDZ-containing proteins targeted by the viral Tax oncoprotein. To assess the overall biological significance of Tax-PDZ interactome, we investigated the expression of human PDZ protein coding genes in different relevant datasets: (i) expression data across 24 different T-cell lines ( Figure 1C ); (ii) differential expression data in HTLV-1-infected versus non-infected cells ( Figure 1D ); (iii) differential expression data in Jurkat cells expressing Tax-1 versus control cells ( Figure 1E ); (iv) mutational landscapes in ATL patient samples ( Figure 1F ); and (v) transcriptome data from a "gene signature" of ATL (Bazarbachi, 2016; Fujikawa et al., 2016 ) ( Figure 1G ). Finally, we performed spatial analysis of functional enrichment (SAFE) (Baryshnikova, 2016) using the above datasets and a STRING network of Tax-1/PDZ interactors (Table S2) . We identified five functional modules including nucleic acid binding, nuclear ubiquitin binding, regulation of cell cycle, cell-cell junction and cytoskeleton organization, and endocytic vesicles membrane assembly ( Figure 1H ). Thus, it appears that the described Tax-1/PDZ interactome is implicated in various Tax functions, from early steps of T-cell infection by HTLV-1 to subsequent disease progression. To characterize the Tax-1-PDZ interactions at the molecular level, we first addressed the association between Tax-1 and the human homolog of the drosophila discs large tumor suppressor (hDLG1/SAP97). hDLG1 is the first PDZ-containing protein identified to interact with Tax-1 and other viral oncoproteins such as HPV 18 E6 and adenovirus E4-ORF1 (Lee et al., 1997) . While the Tax-1-hDLG1 binding detailed here was also functionally validated by numerous previous studies (Aoyagi et al., 2010; Hirata et al., 2004b; Marziali et al., 2017; Suzuki et al., 1999) , its structural basis remained unknown. To obtain molecular-level details of Tax progressively disappear and then reappear elsewhere in the spectrum at the end of the titration, a behavior symptomatic of the slow exchange NMR regime ( Figure 2A ). For both PDZ domains, the strongest effects are observed for the residues in and around the canonical peptide binding site, formed by the β1/β2 loop, the β2 strand and the α2 helix ( Figure 2B ). In particular, the largest binding shifts are detected for the PBM residues G235 and F236 of PDZ1 and G330 and F331 of PDZ2 (Figures 2A-B (Figures 2A-D) . The interaction of Tax-1 10mer with PDZ2 is stronger than that with PDZ1 (KD of 0.9 μM and 2.7 μM, respectively; Figures 2H, L), which is in agreement with our GST-pulldown assay using a Tax-1 protein from HTLV-1 producing MT2 cells ( Figure S2B ). To narrow down further the Tax-1 binding requirements, we performed NMR experiments with the peptides corresponding to the last four and first six amino acids of the Tax-1 10mer. The C-terminal PBM fragment (ETEV) interacts with the 15 N hDLG1 PDZ1+2 tandem in the same way as the Tax-1 10mer ( Figures S3A-D) , while no binding of the Nterminal 6mer peptide (SEKHFR) could be observed ( Figures S4A-B) . These findings demonstrate that the four C-terminal residues of Tax-1, bearing the X-S/T-X-V binding motif, are necessary and sufficient for the interaction with the hDLG1 PDZ1 and PDZ2 domains. This interaction also requires a free C-terminal carboxyl group of Tax-1 as illustrated by smaller chemical shift perturbations and a 360-fold decrease of the binding affinity in NMR experiments with a chemically modified, C-terminally amidated peptide Finally, to assess the binding specificity, we performed NMR experiments with a peptide corresponding to the last 10 amino acids of Tax-2 (NKEEADDNGD), encoded by another member of Deltaretroviruses, the HTLV-2 virus, which is not leukemogenic in human (Martinez et al., 2019) . Tax-2 protein has several structural similarities with Tax-1 but does not contain a PBM at its C-terminus. Compared to Tax-1 10mer (Figure 2A ), the interaction of the C-terminal Tax2 10mer peptide with 15 N hDLG1 PDZ1+2 tandem does not induce binding shifts of residues G235, F236, G330, or F331 ( Figures S2C-D) . Instead, it leads to chemical shift perturbations for several residues in discontinuous patches at the side and the back of the protein, outside the canonical peptide binding sites ( Figure S2E ). Stepwise addition of Tax2 10mer to 15 N hDLG1 PDZ1+2 results in incremental shifts for some of the backbone amide resonances, a behavior typical for the NMR fast exchange regime. The resulting binding curves are very shallow and do not saturate even at the 10-fold molar excess of the added peptide ( Figure S2F ), suggesting a very weak interaction (KD > 5 mM). Overall, the very weak binding to several protein regions outside the canonical peptide binding sites suggests that the interaction of Tax2 10mer with the hDLG1 PDZ1+2 tandem is aspecific. The above SAFE analysis highlighted endocytic vesicles membrane assembly as one of the enriched functions in the Tax-1/PDZ interactome ( Figure 1G ). The subnetwork of this function highlights syndecan binding protein 1 (SDCBP1 also known as syntenin-1), which has the highest connectivity in this subnetwork ( Figure 3A ) and also in our recently released human interactome map (Luck et al., 2020) (HuRI), http://www. interactomeatlas.org/). Syntenin-1 is particularly relevant for viral infections as it controls extracellular vehicles (EVs, also called exosomes) formation and cell-to-cell communication (Imjeti and Menck, 2017; Kim et al., 2015) . A direct interaction between syntenin-1 and programmed cell death 6 interacting protein (PDCD6IP also called ALG-3 interacting protein X (ALIX) (Christ et al., 2016) , was demonstrated and provided fundamental insights on the interplay between exosomes formation and the endosomal sorting complexes for transport (ESCRT) pathway (Baietti et al., 2012) . ALIX and other components of the ESCRT pathway are recruited by structural proteins of enveloped viruses to facilitate budding (Göttlinger et al., 1991; Hanson and Cashikar, 2012; Votteler and Sundquist, 2013) . Importantly, Tax-1, which is a regulator protein, has been shown to localize into EVs from HTLV-1 infected cells. These Tax-1-containing exosomes significantly contribute to viral spread and disease progression (Al Sharif et al., 2020; Pinto et al., 2019) . However, the molecular mechanisms controlling Tax-1 recruitment into EVs are unknown. We therefore tested the possibility that syntenin-1 could mediate Tax-1 localization into