{"paper_id":"26b64200-3be7-4ae7-b2a5-3d06eb8909ce","body_text":"1 \nConformational dynamics of the bacterial E3 ligase SspH1 \n \nCassandra R. Kennedy1, Diego Esposito1, David House2 and Katrin Rittinger1* \n \n1Molecular Structure of Cell Signalling  Laboratory, The Francis Crick Institute, 1 \nMidland Road, London, NW1 1AT, United Kingdom. \n2Crick-GSK Biomedical LinkLabs, GSK, Gunnels Wood Road, Stevenage, \nHertfordshire, SG1 2NY, United Kingdom. \n \n* Correspondence: katrin.rittinger@crick.ac.uk. \n \n \n \n \n \n \n \n \n \n \n \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 2 \nABSTRACT \nThe SspH/IpaH family of novel E3 ligases (NELs)  are found in a number of Gram -\nnegative bacteria and are used to target host enzymes for degradation to support \npathogenesis. These E3 enzymes are autoinhibited in the absence of substrate and \ndifferent models for release of autoinhibition have been suggested. However, many of \nthe molecular details of individual steps during the ubiquitin transfer reaction remain \nunknown. Here, we present the crystal structure of Salmonella SspH1 and an analysis \nof the solution properties of SspH1 on its own and in complex with substrate and \nubiquitin. Our data show that SspH1 exists in a conformation al equilibrium between \nopen and closed states and that substrate binding only modulates the distribution of \nthese states but does not induce major conformational changes. This suggests that \nadditional mechanisms must exist to bring the substrates close to the active site to \nmediate transfer of ubiquitin from the E3~Ub conjugate. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 3 \nINTRODUCTION \nGram-negative bacteria  remain a huge disease burden  globally. Shigella and non -\ntyphoidal Salmonella together account for an estimated 660,000 human deaths \nworldwide each year.1 A key part of Salmonella and Shigella pathogenesis is the \ndelivery of virulence proteins (effectors) into the host cell through a Type 3 Secretion \nSystem (T3SS) to interfere with host immune responses  and support bacterial \nreplication.2,3 Effector proteins perform a range of cellular functions and often have \nenzymatic activities that  include proteases, acetyltransferases,  kinases, \nphosphatases, E3 ubiquitin ligases and deubiquitinases (DUBs).2 Ubiquitination plays \nan important role in the regulation of eukaryotic cellular processes and is a key \nmechanism to target proteins for proteasomal degradation. It is mediated by a 3-step \nenzymatic cascade including E1 activating, E2 conjugating and E3 ligase enzymes. 4 \nIntriguingly, bacteria do not have a canonical ubiquitin system themselves but have \nevolved proteins that mimic and hijack the host ubiquitin system to support their \nsurvival and proliferation .5–7 Many p athogenic bacteria contain E3 ligases most of \nwhich structurally resemble their eukaryotic counterparts, yet some bacteria including \nSalmonella and Shigella have evolved a new class of E3 ligases, the Novel E3 Ligase \nfamily (NELs), that have no structural homology to eukaryotic E3s in their catalytic \ndomain but function via an active site cysteine , analogous to HECT or RBR E3 \nligases.8,9 These catalytic cysteine-containing E3 ligases transfer ubiquitin onto the \nsubstrate in a 2-step reaction: first ubiquitin is transferred from the E2~Ub conjugate \nto form an E3~Ub conjugate in a transthiolation reaction, and subsequently onto lysine \nresidues in the substrate or ubiquitin itself via an aminolysis reaction. \nNELs are found in a number  of gram-negative bacteria such as the mammalian \npathogens Salmonella and Shigella, and plant pathogens Ensifer fredii and Ralstonia \nsolanacearum.10 They are composed of an N -terminal secretion motif, a leucine -rich \nrepeat (LRR) domain and a C -terminal NEL catalytic domain and are structurally \nconserved across the family. Crystal structures and mechanistic studies of NELs have \nprovided insight into the activities of the two key functional domains: the LRR domain,9 \nwhich is responsible for substrate recognition, and the NEL domain, 8 which can be \ndivided into two subdomains: the C -terminal subdomain (CSD) containing the E2 -Ub \nbinding ‘thumb’ (E2-UbBD) and the N-terminal subdomain (NSD) which encompasses \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 4 \nthe catalytic E3 region containing the active site cysteine. 