14-3-3-protein regulates Nedd4-2 by modulating interactions between HECT and WW domains

Neural precursor cell expressed developmentally down-regulated 4 ligase (Nedd4-2) is an E3 ubiquitin ligase that targets proteins for ubiquitination and endocytosis, thereby regulating numerous ion channels, membrane receptors and tumor suppressors. Nedd4-2 activity is regulated by autoinhibition, calcium binding, oxidative stress, substrate binding, phosphorylation and 14-3-3 protein binding. However, the structural basis of 14-3-3-mediated Nedd4-2 regulation remains poorly understood. Here, we combined several techniques of integrative structural biology to characterize Nedd4-2 and its complex with 14-3-3. We demonstrate that phosphorylated Ser342 and Ser448 are the key residues that facilitate 14-3-3 protein binding to Nedd4-2 and that 14-3-3 protein binding induces a structural rearrangement of Nedd4-2 by inhibiting interactions between its structured domains. Overall, our findings provide the structural glimpse into the 14-3-3-mediated Nedd4-2 regulation and highlight the potential of the Nedd4-2:14-3-3 complex as a pharmacological target for Nedd4-2-associated diseases such as hypertension, epilepsy, kidney disease and cancer. Pohl et al. investigated the structural basis of Nedd4-2 regulation by 14-3-3 and found that phosphorylated Ser342 and Ser448 are the main residues that facilitate 14-3-3 binding to Nedd4-2. The authors propose that the Nedd4-2:14-3-3 complex then stimulates a structural rearrangement of Nedd4-2 through inhibiting interaction of its structured domains.

T he neural precursor cell expressed developmentally downregulated 4 (Nedd4-2) is a member of the HECT E3 ubiquitin ligase family. As such, this enzyme targets proteins for ubiquitination in mammalian programmed cell death 1,2 . Mouse knockout studies have confirmed that Nedd4-2 plays a key role in animal physiology by regulating multiple substrates, including the epithelial sodium channel (ENaC). High ENaC activity and blood pressure with aberrant renal Na + reabsorption are observed in Liddle syndrome resulting from mutations in its Nedd4-2-interacting motif 3,4 . In addition to regulating ion transport, Nedd4-2 controls cellular trafficking in different tissues, modulating multiple signaling pathways through these interactions. Unsurprisingly, respiratory distress, hypertension, and electrolyte imbalance and kidney disease stand out among the pathological consequences of Nedd4-2 dysregulation, in line with mouse studies associating numerous SNPs in the Nedd4-2 gene with these conditions and with multiple tumor types 5,6 . Therefore, potential therapeutic interventions may be developed by targeting the ubiquitin system for drug development via Nedd4-2 interactions 7 .
Developing such strategies targeting Nedd4-2 requires thoroughly understanding the structural-functional relationships of this protein. Fortunately, all nine members of the Nedd4 family of mammalian HECT E3 ligases have a similar modular multidomain architecture, typically consisting of an N-terminal C2 domain, two-to-four WW domains, which contain two conserved tryptophan residues and a proline residue, and a C-terminal catalytic HECT domain 5,8 (see Fig. 1). The N-terminal C2 domain enables Ca 2+ -dependent binding to membrane phospholipids, whereas the WW domains specifically bind short protein motifs, either PPXY or LPXY (where X indicates any amino acid), as well as proline-rich motifs of substrate proteins 9 . Through its four WW domains, Nedd4-2 may, nevertheless, interact with many different proteins, and even with several proteins simultaneously, because these domains show different substrate specificities, suggesting distinctive roles 10 . HECT domain is a bilobed domain whose N-terminal N-lobe interacts with E2 enzymes and whose C-terminal C-lobe contains the catalytic cysteine (Cys 942 in Nedd4-2). The C-lobe can freely move around the flexible joint loop connecting this lobe to the Nlobe in the L-shaped structure 11,12 . Furthermore, inter-or intramolecular interactions between WW domains and the PY motif (L 948 PPY 951 ) located within the HECT domain likely inhibit Nedd4-2 auto-ubiquitination, thus increasing its stability 13 . Accordingly, this interaction is disrupted by substrate binding, promoting Nedd4-2 self-ubiquitination and subsequent degradation. As a result, Nedd4-2 is downregulated upon target ubiquitination.
To examine the role of the HECT domain in the overall stability of the Nedd4-2:14-3-3η complex, we also prepared a shorter variant Nedd4-2 335-455 missing both the C2 and the HECT domains. This construct was considerably more soluble and stable than Nedd4-2 186-975 . In the SV-AUC analysis of phosphorylated Nedd4-2 335-455 (pNedd4-2 335-455 ) alone, the single peak with a s w(20,w) of 1.4 (f/f 0 = 1.5) corresponds to a M w of ∼14 kDa (theoretical M w 14 kDa). Accordingly, the absence of HECT domain should not affect Nedd4-2 oligomerization ( Supplementary Fig. S2a). In turn, the analysis of pNedd4-2 335-455 and 14-3-3η mixtures showed the formation of a complex with a s w(20,w) of 4.5 S (f/f 0 = 1.4), which corresponds to a M w of ∼70 kDa and matches a molar stoichiometry of 1:2 (a protomer of pNedd4-2 335-455 bound to a 14-3-3η dimer, theoretical M w 70 kDa) ( Supplementary Fig. S2a). Both direct modeling of SV-AUC data ( Supplementary Fig. S3a) and analysis of the s w isotherm as a function of pNedd4-2 335-455 concentration revealed an apparent K D lower than 30 nM when using a binding model assuming a reversible interaction between the 14-3-3η dimer and one molecule of pNedd4-2 335-455 (Fig. 3b). These results indicate that the HECT domain is not essential for highaffinity Nedd4-2 binding to 14-3-3η.