exosomes. As shown in Figure 3B , a siRNA targeting syntenin-1 expression is associated with a 13 reduction of the Tax-1 amount in exosomes, in accordance with Tax-1 translocation from the cytoplasm to the nucleus ( Figure 3C ). These findings suggest that syntenin-1 is required for Tax-1 translocation into EVs. We have extensively characterized the Tax-1/syntenin-1 interaction, as well as the interaction between Tax-1 and syntenin-2, using yeast two hybrid ( Figure 3D ), co-immunoprecipitations ( Figure 3E ), and GST-pulldown assays ( Figure 3F ). It appears that both syntenin proteins interact with Tax-1, but not with HTLV-2 Tax lacking a PBM, further demonstrating the specificity of Tax-1 PBM for syntenin PDZ domains. Thus, thanks to its PDZ domains, syntenin-1 can act as a molecular bridge, targeting Tax-1 to EVs via the ESCRT pathway. To determine the structural basis of Tax-1/syntenin-1 interaction, we employed solution NMR spectroscopy using natural abundance Tax-1 10mer and isotopically labelled PDZ1, PDZ2 and tandem PDZ1+2 syntenin-1 domains ( Figure 4A ). Addition of Tax-1 10mer to 15 N syntenin-1 PDZ1+2 leads to incremental chemical shift changes for some of the resonances, several of which progressively disappear in the course of the titration, which is symptomatic of the fast-to-intermediate NMR exchange regime. The chemical shift perturbations map onto the canonical peptide binding sites of both PDZ1 (residues I125 and G126) and PDZ2 (residues V209, G210, F211 and F213) domains, encompassing the β1/β2 loop, the β2 strand, and the α2 helix ( Figure 4A ). The binding effects are stronger for the PDZ2 domain and appear to extend to the back of the protein (residues T200, L232 and T266) ( Figure 4B ). These findings are consistent with an earlier study of syntenin-1-peptide interactions by X-ray crystallography that highlighted a binding-induced reorientation of the α2 helix (Grembecka et al., 2006a) , which could explain the chemical shift perturbations of the residues at the back of PDZ2 ( Figure 4B ). To determine the extent to which small molecules could interfere with Tax functions via inhibition of Tax-1/PDZ interactions, we used FJ9 ( Figure 5A ), a previously characterized small molecule inhibitor (Fujii et al., 2007) . FJ9 was able to suppress b-catenin-dependent tumor cell growth in a mouse model, following perturbation of the interaction between the Wnt receptor Frizzled-7 (Frz7) and Dishevelled Segment Protein 3 (DVL3), a PDZcontaining protein overexpressed in several cancer cells (Fujii et al., 2007) . Inhibition of Wnt-b-catenin signaling pathway was also shown to induce cell death in ATL and HTLV-1 transformed cell lines (Ying and Tao, 2009 ). To examine whether FJ9 could also disrupt the Tax-1/PDZ binding, we sought to investigate the FJ9 interaction with syntenin-1 PDZ domains. Unfortunately, the low FJ9 solubility in aqueous buffers precluded the binding analysis by NMR spectroscopy or other biophysical techniques. However, we have successfully performed an in silico molecular docking, which showed that FJ9 engages the canonical peptide binding site of syntenin-1 PDZ2 ( Figure 5B ). The FJ9 molecule makes extensive intermolecular contacts, including hydrophobic interactions with syntenin -1 residues V209, F211, I212, and F213 and a hydrogen bond with the backbone carbonyl atom of F211 ( Figure 5B ). Interestingly, these are the key residues mediating the Tax-1/syntenin-1 interaction ( Figures 4A-B) , suggesting that FJ9 competes with Tax-1 PBM for the binding to syntenin-1 PDZ domains. We next employed YFP-based bimolecular fluorescence complementation (BiFC) assay, which provides direct visualization and quantification of protein interactions in living cells (Hu et al., 2002) . We fused the two YFP-BiFC fragments to the N-and C-terminal ends of Tax-1 and syntenin coding sequences, respectively. Pairs of fusion proteins were expressed in HEK293 cells, and fluorescent complementation was quantified by flow cytometry. As shown in Figures 5C-D, interaction between Tax-1 and syntenin-2 exhibited the highest rate of complementation (33%), compared to Tax-1 and syntenin-1 (14%). This finding is consistent with our co-immunoprecipitation, pulldown, and co-localization as- The Tax-1 PBM has been shown to increase its oncogenic capacity in cell culture (Higuchi et al., 2007) . To examine whether FJ9 could inhibit the Tax-transformation activity, we infected a rat fibroblast cell line (Rat-1) with a VSV-G packaged lentiviral construct harboring the HTLV-1 tax gene and the blasticidin resistant gene (CS-EF-IB-Tax-1). Cells were selected by blasticidin, and pools of resistant cells seeded for a colony formation in soft agar assay (CFSA) in the presence or absence of FJ9 (100 μM). As shown in Figure 5E , HTLV-1 Tax transformed Rat-1 cells form multiple large colonies in CFSA, as previously described (Higuchi et al., 2007) . In the presence of FJ9, Rat-1 cells formed much smaller colonies, and their number was reduced compared to the cells treated with the vehicle. Thus, we conclude that FJ9 inhibits Tax-transformation activity in the Rat-1 To probe the role of the Tax/PDZ interaction in cell-to-cell HTLV-1 transmission, we performed a co-culture experiment using Jurkat T-cell line reporter expressing luciferase under the control of the 5'LTR HTLV-1 promoter, and an HTLV-1 producing cell line (MT2) or as negative control, a human T cell leukemia cell line (CEM) Figure 5G ). The reporter exhibits basal levels of luciferase expression, which increase following co-culture with MT2 cells and transactivation of the 5'LTR by Tax. We showed that at 100 μM, a dose which does not induce significant cell death compared to the vehicle ( Figure 5H ), FJ9 was able to inhibit HTLV-1 transmission, reducing Tax transactivation to the basal level ( Figure 5I ). Despite the discovery of viruses in the early 1900s, and the demonstration of their path- targets. In this work, our goal was to demonstrate that one of such host determinants, the PDZ-domain containing proteins, could provide insights into how to identify small molecule inhibitors of viral-host PPIs. The methodology used here defines a pipeline that could be applied to any pathogenic parasite of interest, without prior knowledge on the druggability of its encoded gene products. First, we chose HTLV-1 Tax (Tax-1), a highly connected viral oncoprotein, and demonstrated that it has the potential to interact with 10% of the human proteome