11 A linker region joins the \nLRR and NEL domains together, however the mechanistic role and importance of \nflexibility of this linker remains poorly defined.  \nThe Shigella NEL family members comprise IpaH proteins, with the best characterized \nmembers being IpaH9.8, IpaH1.4/2.5 and IpaH7.8, whose targets include GBP1, 12–14 \nSte7,15 NEMO,16 LUBAC,17 and Gasdermin B. 18 The Salmonella NEL subfamily \nincludes SspH1, SspH2 and SlrP, which target PKN1, 19,20 NOD1/SGT1,21 and \nthioredoxin,22 respectively. Upon ubiquitination, these host proteins are directed to \nproteasomal degradation, thereby suppressing the host immune and inflammatory \nresponse to bacterial infection. \nThe exact mechanism underlying the regulation of NEL E3 ligase activity has been the \nsubject of many studies. The isolated NEL domain is constitutively active and forms \nfree ubiquitin chains, while ubiquitination activity is suppressed in the full -length \nproteins.8 Auto-inhibition of NEL activity is important to prevent auto-ubiquitination and \nsubsequent proteasomal degradation or formation of unanchored ubiquitin chains \nwhich could stimulate a host immune response. 23–26 Initially, NEL proteins were \nthought to be auto-inhibited due to steric blockage of the catalytic region by the LRR \ndomain, thereby preventing access to the E3 active site, until release of inhibition by \nsubstrate binding.27 This interpretation was based on crystal structures of NEL ligases \nincluding IpaH3,9 SspH2,28 and IpaH9.8,29 capturing two distinct ‘open’ and ‘closed’ \nconformations of the NEL proteins. In the ‘closed’ conformation (SspH2, PDB: 3G06), \nthe LRR domain is oriented towards the NEL domain thereby apparently occluding \naccess to the catalytic cysteine. In the ‘open’ conformations (IpaH3, PDB 3CVR; \nIpaH9,8, PDB 6LOL), the LRR domain is orientated  away from the NEL domain, and \nthese were assumed to represent active conformations. However, elegant biochemical \nexperiments using mutants of the catalytic acid and base residu es have since shown \nthat NELs are not auto -inhibited but can form an E3~Ub intermediate, which \nundergoes rapid non -productive ubiquitin turnover in the absence of substrate, \nindicating that the catalytic cysteine is accessible even in the absence of substr ate \nbinding.30 In addition, a recent study on IpaH9.8 has suggested that two residues in \nthe concave surface of its LRR sense substrate binding to induce a conformational \nchange and release a hydrophobic cluster that mediates auto -inhibition. This \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 5 \nmechanism requires an amplifier loop in the LRR domain which however is not present \nin all family members.29  \nCurrently it is unclear if the se described regulatory mechanisms apply across the \nentire NEL family, nor have the intermediate steps in the ubiquitin transfer reaction \nbeen captured structurally. The first step in substrate ubiquitination, the initial transfer \nof ubiquitin from E2 to the NEL has been described to rely on significant intradomain \nflexibility between the catalytic E3 region and the E2-UbBD thumb11. In contrast, the \nstructural determinants that mediate the switch from unproductive hydrolysis of the \nE3~Ub intermediate to ubiquitin transfer onto substrate lysine residues is not \nunderstood. Similarly, the effect of substrate binding on the conformational flexibility of \nthe LRR and NEL domains with respect to one another has not been studied. \nTo address these questions, we present a comparison of catalytic activities of NEL \nproteins across the Shigella and Salmonella subfamilies, highlighting similarities and \ndifferences between family members. To better understand the molecular mechanism \nof SspH1, we combine the structural characterization of SspH1 by crystallography with \nSAXS and NMR analysis to interrogate the solution state structure and conformational \nflexibility of SspH1 when bound to  its substrate  PKN1 and to ubiquitin.  Our data \nindicate that the open and closed forms of SspH1 are in a conformational equilibrium \nand that substrate binding modulates the distribution between closed and open states \nbut does not induce major conformational changes . Similarly, substrate binding has \nonly a minor effect on the flexibility of ubiquitin in the E3~Ub conjugate suggesting that \nadditional as yet unknown mechanisms must exist to bring the LRR -bound substrate \nclose to the ubiquitin thioester intermediate and initiate ubiquitin transfer.