High-affinity Nedd4-2 binding to 14-3-3η requires both Ser 342 and Ser 448 phosphorylation. All three putative 14-3-3 binding motifs of Nedd4-2 contain arginine residues at -5 and -3 positions with the respect to the phosphoresidue and a serine residue at -2 position. However, only the motif with Ser 448 has a proline residue at +2 position, thus resembling the canonical 14-3-3 binding motif I (RSXpSXP). Considering their differences in primary structure, these three motifs likely differ in their binding affinities to 14-3-3 25,28,29 . To characterize the importance of these phosphorylated motifs for Nedd4-2 binding to 14-3-3η, we prepared three Nedd4-2 335-455 mutants containing a single phosphorylation site (denoted as pS342, pT367 and pS448) by mutating the other phosphorylation sites to alanine and three Nedd4-2 335-455 mutants with combinations of two phosphorylation sites (denoted as pS342 + pT367, pS342 + pS448 and pT367 + pS448), in addition to a variant with all sites mutated to alanine (denoted as no pS), which was used as a negative control.
14-3-3 binding motifs of Nedd4-2 show distinct 14-3-3 isoform binding specificity. The interactions between isolated 14-3-3 binding motifs of Nedd4-2 and 14-3-3 protein isoforms were further characterized by fluorescence polarization (FP). For this purpose, we synthesized three decapeptides of known 14-3-3 binding motifs with phosphorylated pSer 342 , pThr 367 and pSer 448 (denoted as pep-S342, pep-T367 and pep-S448) bordered by the amino acids from the natural sequence of Nedd4-2, with four amino acids before and another five after the corresponding phosphoresidue. We labelled all peptides N-terminally with fluorescein isothiocyanate (FITC) and characterized their binding to all human 14-3-3 protein isoforms by FP ( Fig. 4 and Table 1). Our FP data revealed that all three phosphopeptides have relatively low binding affinities. The pep-S342 showed a measurable binding affinity only to 14-3-3σ (K D of ∼7.8 μM). In contrast, we were able to detect pep-T367 binding, albeit very weak, to all 14-3-3 isoforms except ε and σ. Lastly, we determined that pep-S448 binds to all 14-3-3 isoforms in the low micromolar range, showing the lowest K D values (∼6.1 and 7.2 μM) when bound to isoforms η and γ, respectively. In addition, the K D values of synthetic phosphorylated peptides are two orders of magnitude higher than those of recombinant Nedd4-2 variants, indicating simultaneous binding of two motifs and/or the participation of regions other than phosphorylated motifs in the formation of the complex (Figs. 2 and 3). Overall, these results support our hypothesis that the pSer 448 motif is the dominant 14-3-3 binding motif.
The 14-3-3 dimer simultaneously anchors both pSer 342 and pSer 448 motifs of Nedd4-2. Because our AUC analysis suggested that the pSer 342 and pSer 448 motifs are responsible for highaffinity Nedd4-2 binding to 14-3-3, we investigated the structural basis of their interaction with 14-3-3 in detail. For this purpose, we crystallized the peptides containing the pSer 342 and pSer 448 motifs (pep-S342 and pep-S448) bound to 14-3-3γΔC (lacking the 15 flexible residues at the C-terminus). The 14-3-3γ isoform was selected based on crystal quality. Both phosphopeptide complexes crystallized in the trigonal space group R3 with two 14-3-3 dimers with bound phosphopeptides in the asymmetric unit (PDB ID: 6ZBT and 6ZC9, Table 2 and Fig. 5). The structures of the pep-S342:14-3-3γΔC and pep-S448:14-3-3γΔC complexes were solved by molecular replacement with 14-3-3γ (PDB ID: 2B05) as a search model and refined to resolutions of 1.8 Å and 1.9 Å, respectively. The final electron densities of both peptides allowed us to trace seven of ten amino acids (Leu 338 -Thr 344 ) of pep-S342 and (Pro 444 -Pro 450 ) pep-S448 (Fig. 5a, c). In both cases, we were unable to trace the last three residues of each peptide, presumably due to disorder. The phosphate group, the main-chain conformation and other contacts in the 14-3-3binding groove of both motifs were recognized similarly to those previously observed in other 14-3-3 protein complexes 25,29,31-34 . The pSer 342 moiety of pep-S342 is coordinated through direct contacts enabled by side chains of the 14-3-3 residues Arg 57 , Arg 132 , Lys 50 and Tyr 133 and by water-mediated contacts with 14-3-3 Asp 129 and Asn 178 (Fig. 5b). Other contacts include hydrogen bonds between the main-chain atoms of the Nedd4-2 residues Cys 341 and Val 343 and the side chains of the 14-3-3 residues Asn 229 and Asn 178 , respectively. In addition, the side chain of Nedd4-2 Ser 340 makes a polar contact with the sidechain residues Trp 233 and Glu 185 of 14-3-3. Similar contacts between the phosphopeptide and 14-3-3 were also observed in the 14-3-3γΔC:pep-S448 complex (Fig. 5d). The main difference is the absence of water-mediated contacts between Arg 445 (-5 residue with the respect to pSer 448 ) and the side chains of the 14-3-3 residues Arg 57 , Arg 61 and Glu 136 .