through defined interactome modules. Second, we systematically mapped the interactome of Tax-1 with one of its cellular modules, the PDZ domain, and assessed the biological significance of that particular portion of the Tax interactome by integrating expression and genomic data from HTLV-1-infected patients. Third, we determined the structural basis of Tax-1 C-terminal motif binding to canonical sites of hDLG1 and syntenin-1 PDZ domains. Finally, we demonstrated that a small molecule targeting Tax-1-syntenin interactions could inhibit Tax-1 functions in cellular models. By testing 248 out of 268 PDZ domains identified in 151 human proteins, our results establish the first comprehensive Tax-1-PDZ interactome, highlighting a set of 54 PDZ-domain containing proteins. Comparative analyses with similar unbiased mapping for other viral proteins are now possible. For example, hDLG1 and SCRIB have also been shown to interact with HPV E6, Ad9 E4orf1, IAV NS1, HTLV-1 Env, HIV-1 Env and HCV core proteins, but with different biological outcomes, including perturbations of cell polarity, signaling, or apoptosis (Thomas and Banks, 2018) . Interestingly, the conserved C-terminal PBM motifs of these viral proteins highly correlate with viral pathogenicity, independently of their substantial differences in genome content (RNA and DNA viruses) or mode of infection (acute and persistent families of viruses). Other examples include syntenin-1 and PTPN13 able to interact with distant viral proteins such as HTLV Tax-1 (this study), but also with coronaviral E, 3A and N proteins (Caillet-Saguy et al., 2021; Jimenez-Guardeño et al., 2014). These pan-viral interactors may thus constitute ideal drug targets for the discovery of broad-spectrum antiviral inhibitors. In fact we showed that, compared to hDLG1, syntenin-1 PDZ domains exhibit low affinities for the Tax-1 PBM peptide, as previously reported for other systems (Grembecka et al., 2006a; Latysheva et al., 2006) . The low affinity appears to be a general feature of the syntenin-1 PDZ-mediated binding, underlying its role in orchestrating the dynamic interaction networks, capable of rapid response to environmental changes (Manjunath et al., 2018) . Consistent with the above findings, we previously highlighted the importance of syntenin-1 as a potential recruiter of proteins into extracellular vesicles (EV) (Luck et al., 2020) . Here, we further demonstrate that through its PDZ domains syntenin-1 recruits Tax-1 into EV exosomes ( Figure 3B ), providing a mechanistic explanation for the presence of the Tax-1 protein in exosomes isolated from HTLV-1-infected cell lines, HAM/TSP or ATL patient samples (Barclay et al., 2017; Jaworski et al., 2014; Otaguiri et al., 2018) . A compelling hypothesis is that blocking syntenin-1/viral proteins interactions could elicit a two-pronged inhibitory activity: (i) at the viral budding step of enveloped viruses replication, where syntenin-1 could connect viral particles to the ESCRT pathway (Göttlinger et al., 1991; Hanson and Cashikar, 2012; Votteler and Sundquist, 2013) ; and (ii) in cell-tocell viral transmission and inflammatory immune response induction by exosomes containing syntenin-1/viral proteins complexes (Otaguiri et al., 2018) . Interestingly, a small molecule inhibitor of the syntenin-exosomal pathway was also shown to potentially inhibit tumor exosomal communication in breast carcinoma cells (Leblanc et al. 2020) . Structural insights are essential for design of small molecule inhibitors of interactions between viral proteins and cellular protein modules. X-ray structures of hDLG1 PDZ1 and (Rance et al., 1983) . These findings suggest that the C-terminal part of Tax-1 adopts the canonical binding mode displayed in the X-ray structures of hDLG1 PDZ-peptide complexes ( Figure 2C -D) (Rance et al., 1983; Zhang et al., 2011) . Comparison across different PDZ-peptide systems shows that hDLG1 residues G235, F236, G330, and F331 act as binding hot spots, which should be amenable to traditional drug discovery efforts aiming at designing potent small molecule binders. As a proof-of-concept, we demonstrated a clear correlation between PPI inhibition and impairment of Tax functions, including its transformation ability and HTLV-1 cell-to-cell transmission. The combination of our experimental strategies should allow the development of anti- Tax-1 compounds for future pre-clinical and clinical studies of drug candidates for treatment of HTLV-1-induced diseases. In this work, we have only characterized one portion of the Tax-1 interactome, the PDZome, yet there are other host protein domains and motifs specifically interacting with HTLV-1 Tax (Table S1 ). Therefore, it will be important to obtain structural data for other interactions such as those involving CREB/ATF and NF-kB transcription factors, which are key drivers of the molecular mechanisms underlying HTLV-1 pathogenesis. Furthermore, the FJ9 small molecule used in this study is an indole-2-carbinol-based chemical scaffold compound that was previously identified to inhibit Frz7-DVL3 interaction (Fujii et al., 2007) . We envisage that other chemical structures could mimic the Tax-1 PBM binding to PDZ domain. Subsequent follow-up studies, e.g. employing computational structure-based docking methodologies (Gorgulla and Boeszoermenyi, 2020) , could potentially identify more potent small-molecular inhibitors targeting the Tax-1/PDZ binding interface. Finally, although it has been shown that the Tax-1 PBM is essential to sustain Tcell proliferation in HTLV-1-infected humanized mice (Pérès and Blin, 2018), we did not examine the in vivo impact of inhibiting Tax-1 PBM-PDZ interaction using small molecule inhibitors. 