\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 6 \nRESULTS \nComparison of IpaH1.4, IpaH9.8, SspH1 and SspH2 catalytic activities \nNELs are highly homologous in their catalytic NEL domain, and many residues \npreviously identified as important for catalytic activity , including residues responsible \nfor E2-Ub binding as well as catalytic acid and catalytic base residues, are conserved \nacross the family. The main differences between NEL family members are in the LRR \ndomains, which mediate substrate recognition but may also affect ubiquitin transfer as \nsuggested for the amplifier loop in IpaH9.8. To investigate if the mechanism of ubiquitin \ntransfer of NELs is conserved across the family we compared catalytic activities across \n4 family members: two Shigella, IpaH1.4 and IpaH9.8, and two Salmonella proteins, \nSspH1 and SspH2, as these have well established substrates. A sequence alignment \nof these family members is shown in  Supplementary Figure 1A, 31 highlighting the \nconservation of key catalytic residues but also differences such as the lack of the \nIpaH9.8 amplifier loop in SspH1 and SspH2. The LRR domain of SspH2 is longer \ncompared to the other three proteins, corresponding to 12 LRR repeats compared to \n8 repeats for IpaH9.8 and 1.4.28,29 Structural models generated by AlphaFold2 predict \nan additional globular N -terminal domain in SspH1 and SspH2 which is structurally \nconserved across the two proteins (Supplementary Figure 1B),32–34 and found in other \nbacterial effectors such as SifA, though its function in E3 ligase activity has not been \ninvestigated thus far. \nTo interrogate the mechanisms of ubiquitin transfer across the Shigella IpaH and \nSalmonella SspH proteins, we purified three constructs for each of IpaH1.4, IpaH9.8, \nSspH1 and SspH2: the full -length protein (FL); the NEL domain (NEL); and the full -\nlength protein minus the N-terminal secretion signal or globular domain in SspH1 and \nSspH2 (∆N) ( Figure 1A ). First, we confirmed substrate ubiquitination activity of \nIpaH1.4, IpaH9.8 and SspH1 with their respective substrates LUBAC, GBP1 and \nPKN1 HR1b domain for the ∆N (Figure 1B) and FL constructs (Supplementary Figure \n2A). Substrate ubiquitination by SspH2 could not be investigated due to difficulties with \nproducing sufficiently pure SGT1/NOD1 substrate complex. For each construct of \neach protein, we performed a further three types of assays to assess activity: first we \nperformed in vitro ubiquitination assays, where overall E3 lig ase activity is monitored \nthrough autoubiquitination or formation of free ubiquitin chains. Next, we carried out \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 7 \nE2~Ub complex discharge assays using fluorescently labelled ubiquitin to isolate the \nE2~Ub and E3~Ub formation steps of the ubiquitin cascade. If a stable E3~Ub \nintermediate is formed this can be observed on non -reducing SDS gels. Finally, we \nperformed ubi quitin-loading assays with a covalent ubiquitin probe (ubiquitin vinyl \nsulfone, UbVS) to assess accessibility of the E3 catalytic cysteine.  \nAs expected, we observe some similarities in activities across the four Shigella and \nSalmonella NELs. In substrate ubiquitination assays, we observed equal substrate \nubiquitination with both ∆N and FL proteins indicating that the N -terminal globular \ndomain present in SspH1/2 has no effect on catalytic activity ( Figure 1B , \nSupplementary Figure 2A, Supplementary Figure 3A -C). In contrast, in the absence \nof substrate, no auto-ubiquitination or free chain formation activity was observed with \n∆N and FL proteins ( Figure 1C , Supplementary Figure 2B ). In E2~Ub discharge \nassays, FL and ∆N constructs discharge ubiquitin from UbcH5A~Ub ( Figure 1D and \nE, Supplementary Figure s 2C and 3E ) indicating that their catalytic cysteine is \naccessible but no stable E3~Ub thioester intermediate is formed. This is in agreement \nwith the labelling observed with UbVS in ubiquitin -loading assays ( Figure 1F , \nSupplementary Figure 2D). The isolated NEL domains alone did not auto-ubiquitinate, \nhowever, free ubiquitin chains were synthesized indicating that removal of the LRR \ndomain