Recognition of high-risk HPV E6 oncoproteins by 14-3-3 proteins studied by interactomics and crystallography

In tumors induced by high-risk mucosal human papillomaviruses (hrm-HPVs), HPV E6 oncoproteins inhibit apoptotic processes and sustain cell proliferation. E6 from all hrm-HPVs harbor a C-terminal short PDZ domain-binding motif (PBM), whose phosphorylation down-regulates PDZ binding but triggers E6 binding to 14-3-3 proteins. Here we classify PBMs of E6 proteins depending on their principle ability to be phosphorylated and subsequently acquire a 14-3-3-binding motif III consensus, (pS/pT)XX-COOH. Systematic competitive fluorescence polarization measurements show that the PBMs from four selected E6 oncoproteins bind all seven human 14-3-3 isoforms with distinct, wide-ranging affinities, obeying remarkable trends assigned to 14-3-3 isoform specificity and small E6 sequence variations. We crystallized the hrm-HPV18 E6 PBM bound to 14-3-3ζ, revealing a 14-3-3-motif III complex at 1.9 Å resolution. Using fluorescence polarization and crystallography, we also demonstrate that fusicoccin, a molecule that reinforces many known 14-3-3 complexes, destabilizes the 14-3-3-E6 interaction, indicating the druggability of that complex.


Abstract 22
In tumors induced by high-risk mucosal human papillomaviruses (hrm-HPVs), HPV E6 23 oncoproteins inhibit apoptotic processes and sustain cell proliferation. E6 from all hrm-24 HPVs harbor a C-terminal short PDZ domain-binding motif (PBM), whose phosphorylation 25 down-regulates PDZ binding but triggers E6 binding to 14-3-3 proteins. Here we classify 26 PBMs of E6 proteins depending on their principle ability to be phosphorylated and 27 subsequently acquire a 14-3-3-binding motif III consensus, (pS/pT)XX-COOH. Systematic 28 competitive fluorescence polarization measurements show that the PBMs from four 29 selected E6 oncoproteins bind all seven human 14-3-3 isoforms with distinct, wide-ranging 30 affinities, obeying remarkable trends assigned to 14-3-3 isoform specificity and small E6 31 sequence variations. We crystallized the hrm-HPV18 E6 PBM bound to 14-3-3ζ, revealing 32 a 14-3-3-motif III complex at 1.9 Å resolution. Using fluorescence polarization and 33 crystallography, we also demonstrate that fusicoccin, a molecule that reinforces many 34 known 14-3-3 complexes, destabilizes the 14-3-3-E6 interaction, indicating the druggability 35 of that complex. (https://pave.niaid.nih.gov/) are classified into five phylogenetically distinct genera (alpha, 47 beta, gamma, mu and nu), of which alpha HPVs display almost exclusively mucosal 48 tropism 1 . While most HPV only generate benign proliferative events such as warts, lesions 49 or condylomas, a subset of about 20 HPV types belonging to two alpha species have been 50 associated to cancers and are therefore epidemiologically categorized as "high-risk" 2, 3, 4 . 51 HPV16 and 18 are the most prevalent, causing up to 80% of squamous cervical 52 carcinomas 5 , whereas HPV16 is also highly prevalent in HPV-positive tumors of the 53 oropharynx and the anus 6 . 54 The E6 protein is one of the two main HPV oncoproteins expressed at the early stage of 55 infection. In HPV-transformed cells, E6 participates in counteracting apoptosis, altering 56 differentiation pathways, polarity and adhesion properties, and thereby sustains cell 57 proliferation 1, 7 . Inhibition of E6 in HPV-positive cell lines results in the cell growth arrest 58 and induces apoptosis or rapid senescence 8,9,10,11 . 59 E6 has been found to interact with numerous distinct cellular target proteins involved in a 60 variety of cellular functions 12 . Importantly, E6 proteins from distinct HPV species recognize 61 distinct subsets of the full panel of potential E6 targets 13 . This likely contributes to the 62 particular biological traits of each HPV type in terms of tropism, viral cycle, or 63 pathogenicity. 64 E6 proteins of most mammalian PVs comprise two zinc-binding domains (E6N and E6C) 65 separated by a linker helix 1 (Fig. 1A), which altogether form a charged-hydrophobic pocket 66 that recognizes LxxLL motifs interspaced with acidic residues, found in a variety of host 67 cellular proteins 12 . Hrm-HPV E6 recruits simultaneously the LxxLL motif of ubiquitin ligase 68 E6-Associated Protein (E6AP) and, via a distinct interface, the central "core" domain of 69 tumor suppressor p53 14 . The resulting E6-E6AP-p53 trimer drives the ubiquitinylation and 70 subsequent proteasomal degradation of p53 15 . In HPV-transformed cells, this leads to low 71 levels of p53 protein that probably contribute to the tumorigenic phenotype. It is supposed 72 that E6 similarly stimulates the E6AP-mediated degradation of other host proteins 15,16 . 73 In addition to their LxxLL-binding pocket and p53-binding surface, hrm-HPV E6 proteins 74 harbor a C-terminal PDZ domain-binding motif (PBM) 17 (Fig. 1A , it turns several of them into candidate binding sites for 14-3-3 proteins 18, 21 . Despite 92 being encoded by seven distinct genes, human 14-3-3 proteins share highly similar 93 sequences and biochemical properties. Hence they are commonly referred to as 94 "isoforms", individually named β, γ, ε, ζ, η, σ, and τ (beta, gamma, epsilon, zeta, eta, sigma 95 and tau) 25 , not to be confused with their splice variants. 