20 Despite 60 years of molecular virology research, the arsenal of antiviral drugs remains dangerously small, with only ~90 molecules available today (De Clercq and Li, 2016) . Those are strongly biased towards a single virus, the human immunodeficiency virus (HIV) and its key enzymes (reverse transcriptase, protease and integrase The authors declare no competing interests. by the FJ9 small molecule. Statistical analyses were done using unpaired t-tests, where **** indicate a p value <0.0001, *** a p value <0.001, and ** a p value <0.01. Figure S1 . A comprehensive analysis of the Tax-1 interactome with human PDZcontaining proteins. (A) PPIs between Tax-1 and human PDZ domain-containing proteins identified by yeast 2-hybrid (Y2H) and validated by GPCA, or using a pull-down assay (blue); and known PPIs from the literature (yellow). (B) Literature reported interactions between Tax-1 and human PDZ domain-containing proteins. (C) Heat maps of the Y2H assay by testing a library of 244 individualized PDZ domain-containing proteins fused to AD against DB-Tax-1. A growth score of "0", "1" or "2" indicates a null, weak or strong interaction, respectively. (D) Representative Y2H assay plate lacking tryptophan (Trp), leucine (Leu), and histidine (His) where a growth score of "0", "1" or "2" indicates a null, weak or strong interaction between DB- Tax Figure 2C ,D. The modeled PBM peptide, bound to the canonical PDZ site, is shown in sticks. (E) The spectra of the 15 N hDLG1 PDZ1+2 tandem in the absence and presence of 6mer-N (black and green, respectively).(F) Δδ avg of the hDLG1 PDZ1+2 in the presence of 6 molar equivalents of 6mer-N.The labels identify the sole residue with a significant binding shift. (A) Overlay of 1 H, 15 N HSQC spectra of 15 N hDLG1 PDZ1 in the absence and presence of Tax-1mod peptide (black and red, respectively). The labels identify the backbone amide resonances affected by the Tax-1 10mer binding (cf. Figure 2E ). (B) Δδ avg of hDLG1 PDZ1 at saturating amounts of Tax-1mod. (C) Chemical shift mapping of the Tax-1mod binding. The molecular surface of hDLG1 PDZ1 is colored as in Figure 2G . The modeled C-terminal part of the Tax-1 peptide, bound to the canonical PDZ site, is shown in sticks. (D) NMR chemical shift titration of 15 N hDLG1 PDZ1 with the Tax-1mod peptide. Open and filled symbols refer to the chemical shift perturbations (Δδ) of the backbone H and N atoms, respectively, of F236 (squares), G240 (circles), I258 (triangles), I259 (diamonds), and L296 (stars). The curves were fitted simultaneously to a binding model with the shared K D (Equation 1 ). The solid lines show the best fits with the K D value of 951 ± 34 µM. (A) Δδ avg of 15 N syntenin-1 PDZ1+2 tandem in the presence of 6 molar equivalents of PBM 4mer. Residues with backbone amides broadened beyond the detection limit upon complex formation are indicated by red bars. (B) Chemical shift mapping of the PBM 4mer binding. The molecular surface of syntenin-1 PDZ1+2 tandem is colored as in Figure 4B . The modeled PBM peptides, bound to the canonical PDZ sites, are shown in sticks. (C) Δδ avg of the 15 N syntenin-1 PDZ1+2 tandem in the presence of 7 molar equivalents of 6mer-N. Several residues with significant binding shifts are labelled. Further information and request can be directed to Jean-Claude Twizere (jeanclaude.twizere@uliege.be) Plasmids generated in this study are available upon request and approval of the Material Transfer Agreement (MTA) by the University of Liege. In order to predict potential human host targets of Tax Short linear motifs in Tax sequence were searched using SLiMSearch 2.0 tool from SLiM-Suite (Davey et al., 2011) . SLiMSearch takes as input a set of motif consensus and search them against queried protein sequences, The motif consensus is represented as a regular expression and contains ordinary and special characters. Permitted ordinary characters are single-letter amino acid codes. The special characters are based on regular expression syntax as defined in the Eukaryotic Linear Motif resource (Kumar et al., 2020) .To reduce the number of false positive instances of motif consensus, in our prediction method, only those motif instances which fall in disordered region of Tax sequence were retained. In order to detect motif instances, situated in disordered region of Tax sequence each motif instances was assigned a disorder score (DSc): Where, DSp is the IUPred disorder score of the residue at position p in Tax protein sequence, s is the start position of the consensus match, e is the end position of the consensus match, l is the consensus match length. Motif instances whose disorder score was greater than 0.4 were retained (Zanzoni et al., 2017b; a) . The rationale behind selecting motif instances located in the disordered region is that the functional SliMs are largely located in the unstructured region of protein (Edwards et al., 2012; Fuxreiter et al., 2007) . The Gateway cloning method was used to transfer the DNA encoding the PDZ domain into the AD yeast expression vector pACT2. All PDZ domains were transformed into the yeast haploid strain Y187 (MATa, 112 To detect the Tax-1 partners by pulldown, PDZ domains were expressed in HEK293 cells. GST-Tax-1 or GST-Tax-2 protein as negative control were expressed separately and then purified with glutathione Sepharose beads. The beads now containing the fusion proteins were then incubated with PDZ-flag containing cell lysate. The detection of Flag by immunoblot indicates an interaction between Tax-1 and the PDZ domain protein HEK293 cells were seeded in 24-well plates, then transfected with GL1 and / or GL2 plasmids expressing the fusion proteins 24 hours later. Luciferase activity was measured on the lysates in 96-well plates. The results were normalized relative to the value of the Luciferase control Renilla. The normalized luciferase ratio was calculated as follows: NLR = co-transfection luciferase value (GL1 + GL2) / (GL1 luciferase value alone + GL2 luciferase value alone). An interaction is considered positive or validated when NLR ≥ 3.5. Cell lyses and luminescence measurements were performed in triplicate for each condition. Collected cells were lysed with the passive lysis buffer (Promega). For luciferase measurements, 100 μl of the lysates were used, to which the substrates Firefly and Renilla stop & glo (Promega) were added. Luciferase measurements were performed in 96-well plates using an automated DLR machine. Firefly luciferase values were normalized to Rluc values, and the calculated ratio represents luciferase activity. The Tax-1 10-mer (SEKHFRETEV), Tax2 10mer (NKEEADDNGD), TAX-1 PBM 4-mer (ETEV), and Tax-1 N-ter 6-mer (SEKHFR) peptides were synthesized by Biomatik, Canada, as acetate salts of >98% purity. Plasmids harboring the PDZ domains of hDLG1 tandem PDZ1+2 and syntenin-1 (PDZ1, PDZ2 and Tandem PDZ1+PDZ2) were purchased from GeneScript,hDLG1 PDZ1,2 and 3 were obtained from addgene (Tonikian et al., 2008a) and transformed into E. coli BL21. A single bacterial colony was used to inoculate 5.0 mL LB and grown overnight at 37 °C, from which 1.0 mL was used as starter culture for expressing the proteins in 100 mL LB. For expression, the culture was grown at 37°C, shaking (180rpm) proteins were