released auto -inhibition (Figure 2A, Supplementary Figure 3 D). Similarly, in \nE2~Ub discharge assays with isolated NEL domains some free di -ubiquitin was \nsynthesized (Figure 2B), and all constructs were labelled by UbVS (Figure 2C). \nHowever, we also observed differences between the activities of these proteins. \nLabelling with UbVS was different between the four NEL family members: both FL and \n∆N constructs of IpaH9.8 and IpaH1.4 were approximately 50% labelled after two \nhours, whereas SspH2 was almost completely labelled, in contrast to SspH1 which \nshowed the lowest level of labelling ( Figure 1F, Supplementary Figure 2D). With the \nNEL domains alone, labelling was more similar across the family (Figure 2C), though \nSspH1 still showed the lowest degree of labelling. In NEL domain -only ubiquitination \nassays, we observed a similar propensity to form free ubiquitin chains with IpaH1.4 \nand IpaH9.8, while SspH1 and SspH2 appeared less active (Figure 2A).  \nSimilarly, time course E2~Ub discharge assays with ∆N constructs highlighted that \nIpaH1.4 and IpaH9.8 were much more efficient at discharging ubiquitin than SspH1 or \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 8 \nSspH2 (Figure 1D and E ). To observe complete ubiquitin discharge within the same \ntime frame, a 10 -fold higher concentration of SspH proteins was required when \ncompared to the IpaH proteins. This corroborates previous work by Keszei and Sicheri, \nwhere longer assay time points we re used to study SspH1 activity compared to \nIpaH9.8.30 For E2~Ub discharge assays with isolated NEL domains , we observed \nsimilar rates of free di-ubiquitin synthesis with IpaH1.4 and IpaH9.8, while SspH2 was \nsignificantly less efficient compared to the other three proteins ( Figure 2B ). \nInterestingly, the E3~Ub thioester intermediate was sufficiently stable in these assays \nthat it could be observed at early time points with IpaH1.4, IpaH9.8 and SspH1.  \nWe found it intriguing that SspH1 is the apparently least accessible of the NELs tested \nto ubiquitin -loading with UbVS ( Figure 1F ) and has a lower turnover in discharge \nassays compared to IpaH proteins (Figure 1D and E, Figure 2B) but is the most active \nin substrate ubiquitination assays ( Figure 1B ). Therefore, we tested whether the \npresence of PKN1 HR1b substrate might alter the dynamics of SspH1 and therefore \naccessibility of the catalytic cysteine. However, we observed that PKN1 had no effect \non SspH1 lo ading with UbVS ubiquitin ( Supplementary Figure 4A ). This is \ncomplementary to, and consistent with E2~Ub discharge assays performed by Keszei \nand Sicheri where the addition of PKN1 HR1b to SspH1 had no impact on E2~Ub \ndischarge rate, indicating that substrate binding does not significantly alter access to \nthe SspH1 active site.30 \nΔNSspH1 crystal structure \nTo better understand the mechanism of SspH1 and the apparent differences in activity, \nwe solved the crystal structure of a construct containing the LRR and NEL domains \nbut lacking the first 160 residues, ΔNSspH1 (now referred to as SspH1). The structure \nwas refined at 2.9 Å resolution with one copy in the asymmetric unit and a P622 space \ngroup. Details of data collection and statistics are reported in Supplementary Table 1.  \nThe structure is elongated with the LRR (residues 161 -393) domain extending away \nfrom the NEL domain (residues 405-622) which contains the catalytic cysteine (C492) \nand the a-helices of the E2-Ub binding domain (E2-UbBD, residues 623-697) (Figure \n3A). It is similar to IpaH9.8 and IpaH3 in an ‘open’ conformation, with the position of \nthe E2-UbBD relative to the catalytic loop in an analogous position to other bacterial \nE3 ligase structures (Figure 3B). The E2-UbBD domain of SspH1 consists of two long \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 9 \nanti-parallel α-helices and a small C-terminal helical segment as also seen in SspH2 \n(Figure 3C and Supplementary Figure 1A ). In contrast, IpaH9.8, IpaH1.4 and SlrP \nhave a notable kink in one of the two long anti -parallel α-helices ( Figure 3C). In \nagreement with our sequence alignment (Supplementary Figure 1A), SspH1 lacks the \namplifier loop in LRR6 of the IpaH9.8 LRR domain which was shown to be important \nfor GBP1 substrate sensing. 