14-3-3 proteins form homo-and 96 heterodimers 26 characterized by their ability to bind phosphopeptides 27, 28 . 14-3-3 are 97 involved in the regulation of multiple cellular processes including apoptosis, cell division 98 and signal transduction 29 . Phosphorylated 14-3-3-binding sequences usually correspond 99 to internal motifs I RSX(pS/pT)X(P/G) and II RXY/FX(pS/pT)X(P/G) 28 and to the C-terminal 100 motif III (pS/pT)X 0-2 -COOH 30, 31 , where pS/pT denotes phosphorylated serine or threonine 101 and X denotes any amino acid. The regulation by 14-3-3 binding typically protects 14-3-3 102 targets from dephosphorylation and degradation, affects their activity, intracellular 103 localization, and interactions with other proteins by occlusion of the phosphorylated 104 segments 32 . Given the involvement of 14-3-3 in the development of various pathological 105 conditions associated with the deregulation of the corresponding protein-protein 106 interactions, 14-3-3/phosphotarget complexes are considered with increasing attention 33 . 107 Phosphorylated 16E6, 18E6, and 31E6 have been previously shown to bind to 14-3-3 108 proteins, albeit with different efficiencies 18,21,24 . In contrast, the binding of 109 unphosphorylated or phosphomimetic E6 PBMs to 14-3-3 proteins was not detectable 21, 24 . 110 This suggests that the phosphorylation-dependent interaction with 14-3-3 proteins is a 111 general feature of various E6 PBMs. 112 Here we classified PBMs of hrm-HPV E6 proteins depending on their principle ability to be 113 phosphorylated and subsequently acquire a 14-3-3-binding motif III consensus, (pS/pT)XX-114 COOH. We performed systematic quantitative binding assays of four phosphorylatable E6 115 PBMs with all seven human 14-3-3 isoform homodimers and solved a high-resolution 116 crystal structure of the HPV18 E6 PBM complex with 14-3-3ζ. This combination of 117 approaches has allowed us to rationalize differences in the affinities of tested PBMs and 118 fully confirm this by mutated peptides. Finally, we used binding assays and crystallography 119 to analyze the impact of natural toxin fusicoccin on the 14-3-3-E6 PBM interaction, which 120 proved that the 14-3-3-E6 complexes are nominally druggable by small molecules. 121

122
PDZ-binding motifs of E6 proteins can be classified according to their 123 phosphorylation and 14-3-3-binding propensity 124 Among all 225 HPV type E6 proteins curated in the PaVE database 125 (https://pave.niaid.nih.gov/), 31 E6 proteins from the α-genera HPV, having predominantly 126 mucosal tropism, possess a C-terminal PDZ-binding motif (PBM) that corresponds to the 127 class 1 PBM [X(S/T)X(L/V/I/C)-COOH, where X is any amino acid residue 34,35 ]. It has 128 previously been shown that the E6 PBMs are susceptible to phosphorylation by protein 129 kinases on the conserved Ser/Thr residue at the antepenultimate C-terminal position, 130 which contributes to the class 1 PBM consensus 18,21,23 . This phosphosite is preceded by 131 arginine residues in most of the HPV-E6 PBM sequences with recognizable patterns of the 132 two common basophilic kinase substrate consensus motifs of R(X/R)X(S/T) and 133 RXRXX(S/T) 36, 37 . Additional Arg residues, e.g. at position -6 are most probably less 134 involved in kinase recognition and more important for binding to other partners. For 135 instance, Arg -6 of 18E6 is important for PDZ recognition but its mutation does not affect its 136 phosphorylation efficiency by PKA, while mutation of Arg -5 and -4 blocks 18E6 PBM 137 phosphorylation 38 . Considering the basic consensus motifs, these 31 E6 PBMs can be 138 readily divided into three subgroups: Subgroups 1 and 2 having sequences prone to 139 phosphorylation by the basophilic kinases and the third, orphan subgroup with a less 140 predictable phosphorylation propensity (Fig. 1C). Accordingly, the PBM of HPV66 E6, 141 which in our classification falls into the orphan group, is normally recalcitrant to 142 phosphorylation by PKA in cellulo but becomes prone to it upon introduction of Arg 143 residues at its positions -5 and -4 38 . In addition, more than half of the identified HPV-E6 144 PBM sequences harbor candidate phosphorylation sites of DNA damage response 145 ATM/ATR kinases with a consensus of (S/T)Q 39 . 146 Therefore, the regulation of subgroup 1 and 2 E6 PBMs by phosphorylation is likely a 147 rather general phenomenon. In line with earlier observations 21, 24, 40 , the phospho-PBM 148 sequences from subgroups 1 and 2 can ideally match the C-terminal 14-3-3-binding motif 149 III 30 ( Fig. 1B and C). It is therefore likely that phosphorylation of the C-terminal PBMs from 150 HPV-E6 types belonging to subgroups 1 and 2 by different kinases triggers their 151 recognition by 14-3-3 proteins 18,21,22,24 . The domain organization (top) and three dimensional structure (