produced in E coli grown in the minimal medium following a published protocol (Volkov et al., 2013) . GST-fused proteins were purified using glutathione sepharose 4B beads (GE HealthCare) equilibrated in 1x Phosphate Buffered Saline (PBS) buffer. Bound proteins were eluted with 50 mM Tris, pH 8, 10 mM reduced glutathione. GST tag was then cleaved off using preScission protease (Sigma) overnight at 4°C. The mixture was then loaded back to the column equilibrated in PBS, and PDZ proteins were collected from the flow-through. These were further purified using a HiPrep 16/60 Sephacryl S-100 HR size exclusion column, equilibrated in 20 mM NaPi pH 6.0, 50 mM NaCl, 2 mM DTT. ITC measurements were carried out on an Microcal ITC200 calorimeter at 25 °C. All experiments were performed in 10 mM Tris pH 8.0. The PDZ proteins were loaded into the sample cell at 20 μM concentration, and 250 μM of peptides (dissolved in the same buffer) were loaded into the syringe. Titrations comprised 26 × 1.5 μL injections of peptide into the protein, with 90 s intervals. An initial injection of 0.5 μL was made and discarded during data analysis. The data were fitted to a single binding site model using the Microcal LLC ITC200 Origin software provided by the manufacturer. All NMR spectra were acquired at 298 K in 20 mM sodium phosphate 50 mM NaCl pH 6.0, 2 mM DTT and 10 % D2O for the lock on a Bruker Avance III HD 800 MHz spectrometer (equipped with a TCI cryoprobe) or a Varian Direct-Drive 600 MHz spectrometer (fitted with a PFG-Z cold probe). The data were processed in NMRPipe (Delaglio et al., 1995) and analyzed in CCPNMR (Vranken et al., 2005) . The NMR titration curves were analyzed with a two-parameter nonlinear least squares fit using a one-site binding model corrected for the dilution effect (Kannt et al., 1996) , Eq. 1 : where Δδbinding is the chemical shift perturbation at a given protein/peptide ratio; The double stranded interfering RNAs (siRNAs) were purchased from Eurogentec (Liège, Belgium). The following siRNAs were used in this work:SDCBP sense (5'-GCAAGAC-CUUCCAGUAUAA-3'), SDCBP anti-sense (5'-UUAUACUGGAAGGUCUUGC-3'), and control siRNA (5′-GGCUGCUUCUAUGAUUAUGtt-3'). The cells were lysed and clarified by centrifugation. After incubating the lysate for 5 min at 100 ° C in the presence of Laemmli buffer, electrophoresis was carried out on 10% SDS-PAGE, followed by immunoblotting. The following antibodies were used: anti-Flag, anti-SDCBP, anti-tax and anti-ubiquitin. HeLa cells were seeded on coverslips in 24-well plates for 48 hours. The cells were then washed three times at 37 °C with PBS, fixed with PBS-4% paraformaldehyde for 15 min, and then washed twice more with PBS. Cells were then permeabilized with PBS-0.5% Triton X-100 for 20 min, incubated with the PBS-FBS 20% blocking solution for 30 min, and then washed twice with PBS. After that, the cells were incubated with the corresponding primary antibody diluted in PBS-Triton X-100 0.5% for 2 hours, washed 3 times with PBS, and incubated 1 hour with the corresponding Alexa conjugated secondary antibodies (Invitrogen) diluted. 1/1000 in PBS-Triton. Finally, the cells were washed 3 times with PBS and mounted on glass coverslips with ProLonf Gold Antifade containing DAPI (life technologies). The slides were examined by confocal microscopy with Leica TCS SP2 or Nikon A1R confocal microscope with a 60X objective. The images were taken with a resolution of 1024x1024 pixels and subsequently processed and assembled with LEICA LAS AF Lite 5 or A1R Software. Total RNA was extracted using the GeneJET RNA purification kit (Thermo scientific). Total RNA was reverse transcribed with random primers using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific). QPCR was performed using Roche's SYBER Green detection in the Lightcycler 480 (Roche). Quantification of mRNA was performed by relative normalization to the constitutive gene GAPDH. The relative expression levels were calculated for each gene using the ΔCt or ΔΔCt method. Exosomes were isolated from cell supernatants. First, 50-100 ml of the supernatants were centrifuged at 2000 xg for 20 minutes at room temperature to remove floating cells. Second, the cell-free supernatant was then centrifuged at 12,000 xg for 45 minutes and filtered (0.22 μm) to remove additional cell debris. Third, the samples were ultracentrifuged for 2 h at 100,000 g 4 ° C. The pellet containing the exosomes was washed with PBS, then centrifuge for an additional 2 hours at the same speed, and finally suspended in lysis buffer (50 mM Tris-HCl, 1% SDS, pH 7.5, supplemented with protease and phosphatase inhibitors). The lysate was stored at 4 ° C until use in Western blot. The interaction of Tax-1 with SDCBP was re-validated by immunoprecipitation. HEK 293T cells were transfected with Flag-Syntenin, Tax-1 and Tax-2. Fourty-eight hours after the transfection, the cells were lysed and protein expression verified by a Western blot, using specific anti-Tax-1 and anti-Flag antibodies. HeLa cells were transfected with constructs expressing Tax-1 and control siRNA or SDCBP siRNA. Cells were fixed, permeabilized and labeled with primary anti-Tax-1 antibodies and secondary antibodies conjugated to Alexa 488 24 hours post transfection. These were then analyzed under a confocal microscope. Using Nikon A1R software, we quantified the intensities of ROI (regions of interest) which allowed to determine the cytoplasm / nucleus ratio of protein localization. Statistically significant data are indicated with * (P <0.05), ** (P <0.01), or *** (P <0.001). HeLa cells were transfected with constructs expressing Tax-1 and / or Flag-SDCBP. The cells were fixed, permeabilized and labeled with anti-Tax-1 or anti-Flag antibodies and conjugated secondary antibodies Alexa 488 or Alexa 633 24 hours post transfection. These were then analyzed under a confocal microscope as described for the previous section. The crystal structure of the complex of syntenin-1 PDZ2 and syndecan-4 peptide was retrieved from RSCB Protein Data Bank (PDB:1OBY) (Kang et al., 2003) . Protein preparation wizard in the Schrodinger package was used to pre-process, minimize, and refine the protein (Sastry et al., 2013) . With all water molecules removed, the missing hydrogen/side-chain atoms and appropriate charge and protonation state of the protein at pH 7.0 were assigned using the Protein Preparation wizard module in Maestro (Bell et al., 2012) . Molecular dynamics simulations were performed using