29 SspH2 also lacks an amplifier loop, while IpaH3 is \nmissing electron density for eight amino acids in the corresponding LRR, and SlrP has \nan extended loop in the next LRR.9,28 Whether these loops act as signaling amplifiers \nupon substrate binding is unknown.  \nIn our SspH1 structure, sixty N -terminal residues, those in the LRR -NEL linker (394-\n404) and the loop containing the catalytic C492 have a higher B -factor compared to \nother residues suggesting more flexibility in these regions. A highly dynamic LRR-NEL \nlinker and catalytic loop might be an important mechanistic feature in bacterial E3 \nligases as high B-factors or lack of electron density are also observed for these regions \nin other available NEL structures that contain both LRR and NEL catalytic domains: \n∆N SspH2, IpaH3 and IpaH9.8 do  not have electron density for the LRR -NEL linker, \nwhile IpaH9.8 is also missing density for the catalytic loop. 9,28,29 SlrP, is currently the \nonly available structure of a n LRR-NEL construct that has its substrate thioredoxin \nbound. The complex crystallized as a heterotetramer, with the two SlrP molecules in \nan open, head to tail arrangement that are bridged by 2 molecules of thioredoxin -1, \nthough it is not known if self-association is required for substrate ubiquitination by SlrP. \nIn this structure there is well -defined density for the loop linking the LRR and NEL \ndomains which makes contacts with Trx-1.35   \n \nWe observed a symmetry related molecule in our structure of SspH1 where the E2 -\nUbBD of one molecule sits on the top of the LRR domain, close to its substrate binding \ninterface (Supplementary Figure 5A). To determine if this was a biologically relevant \ninteraction or an artefact of crystal packing, we performed SEC (size -exclusion \nchromatography)-MALLS (multi -angle laser light scattering measurements) with FL \nSspH1 and ∆NSspH1. We confirmed that, in the range of concentrations explored, \nboth are monomeric in solution ( Supplementary Figure 5B and C ), and form a 1:1 \ncomplex with the substrate PKN1 ( Supplementary Figure 5D and E ), in agreement \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 10 \nwith a previous report.11 This is in contrast to SlrP which has been shown to exist in a \nmonomer-dimer equilibrium.35 \nWe next interrogated the importance of residues in SspH1 that make contacts between \nthe LRR and NEL domains to better understand the interplay between the se two \ndomains and their role in autoinhibition .  R351, which is conserved in SspH1/2 and \nSlrP but not IpaH proteins forms key interactions with E493 and D552, while R375, \nwhich is an Arg or His across the NEL family (Supplementary Figure 1) points towards \nan acidic pocket formed by residues surrounding the catalytic acid and base  (Figure \n4A and Supplementary Figure 6A).30 However we observed no significant difference \nbetween mutants R351A, R375A, R351A R375A and WT SspH1 activity in either auto-\nubiquitination or PKN1 ubiquitination assays (Figure 4B), indicating that these residues \ndo not regulate autoinhibition. Additionally, hydrophobic interactions with L550 at its \ncenter further contribute to the LRR -NEL interface (Supplementary Figure 6A ). This \nobservation supports previous mutational analysis of L550 which showed that L550S \npartially released autoinhibition but had no apparent effect on substrate ubiquitination \n27,36. Interestingly, the L550D mutation changed the pattern of substrate ubiquitination, \nsuggesting that this position affects the conformational changes required to bring the \nsubstrate close to the E3~Ub thioester (Supplementary Figure 6B). \nAt present, the SlrP/Trx-1 complex is the only available structure of a near full -length \nNEL bound to its substrate. However, the tetrameric nature of this complex makes it \ndifficult to draw general conclusions about the mechanism of substrate ubiquitination, \ngiven that other NEL fa mily members are assumed to transfer ubiquitin in a 1:1 \nNEL/substrate complex. Furthermore, no structure of apo SlrP is available to assess \nconformational changes induced upon substrate binding. Instead , near full -length \nIpaH9.8 has been structurally characterized in the apo state and the isolated LRR \ndomain bound to its substrate GBP1  providing first insight into the effect of substrate \nengagement.29 Comparison of these two structures allowed identification of