proteins with parallel binding profiles 189
Four phospho-PBMs from E6 proteins of HPV types 16, 18, 33 and 35 belonging to 190 subgroups 1 and 2 (as defined in Fig. 1C) were analyzed for their interaction with the all 191 seven human 14-3-3 homodimers. For comparison, we also measured two non-viral 192 phospho-PBMs originating from protein kinase RSK1 24 . We used a competitive 193 fluorescence polarization assay that measures the displacement of a fluorescent tracer 14-194 3-3-binding peptide in the presence of an increasing amount of a competitor molecule (all 195 binding curves are shown in Supplementary Fig. 1A). 196 All phospho-PBMs (hereinafter called p16E6, p18E6, p33E6, p35E6, RSK1_-1P, and 197 RSK1_-2P) detectably bound to 14-3-3 proteins, in sharp contrast to their 198 unphosphorylated counterparts. The interactions between E6 phospho-PBMs and 14-3-3 199 proteins spanned very wide affinity ranges, from just below 1 μM (p33E6-14-3-3γ) to 200 above 300 μM ( Fig. 2A and Supplementary Fig. 1). Such large binding affinity differences 201 are noteworthy since the four E6 PBM sequences are very similar (Fig. 1B), and all 14-3-3 202 isoforms share highly conserved phosphopeptide-binding grooves. 203 Remarkably, the six phospho-PBMs showed consistent hierarchy in their relative binding 204 preferences towards any of the seven 14-3-3 isoforms, albeit with a systematic overall shift 205 in affinity from one peptide to another. The seven 14-3-3 isoforms clustered as four groups 206 of decreasing affinity, in a conserved order from the strongest to the weakest phospho-207 PBM binder: gamma, eta, zeta/tau/beta and epsilon/sigma (γ, η, ζ/τ/β, and ε/σ) (Fig. 2B). 208 These conserved relative affinity shifts can be quantified by calculating, for two distinct 14-209 3-3 isoforms, their differences of free energy of binding (ΔΔG) towards each individual 210 phosphopeptide, then calculating the average difference (ΔΔG av ) with its standard 211 deviation (Fig. 2C). Between the strongest and the weakest binders (isoforms γ and σ, 212 respectively) the average phosphopeptide-binding energy difference is ΔΔG av = -5.1 ± 1.3 213 kJ/mol, roughly corresponding to a 11-fold K D ratio. 214 Noteworthy enough, another study 42 also measured the binding of seven human 14-3-3 215 isoforms to the completely unrelated cystic fibrosis transmembrane conductance regulator 216 (CFTR) phosphopeptide CFTR-R7, representing an internal 14-3-3-binding motif. A roughly 217 comparable hierarchy of affinities was observed, also showing a 11-fold K D ratio between 218 the strongest binder (gamma isoform, K D = 1 μM) and the weakest binder (epsilon isoform, 219 K D = 11 μM); both values being remarkably close to those obtained herein for the 220 strongest-binding E6 phospho-PBM, p33E6. Similar trends were also detected upon 221 interaction of the Leucine-Rich Repeat Kinase 2 (LRRK2) internal motif phosphopeptides 222 with 14-3-3 isoforms, when gamma and eta were the strongest and epsilon and sigma 223 were the weakest binders 43 . The K D ratio for the gamma and sigma isoforms averaged for 224 the three of LRRK2 phosphopeptides (~20) 43 is remarkably similar to the 11-fold difference 225 described here for the E6 PBM interaction with the strongest and weakest human 14-3-3 226 isoforms. Therefore, our data support the idea that specificity of 14-3-3 isoforms for protein 227 partners follows a general trend. 228 The seven 14-3-3 isoforms also showed consistent profiles in their relative binding 229 preferences towards any of the four E6 phospho-PBMs. For each 14-3-3 isoform, the four 230 phospho-PBMs systematically rank the same way from the strongest to the weakest 231 binder: p33E6, p18E6, p16E6 and p35E6 (Fig. 2C). The average 14-3-3 binding free 232 energy difference between p33E6 and p35E6 was ΔΔG av = -10.9 ± 0.7 kJ/mol, roughly 233 corresponding to a 100-fold K D ratio. four HPV-E6 phospho-PBMs from weakest (white) to strongest (red). C. Averaged ΔΔG 256 values between 14-3-3 isoforms or E6 phospho-PBM pairs, calculated based on their 257 observed order of binding affinities (from weakest to strongest). Individual K D values from 258 Supplementary Fig. 1 were first converted into ΔG values (at T=295 K; excluding cases 259 when K D > 300 μM) and average ΔΔG values (ΔΔG av ) were calculated between the 260 indicated motifs/isoforms. 261

262