GPU-accelerated Amber 18 software package (Lee et al., 2018) . LEAP module and antechamber were used to optimize the protein and the inhibitor, respectively. Subsequently, the restrained electrostatic potential (RESP) (Dupradeau et al., 2010) and the general amber force field (GAFF) (Wang et al., 2004) were used to generate partial atomic charges and assign the force field parameters (Hornak et al., 2006) . The system was neutralized by adding four Cl - [28] water model of 10 Å distance was used, and the simulations were carried out using periodic boundary conditions. The Particle Mesh Ewald (PME)(Harvey and De Fabritiis, 2009) method was employed for long-range electrostatic interactions with a 12 Å cut-off. The SHAKE algorithm (Ryckaert et al., 1977) was used to constrain all the bonds to hydrogen atoms. A 2,500-step geometric minimization was performed to remove any possible steric clashes. The topology and coordinate files were generated, and the systems were subjected to 60ns MD simulations. Finally, a post analysis of the MD trajectories was carried out by computing the binding energies, root-mean-square deviation (RMSD), and root-mean-square fluctuation (RMSF) using CPPTRAJ and PTRAJ modules (Roe and Cheatham III, 2013) implemented in the Amber 18 package. To test the effect of FJ9 on the transformation of cells induced by the HTLV-1 virus, Rat-1 cells were stably transduced (rat fibroblasts) with a vector lentiviral pTax-1-IB. A pool of resistant cells (5 × 10 3 ) was seeded in DMEM / 10% FCS containing 0.33% agarose, superimposed on a layer of 0.5% agarose in a 6-well plate with 100 μM of FJ9 or DMSO. After three weeks in culture, colonies that usually form "foci" were examined under a light microscope and photos taken to quantify the effect of FJ9. The Tax-1 induced foci count was performed using ImageJ software. Virus-producing cells (MT2) were co-cultured with cells containing the gene encoding luciferase under the control of the HTLV-1 LTR5' viral promoter. The cells were grown for 24 hours in the presence or absence of 100 μM of FJ9. Luciferase activation assay was performed as described previously. To estimate the ability of FJ9 to inhibit the interaction between Tax-1 and SDCBP, the fluorescent protein YFP was used. HEK 293T cells were transfected with the C terminal fragment of YFP protein fused to SDCBP and with the N terminal fragment of YFP fused to Tax-1., The cells were then treated with 300 μM FJ9 or DMSO 24 hours later and further cultured for an additional 24 hours before flow cytometry analysis. The results represent the mean and standard deviation of three independent experiments. The values of the graphs are presented as the mean ± standard deviation, calculated on at least three independent experiments. Significance was determined using a t test with two variables (comparison of means). The thresholds of value P (p value) are represented as follows; * = p <0.05; ** = p <0.01 and *** = p <0.001 Zhang, Z., Li, H., Chen, L., Lu, X., Zhang, J., Xu, P., Lin, K., and Wu, G. (2011) . Molecular basis for the recognition of adenomatous polyposis coli by the Discs Large 1 protein. PloS one 6, e23507. IL16 PDZD4 SNTB1 SDCBP2 GRIP1 LMO7 CYTIP SDCBP SHANK1 MAGI1 LNX2 RADIL PDZD7 MAGI3 PARD6A RIMS1 PDZRN3 SHROOM4 GIPC2 GRIP2 SNTG1 DEPTOR DLG4 APBA1 PDLIM2 RHPN1 PDLIM7 PARD6B DLG5 MAGI2 LNX1 PDZD2 PDZD3 RGS3 PDLIM4 DVL3 INTU DLG1 MPDZ DVL3 SCRIB LIN7A APBA3 FRMPD2 SLC9A3R1 TAX1BP3 PREX1 PPP1R9B DLG3 PTPN3 SNTB2 S2-mutPDZ1+wtPDZ2 S2-wtPDZ1+mutPDZ2 S2-mutPDZ1+mutPDZ2 S1 S2 S1-PDZ1 S2-PDZ1 S2-PDZ2 S1-PDZ2 S1-PDZ1+S2-PDZ2 S1-PDZ1+S2-PDZ1 S2-PDZ1+S1-PDZ2 S1-PDZ2+S2-PDZ2 GFP fusion ns Protein Interaction Figure S1 . A comprehensive analysis of the Tax-1 interactome with human PDZ-containing proteins. (A) PPIs between Tax-1 and human PDZ domain-containing proteins identified by yeast 2-hybrid (Y2H) and validated by GPCA, or using a pulldown assay (blue); and known PPIs from the literature (yellow). (B) Literature reported interactions between Tax-1 and human PDZ domaincontaining proteins. (C) Heat maps of the Y2H assay by testing a library of 244 individualized PDZ domain-containing proteins fused to AD against DB-Tax-1. A growth score of "0", "1" or "2" indicates a null, weak or strong interaction, respectively. (D) Representative Y2H assay plate lacking tryptophan (Trp), leucine (Leu), and histidine (His) where a growth score of "0", "1" or "2" indicates a null, weak or strong interaction between DB- Tax PARD6A PARD6B PARD6G PDLIM1 PDLIM2 PDLIM3 PDLIM4 PDLIM5 PDLIM7 PDZD11 PDZD2_1 PDZD2_2 PDZD2_3 PDZD2_4 PDZD2_5 PDZD2_6 PDZD3_1 PDZD3_2 PDZD3_3 PDZD3_4 PDZD4 PDZD7_1 PDZD7_2 PDZD8 PDZK1_1 PDZK1_2 PDZK1_3 PDZRN3_1 PDZRN3_2 PDZRN4 PICK1 PPP1R9A PPP1R9B PREX1_1 PREX1_2 PRX PSMD9 PTPN13_1 PTPN13_2 PTPN13_3 PTPN13_4 PTPN13_5 PTPN3 PTPN4 RADIL RAPGEF2 Figure 2C ,D. The modeled PBM peptide, bound to the canonical PDZ site, is shown in sticks. (E) The spectra of the 15 N hDLG1 PDZ1+2 tandem in the absence and presence of 6mer-N (black and green, respectively).(F) Δδ avg of the hDLG1 PDZ1+2 in the presence of 6 molar equivalents of 6mer-N.The labels identify the sole residue with a significant binding shift. Figure S4 . Control NMR experiments with the C-terminally amidated Tax-1 10mer peptide (Tax-1mod). (A) Overlay of 1 H, 15 N HSQC spectra of 15 N hDLG1 PDZ1 in the absence and presence of Tax-1mod peptide (black and red, respectively). The labels identify the backbone amide resonances affected by the Tax-1 10mer binding (cf. Figure 2E ). (B) Δδ avg of hDLG1 PDZ1 at saturating amounts of Tax-1mod. (C) Chemical shift mapping of the Tax-1mod binding. The molecular surface of hDLG1 PDZ1 is colored as in Figure 2G . The modeled C-terminal part of the Tax-1 peptide, bound to the canonical PDZ site, is shown in sticks. (D) NMR chemical shift titration of 15 N hDLG1 PDZ1 with the Tax-1mod peptide. Open and filled symbols refer to the chemical shift perturbations (Δδ) of the backbone H and N atoms, respectively, of F236 (squares), G240 (circles), I258 (triangles), I259 (diamonds), and L296 (stars). The curves were fitted simultaneously to a binding model with the shared K D (Equation 1 ). The solid lines show the best fits with the K D value of 951 ± 34 µM. Figure S5 . Control syntenin-1 NMR-experiments with Tax-1 PBM 4mer and 6mer-N peptides. (A) Δδ avg of 15 N syntenin-1 PDZ1+2 tandem in the presence of 6 molar equivalents of PBM 4mer. Residues with backbone amides broadened beyond the detection