structural \nchanges induced in the LRR upon substrate binding, especially a rotation of the LRR-\nCT toward the convex side of the LRR, which destabilizes a hydrophobic cluster in the \nC-terminus of the LRR that engages F395 (equivalent to L550 in SspH1) from the NEL, \nthereby enabling release of the NEL domain and autoinhibition. The structure of SspH1 \npresented here, now allows us to carry out the same comparison using the previously \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 11 \npublished structures of the SspH1 LRR in its apo form and in complex with its substrate \nPKN1 HR1b.27 Interestingly, an overlap of the PKN1-bound LRR with SspH1 revealed \nthat the changes induced upon substrate binding are much smaller than those \nobserved in IpaH9.8  with an RMSD of 0.685  Å (Supplementary Figure 6C), without \nany significant effect on the hydrophobic environment of L550 . This suggests that \nsubstrate binding to SspH1 alone might not be sufficient to induce the conformational \nchanges required to release autoinhibition. \nSolution properties of FL and ∆NSspH1, and substrate-bound complexes \nThe observation that LRR and NEL domains adopt different orientations with respect \nto one another in crystal structures  of apo NEL proteins  indicates the presence of \nsignificant interdomain flexibility, however little is known about substrate binding  \naffects this conformational flexibility . To investigate the solution properties of SspH1 \nwe recorded Small -angle X -ray Scattering (SAXS) data on full -length (FL) and \n∆NSspH1 constructs in the presence and absence of PKN1 ( Figure 5A-B and \nSupplementary Figure 7 ). Details of data collection and analysis are reported in \nSupplementary Table 2.  \nWe observed a 75 Å decrease in the maximum dimension (Dmax) and a 14 Å \nreduction in the radius of gyration (Rg) for ΔNSspH1 compared to the FL protein \nindicating a large contraction in the protein volume following the deletion of the small \nN-terminal domain. Interestingly, the change in molecular dimensions is not associated \nwith a change in the respective radii of cross-section (Rc) suggesting that the overall \nstructures maintain elongated shapes (Dmax>>Rg) of similar widths. Rather \nsurprisingly, the Guinier analysis of ΔNSspH1 and FL protein scattering data reveals \nsimilar radii of gyration for both proteins in isolation and bound to the HR1b domain of \nPKN1 (Supplementary Figure 7A). A 10 Å increase in the maximum dimension of the \nnormalized P(R) distributions is observed for both constructs following substrate \nrecruitment, suggesting that only small molecular rearrangements occur upon PKN1 \nbinding. In both cases the molecular weights calculated from the SAXS data are those \nexpected from the primary sequences suggesting the proteins in their apo forms and \nin complex with PKN1 are monomeric in solution with a 1:1 stoichiometry, as reported \npreviously and in agreement with our SEC -MALLS analysis ( Supplementary Figure \n5).11 \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 12 \nWe assessed the flexibility of the  protein constructs  under investigation  using \ndimensionless Kratky plots (Figures 5A and B). The broad bell shape with a maximum \nshifted towards higher q*Rg values, especially apparent in FL SspH1, together with \nthe long tail s in the P(r) distributions, confirms that SspH1 populat es elongated \nconformations. Moreover, the uptrend of the Kratky profile at higher q*Rg indicates the \npresence of overall conformational flexibility which is unaffected by the presence of \nthe substrate. Fitting the solution scattering data using either our ΔNSspH1 crystal \nstructure coordinates or an AlphaFold3 model, both of which are characterized by \ndistinct orientations of the LRR and E2-UbBD relative to the catalytic E3 domain, yields \nlarge χ2 values in both cases . This indicates that neither structure accurately \nrepresents the solution conformation of SspH1, most likely due to its dynamic features. \nTo explore the conformational space available to ΔNSspH1 we modelled solution \nstructure ensembles based on the experimental X-ray scattering data using an Xplor-\nNIH implemented protocol.37 Ensembles were chosen based on the agreement with \nthe experimental data (χ2 ≤ 1.5) and visualized in a plot reporting the angles between \nthe centers of mass of the E2-UbBD, E3 