Atomic structure reveals the 14-3-3ζ / 18E6 PBM interface 263 To get structural insight into the 14-3-3ζ interactions with 18E6 PBM, we designed, as 264 previously reported 44, 45 , a protein-peptide chimera, which we co-expressed with PKA and 265 purified from E. coli cells by using an engineered cleavable His 6 tag ( Supplementary Fig.  266 2A-C). The 14-3-3ζ core was modified to block phosphorylation of the semiconserved 267 Ser58 located in the dimer interface 46 and to facilitate crystallization 47 and was tethered to 268 the 18E6 phosphorylatable peptide RRRETQV-COOH. This approach was justified 269 because of the disordered nature of the E6 PBMs 48, 49 . 270 The 14-3-3ζ-18E6 PBM chimera was stoichiometrically phosphorylated by co-expressed 271 PKA, which gave a distinct downward shift on native PAGE ( Supplementary Fig. 2D), and 272 readily formed diffraction-quality crystals (all statistics is presented in Table 1). Structure 273 determination was carried out at a 1.9 Å resolution with a single 14-3-3 dimer in the 274 asymmetric unit (Supplementary Fig. 3A). The 18E6 phosphopeptide of this dimer was 275 bound in-trans in the amphipathic 14-3-3 grooves of the crystallographic symmetry 276 neighbor molecule, resulting in swapped phosphopeptides ( Supplementary Fig. 3A). Such 277 arrangement, previously observed for 14-3-3σ chimeras with other phosphopeptides 44, 50 278 provided us with high-resolution insight into the 14-3-3/18E6 PBM interaction ( Fig. 3A-C). 279 All residues of the 18E6 phosphopeptides could be traced including the conformation of the 280 C-terminal carboxyl-group. The phosphopeptide is stabilized by multiple polar interactions 281 involving those between phospho-Thr156 (position -2) and conserved residues of the basic 282 pocket in the amphipathic groove of 14-3-3 (Arg56, Arg129, Lys49 and Tyr128, 14-3-3ζ 283 numbering), the phosphopeptide backbone interactions to Asn224 and Asn173, and a 284 remarkable polar contact between the carboxyl-group of the peptide and the side chains of 285 Lys120 and one of the two alternative conformations of Ser45 of 14-3-3ζ (Fig. 3C). The 286 carboxyl-group orientation is further stabilized by in-cis H-bond with the side chain of the 287 preceding residue Gln157 (position -1) (Fig. 3C). The side chain of the C-terminal Val158 288 (position 0) of the peptide makes a hydrophobic contact with Val46 of 14-3-3ζ. A second 289 remarkable hydrophobic contact is observed between the side chain of Val176 of 14-3-3ζ 290 and the methyl-group of phospho-Thr156 (position -2), which hints at a more stable binding 291 of pThr than would be in the case of pSer. 292 Importantly, the conformation of 18E6 phosphopeptide bound to 14-3-3ζ within the chimera 293 is practically identical (RMSD = 0.17 Å upon superimposition of the six core Cα atoms of 294 the peptides) to the 14-3-3σ-bound conformation of a synthetic 16E6 phosphopeptide 295 reported very recently at a much lower resolution (Fig. 3D) 24 . In both cases, the carboxyl-296 group gives a polar contact to the Lys120 side chain and the hydrophobic side chain of the 297 last peptide residue (Val158 in 18E6 and Leu151 in 16E6) faces the side chain of Val46 of 298 14-3-3. This validates the chimeric approach 44 as a strategy to obtain high-resolution 299 crystal structures of 14-3-3/phosphopeptide complexes. Arg residues that are likely to be 300 involved in kinase recognition also contribute to 14-3-3 binding. The Arg residue occupying 301 position -4 forms an in-cis interaction with the phosphoryl group, and the Arg residue 302 occupying position -5 forms in both 16E6 and 18E6 a π-stacking interaction with Arg60 of 303 14-3-3 (Fig. 3D). Most of interface contacts of the 18E6/14-3-3ζ and 16E6/14-3-3σ 304 complexes are similar, suggesting that the two current crystal structures can also serve as 305 templates to build accurate homology models of other 14-3-3/E6 complexes and, more 306 generally, any motif III complexes with a pS/pTXX-COOH consensus. 307 However, some noteworthy differences appear in a subset of the crystallographic 308 conformers of 14-3-3/16E6 and 14-3-3/18E6 complexes. On the one hand, in 1 of the 4 309 conformers observed in the asymmetric unit of the 14-3-3σ/16E6 crystal, the side chains of 310 Arg -7 (Gln in 18E6) and Glu -3 form an additional in-cis salt bridge (Fig. 3D). On the other 311 hand, Arg -6 of 18E6 (Thr in 16E6) mediates a bipartite interaction with 14-3-3 in most of 312 the observed conformers. It simultaneously interacts with the carbonyl of Asp223 and 313 participates in a water-mediated interaction with Asn224 ( Fig. 3C and D). An overlay of the two 14-3-3 bound phosphopeptides from 16E6 (6TWZ.pdb) and 18E6 340 (this work) showing the similarity of the conformation. # denotes the C-terminus (-COOH). 341 wthe water molecule, π -π-stacking interaction. Important positions are numbered 342 according to the PBM convention. E. Averaged ΔΔG values between 14-3-3 isoforms or 343 35E6 phospho-PBM pairs, calculated based on their observed order of binding affinities 344 (from weakest to strongest). Individual K D values from Supplementary Fig. 1 were first 345 converted into ΔG values (at T=295 K; excluding cases when K D > 300 μM) and average 346 ΔΔG values (ΔΔG av ) were calculated between the indicated motifs/isoforms. 347 348 Rescue of the weakest E6-14-3-3 interaction by rational design 349 We hypothesized that the remarkable affinity differences of the four studied E6 phospho-350 PBMs (ΔΔG av = -10.9 ± 0.7 kJ/mol between the weakest and the strongest 14-3-3-binder) 351 might result from two opposed mechanisms: interface peptide-14-3-3 contacts favoring 352 complex formation versus intra-peptide contacts within the unbound peptide disfavoring it.