limit upon complex formation are indicated by red bars. (B) Chemical shift mapping of the PBM 4mer binding. The molecular surface of syntenin-1 PDZ1+2 tandem is colored as in Figure 4B . The modeled PBM peptides, bound to the canonical PDZ sites, are shown in sticks. (C) Δδ avg of the 15 N syntenin-1 PDZ1+2 tandem in the presence of 7 molar equivalents of 6mer-N. Several residues with significant binding shifts are labelled. Extracellular Vesicles in HTLV-1 Communication: The Story of an Invisible Messenger trans activation of the tumor necrosis factor alpha promoter by the human T-cell leukemia virus type I Tax1 protein The PDZ domain binding motif (PBM) of human T-cell leukemia virus type 1 Tax can be substituted by heterologous PBMs from viral oncoproteins during T-cell transformation Syndecan-syntenin-ALIX regulates the biogenesis of exosomes HTLV-I tax induces cellular proteins that activate the kappa B element in the IL-2 receptor alpha gene Isolation of Exosomes from HTLV-Infected Cells Systematic Functional Annotation and Visualization of Biological Networks Tax fingerprint in adult T-cell leukemia How I treat adult T-cell leukemia/lymphoma PrimeX and the Schrödinger computational chemistry suite of programs The human PDZome: a gateway to PSD95-Disc large-zonula occludens (PDZ)-mediated functions The human PDZome: a gateway to PSD95-Disc large-zonula occludens (PDZ)-mediated functions Human T-cell leukemia virus type I Tax associates with and is negatively regulated by the NF-kappa B2 p100 gene product: implications for viral latency The HTLV-1 tax interactome Interaction of HTLV-1 Tax with minichromosome maintenance proteins accelerates the replication timing program Host PDZcontaining proteins targeted by SARS-CoV-2 Discovery of a new human T-cell lymphotropic virus (HTLV-3) in Central Africa Benchmarking a luciferase complementation assay for detecting protein complexes Human Tlymphotropic virus type 1 Tax protein complexes with P-TEFb and competes for Brd4 and 7SK snRNP/HEXIM1 binding /II function as parallel ESCRT-III recruiters in cytokinetic abscission Probing the supramodular architecture of a multidomain protein: the structure of syntenin in solution How I treat adult T-cell leukemia/lymphoma The transcriptionally active factors mediating the effect of the HTLV-I Tax transactivator on the IL-2R alpha kappa B enhancer include the product of the c-rel proto-oncogene SLiMSearch 2.0: biological context for short linear motifs in proteins Approved Antiviral Drugs over the Past 50 Years NMRPipe: a multidimensional spectral processing system based on UNIX pipes ELM-the database of eukaryotic linear motifs The REd. Tools: Advances in RESP and ESP charge derivation and force field library building Interactome-wide prediction of short, disordered protein interaction motifs in humans Comparative biology of human T-cell lymphotropic virus type 1 (HTLV-1) and HTLV-2 Early application of related SCT might improve clinical outcome in adult T-cell leukemia/lymphoma c-fos promoter trans-activation by the tax1 protein of human T-cell leukemia virus type I An antagonist of dishevelled protein-protein interaction suppresses βcatenin-dependent tumor cell growth Polycomb-dependent epigenetic landscape in adult T-cell leukemia Local structural disorder imparts plasticity on linear motifs Signatures of pleiotropy, economy and convergent evolution in a domain-resolved map of human-virus protein-protein interaction networks Antibodies to human T-lymphotropic virus type-I in patients with tropical spastic paraparesis An open-source drug discovery platform enables ultra-large virtual screens Effect of mutations affecting the p6 gag protein on human immunodeficiency virus particle release The binding of the PDZ tandem of syntenin to target proteins The binding of the PDZ tandem of syntenin to target proteins Pharmacological inhibition of syntenin PDZ2 domain impairs breast cancer cell activities and exosome loading with syndecan and EpCAM cargo Taxdependent stimulation of G1 phase-specific cyclin-dependent kinases and increased expression of signal transduction genes characterize HTLV type 1-transformed T cells Physical interaction of human T-cell leukemia virus type 1 Tax with cyclin-dependent kinase 4 stimulates the phosphorylation of retinoblastoma protein Multivesicular body morphogenesis IKKγ serves as a docking subunit of the IκB kinase (IKK) and mediates interaction of IKK with the human T-cell leukemia virus Tax protein Visualization of interactions among bZIP and Rel family proteins in living cells using bimolecular fluorescence complementation PDZ domains: structural modules for protein complex assembly Syntenin mediates SRC function in exosomal cell-to-cell communication Dose-intensified chemotherapy alone or in combination with mogamulizumab in newly diagnosed aggressive adult T-cell leukaemialymphoma: a randomized phase II study Human T-lymphotropic virus type 1-infected cells secrete exosomes that contain Tax protein PCAF interacts with tax and stimulates tax transactivation in a histone acetyltransferase-independent manner The PDZ-binding motif of severe acute respiratory syndrome coronavirus envelope protein is a determinant of viral pathogenesis Role of adapter function in oncoprotein-mediated activation of NF-κB: human T-cell leukemia virus type I Tax interacts directly with IκB kinase γ A new subtype of human T-cell leukemia virus (HTLV-II) associated with a T-cell variant of hairy cell leukemia SUV39H1 interacts with HTLV-1 Tax and abrogates Tax transactivation of HTLV-1 LTR Molecular roots of degenerate specificity in syntenin's PDZ2 domain: reassessment of the PDZ recognition paradigm The role of acidic residues of plastocyanin in its interaction with cytochrome ƒ The HTLV-I Tax oncoprotein targets the retinoblastoma protein for proteasomal degradation The role of cyclin D2 and p21/waf1 in human T-cell leukemia virus type 1 infected cells EVpedia: a community web portal for extracellular vesicles research ELM-the eukaryotic linear motif resource in 2020 Control of cAMP-regulated enhancers by the viral transactivator Tax through CREB and the co-activator CBP Binary Interactome Models of Inner-Versus Outer-Complexome Organization Syntenin-1 is a new component of tetraspanin-enriched microdomains: mechanisms and consequences of the interaction of syntenin-1 with CD63 Binding of human virus oncoproteins to hDlg/SAP97, a mammalian homolog of the Drosophila discs large tumor suppressor protein