Because the 14-3-3 binding motifs pSer 342 and pSer 448 border the WW2 domain ( Fig. 1), we assessed whether this domain Binding of FITC-labeled phosphopeptides with 14-3-3 binding motifs of Nedd4-2 (pSer 342 , pThr 367 or pSer 448 ) to all human 14-3-3 isoforms was characterized by FP titrations. The binding affinities of these peptides were determined by fitting FP data to a one-site binding model. The corresponding sequences of these peptides are shown at the top. Background polarization was subtracted from all values. Results are expressed as mean ± SD (n = 3). Error bars are not visible if they are smaller than the size of the symbol. Table 1 14-3-3 isoform binding specificity of individual 14-3-3 binding motifs of Nedd4-2.
participates in 14-3-3 binding by crystallizing the Nedd4-2 335-455 construct in a complex with 14-3-3ηΔC. The phosphorylation site Thr 367 was mutated to Ala to prepare Nedd4-2 335-455 phosphorylated only at Ser 342 and Ser 448 for high-affinity binding and for homogeneous crystallization. The 14-3-3 ηΔC:Nedd4-2 335-455 complex crystallized in the monoclinic space group C121, with one Nedd4-2 molecule bound to the 14-3-3ηΔC dimer in the asymmetric unit. The structure was solved by molecular replacement with 14-3-3η (PDB ID: 2C63) as a search model and refined to a resolution of 2.3 Å. Interpretable electron density was found for 21 residues, nine around the pSer 342 site and twelve around second pSer 448 site (Leu 338 -Ala 346 and Pro 444 -Ser 455 ) (PDB ID code 7NMZ, Fig. 5e, Table 2). Although the structure revealed simultaneous binding to both phosphorylated motifs by a 14-3-3η dimer, no electron density was found for the 97 residues between the phosphorylated motifs (Val 347 -Arg 443 ), thus indicating that this region, including the WW2 domain, remains flexible when bound to 14-3-3 (Fig. 5e). The recognition of the phosphate group and contacts with the 14-3-3 ligand binding grooves are similar to those observed in the structures of the individual motifs (Fig. 5b, d). Other contacts were not detected in the crystal structures of short phosphopeptides, including the 14-3-3 residue Asn 43 from the α3 helix of both protomers, which interacts with the main chain of the Nedd4-2 residues Ala 346 , Val 452 , Thr 453 and Ser 455 , and the polar contacts between the 14-3-3 residue Glu 15 from the α1 helix and the Nedd4-2 residues Thr 453 and Ser 455 . A similar contacts have been previously observed in the structures of the Nth1:14-3-3 and caspase-2:14-3-3 complexes (PDB ID: 5N6N 32 and 6SAD 33 , respectively).
Crosslinking Nedd4-2 186-975 :14-3-3η mixtures (mixed in 1:2 stoichiometry) using disuccinimidyl suberate (DSS) and DSG yielded eight intermolecular crosslinks (Supplementary  Table S4, Fig, 6c). Most of these crosslinks connect the α-helices that form the 14-3-3η ligand binding groove (α3, α5 and α9) with the WW3 domain and the HECT domain Crystal structures of Nedd4-2 peptides containing the 14-3-3 binding motifs pSer342 and pSer448 bound to 14-3-3γ and 14-3-3η. a Detailed view of the crystal structure of the 14-3-3γΔC:pep-S342 complex. The 2F o -F c electron density map is contoured at 1σ. b Polar contacts (black lines) between the residues of 14-3-3γ (black) and the pSer 342 binding site of Nedd4-2 (red). Water molecules are shown as yellow spheres. c Detailed view of the crystal structure of the 14-3-3γΔC:pep-S448 complex. The 2F o -F c electron density map is contoured at 1σ. d Polar contacts (black lines) between the residues of 14-3-3γ (black) and the pSer 448 binding site of Nedd4-2 (red). Water molecules are shown as yellow spheres. e Crystal structure of Nedd4-2 335-455 containing two phosphorylation sites (pSer 342 and pSer 448 ) bound to the 14-3-3ηΔC dimer: Top view of the intervening linker sequence of the two 14-3-3-binding motifs of Nedd4-2 and of the polar contacts (black dashed lines). The final 2F o -F c electron density map is contoured at 0.8σ (blue mesh). Nedd4-2 residues are labeled in red, and 14-3-3 residues are labeled in black. Water molecules are shown as cyan spheres. Structure figures were generated using PYMOL.