353
On the one hand, position -6 is an Arg in the two strongest 14-3-3-binders (18E6 and 354 33E6) versus a Thr in the weakest ones (35E6 and 16E6). Accordingly, the crystal 355 structures have shown that Arg -6 can mediate more interactions with the generic 14-3-3 356 interface (Fig. 3D). On the other hand, all E6 phospho-PBMs have a delicate charge 357 distribution, with an acidic C-terminus (that is also involved in PDZ-domain binding) and a 358 basic N-terminal segment (that is also involved in kinase recognition). These local charged 359 segments may form transient in-cis interactions within the unbound phosphopeptide, so-360 called "charge clamps" 51 , potentially unfavorable for 14-3-3 binding. We speculated that 361 Glu -1 in p35E6, the weakest 14-3-3 binder, might participate in such a charge clamp, 362 thereby disfavoring the binding. 363 To test these mechanisms, we synthesized three variants of p35E6. The first variant 364 contained a T-6R substitution, which in principle could allow a more stable bound 365 conformation, but may also stabilize charge-clamps in the free form of the motif. The 366 second variant contained an E-1A substitution, which in principle could destabilize in-cis 367 charge-clamps. A third variant contained both substitutions. All substitutions turned out to 368 reinforce the binding affinities of 35E6 without altering the apparent preferences of the 369 different 14-3-3 isoforms ( Fig. 2A and Fig. 3E). Taken individually, T-6R moderately 370 increased binding (ΔΔG av = -1.1 ± 0.5 kJ/mol, 1.5-fold K D ratio), while E-1A strongly 371 reinforced it (ΔΔG av = -5.1 ± 0.2 kJ/mol, 11-fold K D ratio). When combined, the two 372 substitutions synergistically increased binding (ΔΔG av = -8.7 ± 0.4 kJ/mol, 35-fold K D ratio), 373 thereby turning p35E6 from the weakest 14-3-3 binder into the second strongest one, just 374 below p33E6. 375 376

Fusicoccin partially destabilizes the 14-3-3/E6 PBM interaction 377
Fusicoccin (FSC) is commonly used as a stabilizer of 14-3-3 complexes, when its binding 378 in the distinct pocket in the 14-3-3/phosphopeptide interface is allowed by phosphopeptide 379 side chains of the amino acids in downstream positions relative to the phospho-residue 42, 380 52, 53, 54 . Binding of FSC is known to stabilize the 14-3-3 interaction with few internal motif I 381 phosphopeptides (e.g., Gab2 peptide, 5EXA.pdb 55 and CFTR peptide, 5D3F.pdb 42 ) and 382 numerous peptides with the short motif III consensus, pS/pTX-COOH 31, 56, 57 . Most of the 383 internal motif phosphopeptides, when bound to 14-3-3, apparently leave insufficient space 384 for FSC and therefore hinder formation of the ternary complexes with FSC. Many motif III 385 peptides with a consensus pS/pTX-COOH enable synergistic binding of the 386 phosphopeptide and FSC, reinforcing the 14-3-3/phosphopeptide complex 31,52,53,57,58,59 . 387 At the same time, very little is known about the effect of FSC on binding of the longer motif 388 III peptides to 14-3-3, such as in the case of E6 PBMs. 389 We investigated whether binding of phosphorylated E6 PBM to 14-3-3 might be affected by 390 this small molecule and analyzed its effect on the affinity of the E6 PBMs to 14-3-3. We 391 used FP experiments to measure equilibrium binding affinity constants of complexes 392 between the four HPV-E6 phosphopeptides and 14-3-3 isoforms ζ and γ, in the presence of 393 100 µM FSC (Supplementary Fig. 1B and Fig. 4A). The addition of FSC consistently 394 decreased by 1.5 to 2 fold the affinities of all interactions (ΔΔG av = -1.3 ± 0.5 and -1.8 ± 0.4 395 kJ/mol for ζ and γ, respectively) without altering the apparent preferences of the different 396 peptides ( Fig. 4A and B). Therefore, fusicoccin moderately destabilized all studied 14-3-397 3/E6 interactions, in contrast to its much broadly reported, stabilizing effect on 14-3-3 398 complexes 31, 42, 52 . Nevertheless, the binding of E6 peptides and FSC was not mutually 399 exclusive. experiments are presented. The binding curves are shown in Supplementary Fig. 1. B.