GPU-accelerated molecular dynamics and free energy methods in Amber18: performance enhancements and new features Protein-protein interactions and gene expression regulation in HTLV-1 infected cells The HTLV-1 Tax protein inhibits formation of stress granules by interacting with histone deacetylase 6 HTLV-1 transactivator induces interleukin-2 receptor expression through an NF-κB-like factor Solution structure of the hDlg/SAP97 PDZ2 domain and its mechanism of interaction with HPV-18 papillomavirus E6 protein Phosphoinositide-3 kinase-PKB/Akt pathway activation is involved in fibroblast Rat-1 transformation by human T-cell leukemia virus type I tax A reference map of the human binary protein interactome Structure function relations in PDZdomain-containing proteins: Implications for protein networks in cellular signalling Comparative virology of HTLV-1 and HTLV-2 Evidence for aberrant activation of the interleukin-2 autocrine loop by HTLV-1-encoded p40x and T3/Ti complex triggering Interference of HTLV-1 Tax Protein with Cell Polarity Regulators: Defining the Subcellular Localization of the Tax-DLG1 Interaction 3did: a catalog of domain-based interactions of known three-dimensional structure Overview of Targeted Therapies for Adult T-Cell Leukemia/Lymphoma Human Tlymphotropic virus, type 1, tax protein triggers microtubule reorientation in the virological synapse HTLV-I Tax and cell cycle progression Cell cycle dysregulation by HTLV-I: role of the tax oncoprotein The Tax protein of HTLV-1 promotes oncogenesis in not only immature T cells but also mature T cells TAX-mRNA-Carrying Exosomes from Human T Cell Lymphotropic Virus Type 1-Infected Cells Can Induce Interferon-Gamma Production In Vitro How to Control HTLV-1-Associated Diseases: Preventing de Novo Cellular Infection Using Antiviral Therapy Role for Akt/protein kinase B and activator protein-1 in cellular proliferation induced by the human T-cell leukemia virus type 1 tax oncoprotein PDZ domain-binding motif of Tax sustains T-cell proliferation in HTLV-1-infected humanized mice HTLV-1 Extracellular Vesicles Promote Cell-to-Cell Contact Isolation of a new type C retrovirus (HTLV) in primary uncultured cells of a patient with Sezary T-cell leukaemia Improved spectral resolution in COSY 1H NMR spectra of proteins via double quantum filtering PTRAJ and CPPTRAJ: software for processing and analysis of molecular dynamics trajectory data Interpreting cancer genomes using systematic host network perturbations by tumour virus proteins Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments PDZ domains and the organization of supramolecular complexes Host-pathogen interactome mapping for HTLV-1 and-2 retroviruses Host-pathogen interactome mapping for HTLV-1 and -2 retroviruses Tax oncoprotein of HTLV-1 binds to the human homologue of Drosophila discs large tumor suppressor protein, hDLG, and perturbs its function in cell growth control The landscape of viral expression and host gene fusion and adaptation in human cancer HTLV-1 and apoptosis: role in cellular transformation and recent advances in therapeutic approaches Upsetting the Balance: When Viruses Manipulate Cell Polarity Control A specificity map for the PDZ domain family A specificity map for the PDZ domain family PDZ domain-binding motif of human T-cell leukemia virus type 1 Tax oncoprotein is essential for the interleukin 2 independent growth induction of a T-cell line Conformational characterization of synapse-associated protein 97 by nuclear magnetic resonance and small-angle X-ray scattering shows compact and elongated forms Interaction of retroviral Tax oncoproteins with tristetraprolin and regulation of tumor necrosis factor-α expression Human T-cell leukemia virus type-1 Tax oncoprotein regulates G-protein signaling Recent advances in the treatment of adult T-cell leukemia-lymphomas The viral oncoproteins Tax and HBZ reprogram the cellular mRNA splicing landscape Expression, purification, characterization, and solution nuclear magnetic resonance study of highly deuterated yeast cytochrome c peroxidase with enhanced solubility Virus budding and the ESCRT pathway The CCPN data model for NMR spectroscopy: development of a software pipeline Development and testing of a general amber force field Influenza virus-host interactome screen as a platform for antiviral drug development Human T-cell leukemia virus type 1 Tax enhances serum response factor DNA binding and alters site selection Emergence of unique primate T-lymphotropic viruses among central African bushmeat hunters Protein profile of tax-associated complexes Retroviral oncoprotein Tax deregulates NF-κB by activating Tak1 and mediating the physical association of Tak1-IKK Retroviral oncoprotein Tax induces processing of NF-κB2/p100 in T cells: evidence for the involvement of IKKα PDZ binding motif of HTLV-1 Tax promotes virus-mediated T-cell proliferation in vitro and persistence in vivo PDLIM2 suppresses human T-cell leukemia virus type I Tax-mediated tumorigenesis by targeting Tax into the nuclear matrix for proteasomal degradation HTLV-I Tax protein binds to MEKK1 to stimulate IκB kinase activity and NF-κB activation Epigenetic disruption of the WNT/ß-catenin signaling pathway in human cancers Multiple viral strategies of HTLV-1 for dysregulation of cell growth control Expression of CCR4 in adult T-cell leukemia Frequent expression of CCR4 in adult T-cell leukemia and human T-cell leukemia virus type 1-transformed T cells Perturbed human sub-networks by Fusobacterium nucleatum candidate virulence proteins Perturbed human sub-networks by Fusobacterium nucleatum candidate virulence proteins Harrod, R., Tang, Y., Nicot, C., Lu, H.S., Vassilev, A., Nakatani, Y., and Giam, C.-Z. (1998 ). An exposed KID-like domain in human T-cell lymphotropic virus type 1 Tax is responsible for the recruitment of coactivators CBP/p300. Molecular and Cellular Biology 18, 5052-5061. Harvey, M., and De Fabritiis, G. (2009) . An implementation of the smooth particle mesh Ewald method on GPU hardware. Journal of chemical theory and computation 5, 2371-2377.Hasegawa, H., Sawa, H., Lewis, M.J., Orba, Y., Sheehy, N., Yamamoto, Y., Ichinohe, T., Tsunetsugu-Yokota, Y., Katano, H., and Takahashi, H. (2006) . Thymus-derived leukemialymphoma in mice transgenic for the Tax Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., and Simmerling, C. (2006) . Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins: Structure, Function, and Bioinformatics 65, 712-725.