Scattering data from two regions of the elution peak of Nedd4-2 186-975 alone were analyzed (Supplementary Fig. S4a). Based on our Guinier analysis, the data from the top of the peak of Nedd4-2 186-975 revealed the presence of aggregates, whereas the data from the right shoulder of the elution peak showed homogenous particles with consistent R g values of ∼4.4 nm and an estimated M w of ∼101 kDa, in line with the theoretical M w of Nedd4-2 186-975 (92 kDa; Supplementary Fig. S4b and Table 3). In turn, scattering data from the top of the elution peak of 14-3-3η indicated homogenous particles with an estimated M w of 56 kDa, matching the theoretical M w of 14-3-3η dimer (57 kDa) ( Supplementary Fig. S4c, d and Table 3). The Nedd4-2 186-975 :14-3-3η complex was prepared with the 1:2 stoichiometry at a concentration 60 μM, which is approximately three orders of the magnitude higher than the K D value derived from SV-AUC (∼50 nM, Fig. 2). The complex eluted as a single peak and frames from two regions of the elution profile were analyzed (Supplementary Fig. S4e). As in Nedd4-2 186-975 , the Guinier plot of the scattering data from the top of the elution peak indicated the presence of aggregates, but the scattering data from the right side of the peak showed a R g of ∼4.8 nm and a M w of ∼157 kDa, corroborating the theoretical M w of the Nedd4-2 186-975 :14-3-3η complex with a 1:2 stoichiometry (149 kDa) ( Supplementary  Fig. S4f).
After calculating distance distribution functions P(r), we found similar values of maximal distance within the particle (D max ) of  Table S1) and quantitative crosslinks with a 50-fold molar excess of 12 C and 13 C DSA in pale green (Supplementary Table S3). b Schematic representation of intramolecular crosslinks of Nedd4-2 186-975 in a complex with 14-3-3η (black) with a 50-fold molar excess of DSG (Supplementary Table S2) and quantitative crosslinks with a 50-fold molar excess of 12 C and 13 C DSA in pale green (Supplementary Table S3). c Schematic representation of intermolecular crosslinks between Nedd4-2 186-975 and 14-3-3η mixed at a 1:2 stoichiometry with a 50-fold molar excess of DSG or DSS (black) crosslinking agent (Supplementary Table S4). Nedd4-2 186-975 diagram: the C2 domain, not present in the Nedd4-2 186-975 construct, is shown in grey, the WW domains (denoted as W1-4) are shown in yellow, teal, blue and magenta, and the HECT domain N-and C-lobes are shown in raspberry and salmon, respectively. The beginning and the end of the Nedd4-2 186-975 construct are labeled in red. 14-3-3 diagram: colored regions represent α-helices of the 14-3-3 molecule (shown in green), and the helices that form the ligand binding groove (α3, α5, α7 and α9) are colored in dark green. This figure was prepared using the xiVIEW (https://xiview.org/) and InkScape (http://www.inkscape.org/) programs.
Nedd4-2 186-975 and its complex with 14-3-3η ( Fig. 7a and Table 3). Concurrently, the dimensionless Kratky plot of scattering data ((sR g ) 2 I(s)/I 0 versus sR g ) suggested their conformational flexibility, as indicated by the bell-shaped profiles of the complex and Nedd4-2 186-975 alone with maxima of 1.2 at sR g ∼ 2.1 and 1.4 at sR g ∼ 2.2, respectively (Fig. 7b, green and red trace) because the scattering data on a compact globular particle such as 14-3-3η (Fig. 7b, blue trace) peaks at 1.104 at sR g value of ∼1.73. Nevertheless, the complex exhibits a lower conformational flexibility than Nedd4-2 186-975 alone.
By combining rigid body modeling of SAXS profiles with distance constraints assessed by crosslinking coupled to MS (Supplementary Tables S1-S4), we calculated models of Nedd4-2 186-975 alone and in a complex with 14-3-3η. The best-scoring CORAL model of Nedd4-2 186-975 alone fitted the experimental SAXS data with a χ 2 of 1.26 ( Supplementary Fig. S5) and matched all intramolecular crosslinks (Supplementary Table S1 and S3). In this model, the WW2 and WW3 domains (in teal and blue, respectively) are positioned close to the HECT domain, WW2 is located between the N-and C-lobes and WW3 interacts with the C-lobe of the HECT domain (Fig. 7c). In turn, WW1 and WW4 are distant from the catalytic domain ( Supplementary Fig. S5). This arrangement corroborates a previous report by Grimsey at el. 35 . According to these authors, in the closed, autoinhibited form of Nedd4-2, the HECT domain interacts with the region preceding the WW3 domain.
The Nedd4-2 186-975 :14-3-3η complex was first modeled as a 14-3-3η dimer attached to Nedd4-2 via pSer 342 and pSer 448containing motifs and assuming that the whole HECT domain is a rigid body. However, the models derived from these simulations did not fit well the experimental SAXS data and had χ 2 values of ∼6.4. Therefore, we subsequently allowed the N-and C-lobes of HECT to move freely with respect to each other. From these simulations, the best-scoring CORAL model ( Supplementary  Fig. S6) showed a considerably better agreement with the experimental SAXS data (χ 2 of 1.63), matching all intermolecular crosslinks (Supplementary Table S4), the intramolecular crosslinks of complexed Nedd4-2 (Supplementary Table S2 and S3) and the ab initio shape reconstruction ( Supplementary Fig. S6b,  S4f). The model of the complex suggests that the WW3 domain of Nedd4-2 (shown in blue) is positioned within the central channel of the 14-3-3η dimer and interacts with the 14-3-3η helices α1, α3 and α9 (Fig. 7d). The HECT domain is located outside the central channel of 14-3-3η dimer, its C-lobe (shown in salmon) is close to the 14-3-3η α8-α9 loop, and the N-lobe (shown in raspberry) is positioned near the 14-3-3η helices α6 and α8. Although the HECT domain interacts with one 14-3-3η protomer, the WW4 and WW2 domains are located above the second 14-3-3η protomer, far from the HECT domain ( Supplementary Fig. S6a). Thus, by comparison with the model of Nedd4-2 186-975 alone, 14-3-3η binding changed the relative positions of the structured domains of Nedd4-2, including the N-and C-lobes of HECT, where the C-lobe interacts with the helices α6, α7 and α10 of the N-lobe. Taken together, our SAXS and crosslinking data highlight that 14-3-3 binding induces a conformational rearrangement of Nedd4-2 by changing the relative positions and interactions of its structured domains, including the N-and C-lobes of the catalytic HECT domain, and by masking the surface of the WW3 domain.