424
Averaged ΔΔG values between 14-3-3-E6 phospho-PBM pairs in the absence or in the 425 presence of FSC, calculated based on their observed order of binding affinities (from 426 weakest to strongest). Individual K D values from Supplementary Fig. 1 were first converted 427 into ΔG values (at T=295 K; excluding cases when K D > 300 μM) and average ΔΔG values 428 (ΔΔG av ) were calculated between the indicated motifs/isoforms. C. An overall view on the 429 ternary complex between 14-3-3ζ (subunits are shown by surface using two tints of grey), 430 18E6 phosphopeptide (cyan sticks) and FSC (pink sticks). FSC was soaked into the 14-3-431 3ζ-18E6 chimera crystals. To get structural insights into the peculiar destabilizing effect of FSC on the 14-3-3/E6 PBM 447 interaction, we soaked the 14-3-3ζ/18E6 PBM chimera crystals with FSC and solved the 448 crystal structure of the 14-3-3ζ/18E6 PBM/FSC ternary complex (Fig. 4C-F and Table 1).

449
FSC soaking did not disrupt the crystal lattice of the 14-3-3ζ/18E6 PBM chimera crystals 450 ( Supplementary Fig. 3) and the overall assembly ( Fig. 4C and D); however, significant local 451 structural rearrangements could be observed. FSC (Fig. 4E) binding results in a ~4 Å 452 closure of the last α-helix of 14-3-3ζ (Fig. 4F), similar to what has already been described 453 for other 14-3-3 complexes containing FSC 58 . Crystal lattice obtained by FSC soaking 454 preserved the characteristic phosphopeptide swap stabilizing the contacts with the 455 neighboring chimera molecules, and FSC was bound in each 14-3-3ζ subunit, side-to-side 456 with the C-terminal end of the 18E6 PBM ( Fig. 4C and D, Supplementary Fig. 3). All 457 residues and atoms within the analyzed area could be observed in the electron density 458 thanks to the high resolution (Table 1). 459

FSC occupies its well-defined cavity where it is positioned by hydrophobic interactions with 460
Phe117, Ile166, Ile217 and Leu216, polar contacts with residues Asn42 and Asp213, and a 461 remarkable H-bond involving its 3-metoxy oxygen and the side chain of Lys120 of 14-3-3ζ 462 (Fig. 4D). The latter contact breaks the Lys120 interaction with the carboxyl-group of the 463 18E6 PBM formed in the absence of FSC, displacing the carboxyl to another position, 464 where it establishes a new contact with the 12-hydroxy group of FSC ( Fig. 4D and E). 14-3-465 3ζ Lys49 also switches its position, and loses a contact to the backbone carbonyl of 466 pThr156 (Fig. 3C) to establish instead a contact with the 12-hydroxy group of FSC. The 467 side chain of the C-terminal Val158 of the 18E6 PBM shifts 3.5 Å towards the phosphate 468 moiety of Thr156, breaking the hydrophobic contact with Val46 of 14-3-3ζ and significantly 469 dispersing the local electron density (Fig. 4F). As a result, while most of the peptide 470 conformation remained unchanged, the B-factors of the last 18E6 PBM residue in the 471 refined FSC-bound structure increased significantly (Fig. 4F). Therefore, FSC binding 472 causes an apparent strain in the conformation of the C-terminus of 18E6 PBM (Figs 4 and  473  5). In addition, the presence of FSC significantly reshuffles the water network that 474 surrounds the E6 carboxy-terminal extremity, which directly contacts the bound FSC (Fig.  475  5). The simultaneous binding of the 18E6 PBM and FSC in the amphipathic groove of 14-3-476 3 and the mild destabilizing influence of FSC indicate that this ternary complex can be used 477 as a starting point to design both stabilizers and inhibitors of 14-3-3/E6 interactions. shown in light grey, 18E6 residues are in cyan, phospho-group of Thr156 is shown by 499 orange sticks. FSC is shown by thin magenta sticks, water molecules affected by FSC 500 binding are shown by lime green, those similar in two structures are black. The C-terminal 501 18E6 residue (V158) is denoted by #. Note the significant redistribution of water molecules 502 upon FSC binding. 503

505
E6 oncoproteins of all high-risk alpha HPV types contain a conserved C-terminal PDZ-506 Binding Motif which can turn into a potential 14-3-3-binding motif upon phosphorylation of 507 the conserved Thr/Ser residue at position -2 (Fig. 1). We demonstrated here that the 508 phosphorylated forms of four selected alpha HPV-E6 PBMs detectably bound to all seven 509 human 14-3-3 isoforms. Unexpectedly, the binding affinities of these 28 distinct 14-3-3/E6 510 complexes showed large variations, spanning a 10-fold K D range for different 14-3-3 511 isoforms binding to a given PBM, and a 100-fold K D range for different E6 phospho-PBMs 512 binding to a given 14-3-3 isoform. Furthermore, affinity variations followed trends that were 513 all but random. The 14-3-3 binding profiles of all E6 phosphopeptides turned out to be 514 remarkably parallel, and so were the E6-binding profiles of all 14-3-3 isoforms ( Fig. 2B and  515 C). 