Discussion
Nedd4-2 primarily ubiquitinates membrane-bound proteins, such as channels and receptors, and its dysfunction leads to multiple diseases, including epilepsy, hypertension, cystic fibrosis, pulmonary edema or Liddle syndrome [36][37][38][39][40] . Previous studies have shown that Nedd4-2 phosphorylation triggers binding to the adaptor protein 14-3-3 and that this interaction may have different functional consequences depending on the tissue. In the brain, for example, this interaction promotes the ubiquitination of the GluA1 subunit of the AMPA receptor 36 . In contrast, in the kidney, Nedd4-2 binding to 14-3-3 weakens ENaC ubiquitination, thereby increasing sodium absorption by elevating the surface expression of ENaC 22 . However, key structural aspects of the 14-3-3-mediated regulation of Nedd4-2 have remained unclear until now. Thus, our study provides detailed structural insights into the interaction between 14-3-3 protein and Nedd4-2 by combining several structural biology approaches.
The interaction between the 14-3-3 protein and Nedd4-2 is mediated by phosphorylation of three residues bordering WW2 domain: pSer 342 , pThr 367 and pSer 448 (Fig. 1) 17,23 . Because 14-3-3 protein dimers frequently interact with their binding partners by simultaneously anchoring two phosphorylated motifs separated by at least 20 amino acid residues 41 , we hypothesized that the interaction between 14-3-3 and Nedd4-2 is also mediated by such bidentate binding. Our LC-MS analysis confirmed that Nedd4-2 is phosphorylated by PKA in vitro on all three aforementioned sites. Mutating all three residues abolished Nedd4-2 binding to the 14-3-3 protein, whereas simultaneous phosphorylation of the Ser 342 and Ser 448 sites led to the highest affinity binding of Nedd4-2 to the 14-3-3η dimer, in the nanomolar range (K D < 30 nM) (Fig. 3, S3). The binding affinity decreased upon single phosphorylation of Ser 448 or other double phosphorylation combinations involving Ser 448 . Our fluorescence polarization measurements also highlighted the importance of the Ser 448containing motif for high-affinity Nedd4-2 binding to 14-3-3η. Concurrently, our crystallographic analysis of the phosphorylated Nedd4-2 335-455 bound to 14-3-3η confirmed the bidentate interaction with the 14-3-3 dimer through both 14-3-3 binding motifs, pSer 342 and pSer 448 , also showing that the complex establishes more contacts than the structures of the isolated singly phosphorylated motifs (Fig. 5). These findings support the  assumptions that the key residue for 14-3-3 protein binding is Ser 448 , phosphorylated by either SGK or PKA, and that the interaction between Nedd4-2 and 14-3-3η depends on a Pro residue located at position +2 from pSer 448 17,20,22 . Although pSer 448 alone promotes Nedd4-2 binding to 14-3-3, our data clearly show that the most stable interaction involves simultaneous binding of two phosphorylated motifs. In line with our results, decreased phosphorylation of Nedd4-2 at Ser 342 has been recently shown to promote its association with ribosomal proteins during endoplasmic reticulum stress 42 , but whether this process also involves changes in the interaction between Nedd4-2 and 14-3-3 is still unknown. Nevertheless, our structural analysis of this interaction did show that 14-3-3 changes the relative positions of all Nedd4-2 domains. 14-3-3 proteins are well known to regulate their binding partners by modulating their structure and/or masking structural and functional features on their surface. The conformational change upon the 14-3-3 protein binding was demonstrated for the serotonin N-acetyltransferase (AANAT), an enzyme that controls the daily rhythm in melatonin synthesis. The 14-3-3 binding modulates structure of the substrate binding sites of AANAT, thereby increasing the affinity of AANAT for its substrates with an accompanying increase in activity 43,44 . In our previous work on 14-3-3-mediated regulation of the neutral trehalase Nth1, we demonstrated that 14-3-3 protein triggers its enzyme activity by stabilizing the interaction between its catalytic and regulatory domains 32 . Another example of 14-3-3-induced structural changes upon binding is the negative regulation of B-RAF kinase, which 14-3-3 protein maintains in an inactive state by blocking the membrane recruitment of B-RAF and by preventing B-RAF kinase dimerization through steric occlusion of its domains 45,46 . Our results suggest that the 14-3-3-mediated regulation of Nedd4-2 requires simultaneous binding of two phosphorylated motifs followed by both the conformational change and steric occlusion of several functional domains, thus resembling regulations mentioned above. The CORAL model of Nedd4-2 186-975 alone indicated close interactions between the HECT and WW2/3 domains and full exposure of the WW1 and WW4 domains to the solvent (Fig. 7c and S5). This arrangement of domains corroborates previous evidence of the interaction between WW domains of Nedd4-2 and its own weak PY motive located within the C-lobe of HECT (Fig. 7c) 13,47 . On the other hand, 14-3-3 binding blocked interactions between HECT and WW2/3 by sequestering WW3 within the central channel of the 14-3-3 dimer and by weakening the interaction between the Nand C-lobes of the HECT domain ( Fig. 7d and S6a). Furthermore, in a complex with 14-3-3, the WW4 domain of Nedd4-2 is located near the surface of the 14-3-3η dimer, whereas WW1 and WW2 are fully exposed to the solvent (Supplementary Fig. S6a). Such position of WW2, far from the surface of the 14-3-3η dimer, may also explain why we were unable to trace this domain in the electron density map of Nedd4-2 335-455 bound to 14-3-3η (Fig. 5e). Because WW domains presumably mediate the interaction between Nedd4-2 and its substrates, such occlusions or exposures likely affect substrate ubiquitination, accounting for the 14-3-3-mediated modulation of the ubiquitination of some Nedd4-2 substrates 36,48-50 . Considering the above, our data provide the structural glimpse into 14-3-3-mediated Nedd4-2 regulation, showing that 14-3-3 protein regulates multidomain binding partners through several common mechanisms, either promoting or blocking interdomain interactions and sterically occluding functional surfaces, among other alterations.
In conclusion, Nedd4-2 is phosphorylated on multiple sites by PKA, but dual phosphorylation on the sites Ser 342 and Ser 448 mediates high-affinity Nedd4-2 bonding to 14-3-3η. Upon binding, 14-3-3η induces a structural rearrangement of Nedd4-2 by altering interactions between the structured domains of Nedd4-2, including the N-and C-lobes of the catalytic HECT domain. Changes in the exposure of WW domains may explain how 14-3-3 modulates the ubiquitination of some Nedd4-2 substrates. For this reason, further studies should be conducted to identify the exact mechanisms of 14-3-3-dependent regulation of ubiquitination of particular substrates and to assess whether these mechanisms include variations in Nedd4-2 phosphorylation and whether the suggested conformational change of the HECT domain has any functional consequences. For now, our findings lay the foundations for future research aimed at understanding the versatile regulatory roles of 14-3-3 proteins in the regulation of signaling pathways and processes linked to protein degradation through the ubiquitin-proteasome system. Such studies are particularly relevant because many other E3 ubiquitin protein ligases, e.g., parkin, ZNRF2, CBL, ITCH and SMURF1, are also regulated in a 14-3-3-dependent manner [51][52][53][54][55][56] . Moreover, recent advances in the development of small molecule compounds targeting protein-protein interactions have demonstrated the potential to modulate the activity of key 14-3-3 binding partners in various physiological processes by targeting their distinct interactions with 14-3-3 57 . In this context, the interaction between Nedd4-2 and 14-3-3 proteins should be a promising target for the treatment of Nedd4-2-associated diseases.
Heterologous expression, purification and phosphorylation of Nedd4-2. Nedd4-2 coding sequences (residues 335-455 and 186-975) were PCR-amplified from the plasmid hNedd4-2 (a gift from Christie Thomas, Addgene plasmid # 83433) 59 . The PCR product containing residues 335-455 was ligated into the pRSFDuet-1 (Novagen) using the NcoI/NotI restriction sites. Modified pRSFDuet-1 contained the sequence of the His 6 -tagged GB1 domain of protein G inserted into the first multiple cloning site (a gift from Evzen Boura, IOCB CAS). The PCR product, including residues 186-975, was subcloned into the expression vector pST39 (a gift from Evzen Boura, IOCB CAS) using the XbaI/KpnI restriction sites. The entire cloned regions were confirmed by sequencing. Mutations of Nedd4-2 with a different number of PKA phosphorylation sites (at positions 342, 367 and 448) were generated by mutating other sites to alanine, using the QuikChange TM approach (Stratagene), and confirmed by sequencing. Oligo sequences are provided in Supplementary Table S5.
The eluted Nedd4-2 335-455 protein and its variants were dialyzed against the buffer containing 20 mM Tris-HCl (pH 7.5), 2 mM EDTA, 2 mM 2-mercaptoethanol and 10% (w/v) glycerol. The His 6 -GB1 tag was cleaved by incubation with the TEV protease (750 U of TEV per 1 mg of fusion protein) in dialysis overnight at 4°C. Nedd4-2 335-455 was phosphorylated by incubation at 30°C for 2 h and then overnight at 4°C with 1300 U of PKA (Promega) per 1 mg of protein in the presence of 0.75 mM ATP and 20 mM MgCl 2. TEV and PKA were removed through another immobilized metal affinity chromatography and subsequent size-exclusion chromatography using a HiLoad™ 26/600 Superdex™ 75 pg column (GE Healthcare) in a buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM TCEP and 10% (w/v) glycerol. The typical yield was 5 mg of pure protein per one liter of LB media.