516 When considering the present data together with previously published literature, 14-3-3 517 interactions with phosphorylated motifs from different origins have a strikingly wide affinity 518 range, spanning from low nanomolar to low millimolar detectable dissociation constants 519 (Fig. 6). Despite variable interaction modes (e.g. monovalent vs. divalent), the 14-3-3 520 family-wide affinity trends are surprisingly parallel, with 14-3-3γ and 14-3-3η consistently 521 being the strongest binders and 14-3-3σ and 14-3-3ε being the weakest binders. 522 Independently of the nature of the target motif, the average maximal K D ratio between the 523 strongest-binding and the weakest-binding 14-3-3 is around 10-fold in our present work 524 and also in the literature (Fig. 6), Conversely, phosphopeptides, even very similar in 525 sequence, can sample much wider ranges of binding affinities for a given 14-3-3 isoform. 526 For instance, for 14-3-3γ, the K D ratio between the strongest and the weakest binding 527 phosphopeptide is almost 625-fold in the present work, and 39,000-fold when taking into 528 account other reports 42, 43, 60, 61 (Fig. 6). However, the discussed weak interactions do not 529 necessarily lead to the absence of complex formation in cellulo, as 14-3-3 proteins are 530 among the most abundant proteins in human cells 62, 63 . Indeed, according to Protein 531 Abundance Database 63 , 14-3-3ε is the 48 th most abundant human protein (2479 ppm) and 532 14-3-3ζ is the 72 nd (1680 ppm), whereas all seven 14-3-3 isoforms are within the 820 most 533 abundant proteins (i.e., the top 4.1%) out of 19949 proteins in the integrated whole human 534 body dataset. Given such high abundance, strong 14-3-3 interactions might lead to 535 extremely tight complex formation, while weak interactions might lead to transient, more 536 dynamic 14-3-3 complexes. 537 The observed strong and parallel affinity variations are noteworthy if one considers the high 538 degree of conservation of the amphipathic grooves of the seven 14-3-3 isoforms and of the 539 amphipathic groove-binding residues of the four E6 PBMs. Indeed, affinity differences may 540 stem not only from structural particularities of the bound complexes but also from intrinsic 541 properties of the unbound partners. Conformations unfavorable for complex formation may 542 exist in various proportions for each 14-3-3 isoform and each E6 phosphopeptide, thereby 543 influencing at various degrees the formation of the complexes and their resulting binding 544 affinities. On the one hand, full-length 14-3-3 proteins feature the flexible C-terminal tails 545 that are the most variable elements among isoforms; at least for some isoforms they have 546 been shown to sample conformations also occupying the amphipathic 14-3-3 grooves and 547 thereby regulating their affinities for target peptides 64,65 . Conformational dynamics of the 548 14-3-3 dimers along the open-closed state trajectory 26, 44 may also differ among 14-3-3 549 isoforms, which can probably affect the effectiveness of adopting the peptide-bound 550 conformation. On the other hand, it has been shown that a phosphoryl-group in a 551 disordered segment can form local "charge clamps" in-cis with a neighboring Arg/Lys 552 residue, thereby decreasing its apparent availability for binding partners 51 . The four E6 553 PBMs studied herein display distinctive charged, polar or non-polar residues at several 554 positions (-1; -6; -7; -8; -9) that may influence their explorable conformational spaces. By 555 replacing the Glu -1 by Ala in the weakest 14-3-3 binder, p35E6, we obtained a significantly 556 stronger 14-3-3 binder, suggesting that we successfully disrupted a charge clamp that is 557 unfavorable to binding of the p35E6 motif. Nonetheless, fine differences of contact 558 networks at the interfaces of E6/14-3-3 complexes may also influence affinity. In this line, 559 comparison of the present p18E6/14-3-3ζ structure to the previously solved p16E6/14-3-3σ 560 structure 24 suggested that an Arg residue at position -6 established favorable interface 561 contacts ( Fig. 3C and D). Accordingly, replacing Thr -6 of p35E6 with Arg -6 as found in 562 p18E6 and p33E6, effectively reinforced the binding event. Finally, combining both 563 Glu→Ala and Thr→Arg substitutions at positions -1 and -6, respectively, resulted in a 564 p35E6 variant that bound 14-3-3 proteins almost as strongly as the strongest studied E6 565 binder, p33E6 ( Fig. 2A and Fig. 3E). Interestingly, the RSK1 phospho-PBMs 566 (RRVRKLPSTpTL-COOH and RRVRKLPSpTTL-COOH), featuring non-charged residues 567 Thr and Leu in the downstream position relative to the phosphothreonine, and therefore 568 being unable to form the equivalent charge-clamp that is suggested in the case of p35E6 569 (SKPTRREpTEV-COOH), show remarkably high affinities to all 14-3-3 isoforms per se 570 ( Fig. 2A and Supplementary Fig. 1). Our data prove that charge clamps can indeed 571 strongly contribute to the binding preferences of different phosphopeptide motifs. This 572 principle needs to be