The level of the phosphorylation of both proteins was confirmed by the mass spectrometry service provided by CMS Biocev.
Heterologous expression and purification of Uba1, Ube2d2 and Ub. The coding sequences of mouse Uba1 (ubiquitin-like modifier-activating enzyme 1), mouse Ube2d2 (ubiquitin-conjugating enzyme E2D2) and human Ub were kindly provided by Dr. Silhan (IOCB, CAS). Both enzymes and human ubiquitin were expressed in Escherichia coli BL21 (DE3) cells (Novagen). The E1 enzyme Uba1 was subcloned into a pET28a plasmid with a TEV-cleavable 6 × His-tag at the Nterminus and expressed at 16°C O/N. The E2 enzyme Ube2d2 and human ubiquitin Ub were subcloned into a pET15b plasmid with a TEV-cleavable 6 × His-tag at the N-terminus and expressed at 25°C O/N. All these fusion proteins were expressed in Luria-Bertani media by induction at OD 600 = 0.8 for 20 h. The pelleted cells were suspended in lysis buffer (1 × PBS, 1 M NaCl, 4 mM βmercaptoethanol and 2 mM imidazole) and purified using a Chelating Sepharose™ Fast Flow column (GE Healthcare) according to the standard protocol. The fusion proteins were dialyzed into a buffer containing 20 mM Tris·HCl (pH 7.5), 150 mM NaCl, 2 mM DTT and 10% (w/v) glycerol. The His 6 -tag of Uba1 and Ube2d2 were cleaved by incubation with TEV protease at 30°C for 2 h. The final purity of all proteins was confirmed by size-exclusion chromatography on Superdex™ 200 Increase 10/300 GL (GE Healthcare) or Superdex™ 75 Increase 10/300 GL (GE Healthcare) columns in a buffer containing 20 mM Tris·HCl (pH 7.5), 150 mM NaCl, 2 mM DTT and 10% (w/v) glycerol. The proteins were concentrated as required, aliquoted and flash-frozen in liquid nitrogen.
Ubiquitination assay. For in vitro ubiquitination assays, 5 µg of human Ub, 50 ng of E1 (mouse Uba1), 200 ng of E2 (mouse Ube2d2) and 500 ng of E3 (phosphorylated Nedd4-2 186-975 ) were incubated in 30 µl of the reaction mixture in a buffer containing 50 mM Tris pH 7.5, 2 mM ATP, 5 mM MgCl 2 and 2 mM DTT and incubated for 5, 10 and 15 min at 30°C. The reactions were stopped by adding 5 × SDS-PAGE loading buffer (250 mM Tris pH 6.8, 50% glycerol, 500 mM DTT and 10% SDS) and incubating at 95°C for 3 minutes to denaturate the samples. For reactions with 14-3-3η, Nedd4-2 186-975 was mixed with 14-3-3η in a 1:2 molar ratio and incubated for 30 minutes on ice before starting the reaction. In addition, 20 µl of reaction was separated by conventional SDS-PAGE on a 10% acrylamide gel and electro-blotted against a PVDF membrane in 20 mM Tris pH 7.5, 154 mM glycin and 10% methanol buffer overnight at 4°C. Rabbit anti-Ub polyclonal antibody (Enzo) and anti-rabbit IgG, HRP-linked Antibody (Cell Signaling Technology) were used to visualize the transferred polyubiquitin chains. Chemiluminescence was induced by ECL and detected using Fusion Solo S (Vilber). The detected bands of E3-Ub were quantitated using the Image Lab software (Bio-Rad).
Analytical ultracentrifugation (AUC) measurements. Sedimentation velocity (SV) experiments were performed using a ProteomLab TM XL-I, Beckman Coulter analytical ultracentrifuge 64 . The samples were dialyzed against a buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 1 mM TCEP before analysis. The samples with the longer construct of Nedd4-2 186-975 were dialyzed against a buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 1 mM TCEP and 0.01% (v/v) Tergitol NP-40 (Sigma). SV AUC experiments were conducted at various loading concentrations of 14-3-3η, Nedd4-2 186-975 , Nedd4-2 355-455 and its mutant variants in charcoal-filled Epon centerpieces with a 12-mm optical path length at 20°C, and at 40000 or 46000 rpm rotor speed (An-50 Ti rotor, Beckman Coulter). The buffer density, viscosity, and partial specific volume of all proteins were estimated using the program SEDNTERP. All sedimentation profiles were collected with absorbance optics at 280 nm. The sedimentation coefficients c(s) distributions were calculated from raw data using the SEDFIT software package 65 . The calculated distributions were integrated to establish the weight-average sedimentation coefficients corrected to 20°C and the density of water, s w(20,w) . The s w values were plotted as a function of Nedd4-2 355-455 or 14-3-3η concentration to construct s w isotherms. The resulting isotherms were fitted with a A + B ⇌ AB model, as implemented in the SEDPHAT software package with previously known s w values of each component. The parameters were verified, and the loading concentrations were corrected using global Lamm-equation modeling, also implemented in the SEDPHAT software 66 Chemical Crosslinking combined with mass spectrometry. Both Nedd4-2 186-975 alone and in a complex with 14-3-3η were crosslinked with homobifunctional crosslinkers DSG and DSS in 1:1 (mol/mol) mixtures of non-deuterated and fourtimes-deuterated compounds (d0/d4). For the crosslinking reaction, the proteins were dialyzed in a buffer containing 20 mM HEPES (pH 7.5), 150 mM NaCl and 1 mM TCEP. In all samples, the protein concentration was 0.6 mg.ml −1 . Freshly prepared stock solutions of crosslinkers (10 mg.ml -1 in DMSO) were added to proteins at a 50 × molar excess in a total reaction volume of 20 μl and incubated for 1 h at room temperature.
The proteins were reduced with 10 mM dithiothreitol, alkylated with 30 mM iodoacetamide, and digested overnight with trypsin at 37°C. The peptides were injected into a Luna Omega 5μm Polar C18 100 Å 20 × 0.3 mm column (Phenomenex) and desalted at 20 μL/min for 5 min. Peptides were then separated by reversed-phase chromatography with a Luna Omega 3μm Polar C18 100 Å 150 × 0.3 mm column (Phenomenex) at 10 μL/min using a capillary UHPLC 1290 system (Agilent Technologies) with a gradient sequence of 1−10% for 1 min, 10−45% for 19 min, and 45−95% for 5 min of buffer solvent A (0.1% formic acid, 98% acetonitrile in water) in buffer solvent B (0.1% formic acid, 2% acetonitrile in water). The column was heated to 50°C and directly connected to a 15 T solariX XR FT-ICR mass spectrometer (Bruker Daltonics, USA) operated in positive data dependent mode. StavroX software 67 was used to identify crosslinked peptides. Fixed carbamidomethylation of cysteines and variable methionine oxidation were set as modifications. The modification sites of DSG and DSS were N-termini, lysines, serines, threonines and tyrosines. The mass error threshold was set to 1 ppm, and all assigned fragments were manually curated. For quantitative studies, the Links software 68,69 was used to identify crosslinked peptides. The nonoverlapping isotopes of peptides crosslinked with 12 C and 13 C DSA were used to calculate isotope ratios 70 .
The loading concentrations of the samples were 2.2 mg.ml −1 for Nedd4-2 186-975 , 2 mg.ml −1 for 14-3-3η and 8.9 mg.ml −1 for the Nedd4-2 186-975 :14-3-3η complex (1:2 molar stoichiometry). The injection volume of the samples was 52 μl. Elution profiles with the corresponding frames were processed using ATSAS 3.0.2 software CHROMIXS 71 . ATSAS software PRIMUS and GNOM were used to determine the excluded volume of the hydrated particle (V p ), the radius of gyration (R g ) and maximum particle dimension (D max ) 72,73 . The molecular mass (M W ) was determined using a methods described by Rambo et al. 74 . Scattering profiles, P(r) functions, normalized Kratky plots and Guinier approximations were visualized using GraphPad Prism version 8.0.1 for Windows (GraphPad Software, La Jolla California USA, www.graphpad.com). The program DAMMIF was used to calculate the ab inicio molecular envelopes 75 . For the complex, 10 surfaces were generated in slow mode and averaged using DAMAVER 76 . The averaged surfaces were used as the final filtered model of the complex, which was superimposed to the structural model using the SUPCOMB program (EMBL, Hamburg, Germany) 77 .
Structural modeling. The three-dimensional models of Nedd4-2 186-975 and the Nedd4-2 186-975 :14-3-3η complex were generated using CORAL. SAXS data and distance constraints derived from intramolecular crosslinks of Nedd alone (for model of Nedd4-2 186-975 alone), intermolecular crosslinks between 14-3-3η and the Nedd4-2 186-975 and intramolecular crosslinks of the complexed Nedd4-2 186-975 (for model of the Nedd4-2 186-975 :14-3-3η complex) were used 78 . The starting model of the complex was prepared using the crystal structures of 14-3-3η (PDB ID: 2C63) 79 and 14-3-3γ with the Nedd4-2 phosphopeptides pep-S342 and pep-S448 (PDB ID: 6ZBT and 6ZC9), as well as known solution structures of the WW1-3 (PDB ID: 1WR3, 1WR4, 1WR7) and HECT (PDB ID: 5HPK) domains 80 . The unknown structure of the WW4 domain was generated using the SWISS-MODEL server (https://swissmodel.expasy.org/). Using the CORAL program, disordered loops missing in the crystal structures were modelled as interconnected dummy residue chains attached to the appropriate Cα atoms in rigid domains, but the linker length was limited to 5−100 residues. For this reason, the part of the Nedd4-2 region connecting the WW1 domain to the pSer 342 motif (residues Ser 233 -Phe 249 ) was modeled as the α-helix based on secondary structure prediction by PSIPRED (psipred@cs.ucl.ac.uk). The best-scoring CORAL model was chosen according to the lowest χ 2 , and the distances were derived from intra-and intermolecular crosslinks.
Statistics and reproducibility. Results from the FP assay (Fig. 4) and immunoblot ( Supplementary Fig. S1) are represented as means ± SD from three replicates as indicated in the figure legend. Statistical analysis for the data from immunoblot were performed using Graph-Pad Prism 8.4. Student's t-test was used for comparison of relative changes of samples with and without 14-3-3η at selected timepoints (ns, non-significant P > 0.05; *, P < 0.05).