further examined for all kinds of phosphorylation-regulated 573 interactions as charge-clamp formation might be an often ignored biochemical property of 574 practically all phosphorylation sites. to the strongest partner-binder 14-3-3γ, all studied peptides are following very similar affinity trends 586 between the different 14-3-3 isoforms. Between the strongest and the weakest partner-binding 14-587 3-3 isoform, a 12-fold dissociation constant decrease was found on average. C. When normalizing 588 to the generally strongest studied 14-3-3-binding motif, the isoform-specific affinity variations for 589 each peptide are compensated to clearly show the average differences between different binding 590 motifs. Between the strongest and the weakest analyzed 14-3-3-binding motif, a 34,000-fold K D 591 decrease was found on average for all 14-3-3 isoforms. Only counting the very similar HPV-E6 592 peptides sequence-wise, this affinity difference can still be larger than 100-fold. Therefore, the 593 motif-to-motif affinity differences are expected to be much higher than the 14-3-3 isoform-to-isoform 594 affinity differences. Binding motifs that are analyzed in other studies are highlighted with a grey 595 background 42,43,60,61 . The color scale is either based on affinity values or in K D ratios. √ denotes 596 affinities weaker than the limit of quantitation of the fluorescence polarization assay.

597
When phosphorylated, all the E6 PBMs studied here ideally match motif III consensus 599 recognized by 14-3-3 with two C-terminal residues downstream of the phosphorylated 600 threonine (pS/pTXX-COOH) (Fig. 1B) proteins commonly use their elements, efficiently mimicking the host 14-3-3-binding 614 peptides, to hijack the cellular functions controlled by 14-3-3 proteins. One of the possible 615 mechanisms is the 14-3-3-mediated stabilization of viral proteins to evade 616 dephosphorylation and degradation, which may prolong the half-life and increase chances 617 for successful replication and further infections 21, 68 . 618 The phospho-PBM conformation revealed by our crystal structures, ideally congruous to 619 the amphipathic groove of 14-3-3, leaves vacant the cavity that is known to be druggable 620 and occupied by a fungal toxin called fusicoccin (FSC) 33,54,56 . Benefitting from the well-621 diffracting crystals of the 14-3-3ζ-18E6 PBM chimera, we proceeded with their soaking with 622 FSC, which resulted in a 1.85-Å crystal structure of the ternary 14-3-3ζ/18E6 PBM/FSC 623 complex. The direct comparison of the 14-3-3ζ/18E6 complex in the absence and in the 624 presence of FSC revealed a significant conformational change in the last α-helix of 14-3-625 3ζ, a rearrangement of bound water molecules and a remarkable strain in the C-terminal 626 part of the 18E6 PBM peptide (Figs 4 and 5). This strain forced the translocation of the 627 carboxyl-group and increased the displacement factors for the last C-terminal residue, 628 suggesting destabilization of the 14-3-3ζ/18E6 PBM complex by FSC. This partial 629 destabilization, and thus the less-documented inhibitory action of FSC, was confirmed by in 630 vitro binding assays using fluorescence polarization ( Fig. 4A and B). Noteworthily, the 631 structural data on the effect of FSC on the C-terminal end of the E6 PBM is very consistent 632 with the preservation of the affinity differences for various 14-3-3/phospho-PBMs 633 complexes in the presence of FSC (Fig. 4A). This supports the notion that those affinity 634 differences are dictated by positions beyond the last C-terminal residue, including those 635 involved in formation of charge clamps. 636 We could find only one earlier reported example when FSC decreased the binding affinity 637 of a 14-3-3 to a motif III peptide (the interleukin 9 receptor alpha chain (IL-R9α) peptide, 638 RSWpTF-COOH 31 ), but in that case no experimental structural information was available. 639 In addition, it was reported that binding of the shortest motif III peptide from the cyclin-640 dependent kinase inhibitor (p27 Kip1 ) RRQpT-COOH to 14-3-3 is not affected by FSC, most 641 likely due to the absence of direct contacts between FSC and the phosphopeptide, which 642 both bind in the amphipathic groove of 14-3-3 independently 31 . The structural basis for the 643 inhibitory action of FSC has very recently been reported for several internal motif I 14-3-3-644 binding peptides 58 . Therefore, complementing the range of reports on the stabilizing effect 645 of FSC 31,42,54,55,56 , our results provide the first structural evidence that FSC can be a 646 negative regulator of 14-3-3 interactions with typical motif III peptides. 647 Thus, our structural and in vitro binding data with FSC confirm the druggability of the 14-3-648 3-E6 interaction and suggest that appropriate modification and optimization of the small 649 molecule may provide promising opportunities for selective modulation of viral protein-14-3-650 3 interactions in the future. 651