and LRR domains as a function of the \ndistances between the E2 -UbBD and LRR  centers. Relative positions of the same \ndomains in our ΔNSspH1 crystal structure, as well as the available structures of \n∆NSspH2, IpaH3 and IpaH9.8  are also repo rted. ΔNSspH1 ensembles conformers \npossess significant interdomain flexibility, implying that the protein can adopt both \nextended (open) and more compact  (closed) conformations (Figure 5C). Structure \nenvelope maximum dimension s (Supplementary Figure 7C) and radii of gyration \n(Supplementary Figure 7D) for the crystal structures of ΔNSspH1, ΔNSspH2, \n∆NIpaH9.8 and ∆NIpaH3 all fall within overall ensemble  distributions for these \nparameters. Notably, the ensemble best agreeing  with the experimental scattering \ndata (lowest χ2 value) comprises two extended structures and a closed conformation, \nwherein the  substrate binding interface on the  LRR domain is oriented toward the \ncatalytic loop (Figure 5D). \nIn conclusion , our SAXS data reveal that SspH1 exhibits significant interdomain \nflexibility suggesting that available crystal structures likely represent specific \nconformational snapshots within a continuous structural landscape . While PKN1 \nbinding does not fundamentally alter SspH1 overall flexibility, we observed a subtle \nshift in the Kratky profile maximum towards smaller q*Rg for both the FL protein and \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 13 \nΔNSspH1 upon PKN1 binding (Figures 5A and B). This suggests that PKN1 binding \nmight modulate the distribution of SspH1 closed and open conformations resulting in \na small increase in globularity of the  average solution structure of the complex \ncompared to the apo states. \nDynamic properties of the SspH1-ubiquitin conjugate  \n \nWe sought to better understand the structural flexibility of ubiquitin when loaded to \nSspH1 alone (SspH1~Ub) and in complex with  PKN1. Our SAXS analysis indicated \nsubtle changes in SspH1 conformation upon PKN1 binding . Given the low efficiency \nof UbVS labelling ( Figure 1F), which we used to form a stable SspH1 -Ub covalent \nconjugate, we employed NMR to study how E3 conjugation affects ubiquitin dynamic \nbehaviour.  \nWe first examined the non-covalent interaction of SspH1 with 15N-labelled ubiquitin. \nby monitoring the effect of SspH1 addition on the backbone amide resonances in the \n1H-15N HSQC spectrum of an 15N-labelled ubiquitin sample (Figure 6). Fast exchange \nin a subset of ubiquitin backbone amide resonances was observed upon SspH1 \naddition (Figure 6A). The perturbed resonances are located at the ubiquitin N- and C-\nterminal residues  and amide protons  around the ubiquitin I44 hydrophobic patch \n(Figure 6D and G). Incremental line broadening was also observed for the perturbed  \nresonances in the ubiquitin spectrum, likely due to the increase in molecular size and \nlonger correlation times upon interaction with SspH1. Next, we probed the interaction \nof the ubiquitin with the individual SspH1 domains. Chemical shift perturbations \n(CSPs) were still visible but fewer resonances we re affected in the 1H-15N HSQC \nspectrum when the isolated NEL domain was added to 15N-labelled ubiquitin \ncompared to SspH1 (Figure 6B). The differences in the two interfaces are  located \nmainly in the ubiquitin C-terminal tail (Figure 6E and H ). Resonance line broadening \nis also less pronounced in the titration of 15N-ubiquitin with the NEL alone compared \nto SspH1 (Figure 6A and B). These observations point to a direct participation of the \nLRR domain in the SspH1/ubiquitin interaction. In fact, a titration of ubiquitin with \nSspH1 featuring a largely flexible and independent LRR domain would have shown \nthe same profile of CSPs and line broadening observed for the titration of ubiquitin \nwith the isolate NEL domain. Addition of the LRR domain alone to 15N-labelled ubiquitin \nled t o a largely unperturbed 1H-15N HSQC spectrum, with some cross peaks \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 14 \nexperiencing small CSPs and associate resonance broadening (Figure 6C). A residue-\nspecific line-broadening analysis identified residues primarily located in the N- and C-\nterminal regions of ubiquitin experiencing larger effects upon interaction with the LRR \n(Figure 6F and I ). While this LRR/Ub interaction is weak, its surface appears to be \ncomplementary to the NEL/Ub interface, suggesting the two domains use different \nubiquitin surfaces for their interaction.  \nTo try and rationalize these observations we used AlphaFold3 (AF3) to model the \nstructure of ∆NSspH1 bound to Ub (Supplementary Figure 8A).38 In the predicted \nSspH1/Ub complex, the SspH1/Ub interface aligns with the cluster of ubiquitin \nresidues experiencing CSP s observed in the NMR titration  experiments with the \n∆NSspH1 and NEL domain. To investigate whether SspH1 uses the predicted surface \nto interact with  ubiquitin, we mutated a glutamate residue (E487) in SspH1 that \nappears to serve as a linchpin for binding, being packed between the side chains of \nR72 and R42 of  ubiquitin (Supplementary Figure 8A). We hypothesized that, if this \nwas the correct E3/Ub binding mode, E487 mutation to alanine would weaken ubiquitin \nbinding, while mutation to arginine would completely abrogate it. Whilst there was no \ndifference in the effect on the backbone amide resonances in the 1H-15N HSQC \nspectrum of a 15N-labelled ubiquitin sample upon addition of either WT or E487R \nSspH1 (Supplementary Figure 8B ), we observed disrupted discharge and substrate \nubiquitination activity for the E487A mutant, and its ablation for the E487R mutant \n(Supplementary Figure 8C and D). This indicates that whilst E487 has a role in SspH1 \nactivity, the AF3 predicted interface between the ubiquitin and SspH1 is likely not the \nprimary function of this residue . Instead, E487 interacts  with H498 which has \npreviously been shown to be important for the transfer of Ub from E3~Ub to the \nsubstrate (Supplementary Figure 8E).11 Interestingly, these two residues are located \non the helices either side of the catalytic loop containing C492 suggesting that their \ninteraction may stabilize its position to support catalysis. \nNext, we investigated the dynamic behavior of ubiquitin in the SspH1 ~Ub conjugate \nand how it is altered by PKN1 binding. We recorded 1H-15N HSQC spectra of SspH1 \npre-loaded with 15N-labelled ubiquitin UbVS probe and monitored the effect on \nubiquitin resonances in the SspH1 ~Ub conjugate when PKN1 is added (Figure 7). \nWhile the charging of SspH1 with 15N-ubiquitin probe is inefficient, this approach is \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 15 \nsuitable for NMR experiments as only the charged 15N-labelled ubiquitin in the \nSspH1~Ub conjugate is visible in the spectra. We first confirmed that the 1H-15N HSQC \nspectrum of the free 15N-ubiquitin UbVS probe remains largely indistinguishable from \nthat of the wild -type protein, except for a few C -terminal residues, implying that the \naddition of the vinyl sulfone warhead does not induce structural changes in ubiquitin. \nUpon conjugation to the SspH1 catalytic cysteine , most backbone resonances in the \n15N-labeled ubiquitin 1H-15N HSQC spectrum probe decrease in intensity with a subset \nexperiencing a stronger line broadening effect (Figure 7A). The remaining visible cross \npeaks (backbone and side chain amide resonances) in the 1H-15N-HSQC spectrum \nshow no change in their chemical shift s. A possible interpretation of this observation \nis that in the SspH1~Ub conjugate, the ubiquitin exists in fast exchange between two \nstates: a predominant, fle xibly tethered state, visible by NMR,  and a less abundant, \nlocked state, invisible in the 1H-15N HSQC ubiquitin spectrum due to its high molecular \nweight (~70 kD a). This equilibrium  would explain  the observed pattern of line \nbroadening and the persistence of some visible peaks. It suggests that while ubiquitin \nis not permanently locked in a rigid conformation with SspH1, it does transiently \nbecome locked in a larger complex, possibly involving interactions with both the NEL \nand LRR domains.  Upon addition of PKN1 substrate  to the SspH1~Ub conjugate, \nthere is an overall increase in the ubiquitin cross peaks intensity and a subset of \nubiquitin resonances , that were previously invisible or weak,  either reappear or \nincrease in intensity (Figure 7B). This indicates a shift in the fast exchange equilibrium \nbetween the locked  and flexible ubiquitin states towards a more dynamic ubiquitin.  \nThis shift may result from the loss of one or more interaction surfaces between \nubiquitin and SspH1 upon PKN1 binding. Intriguingly, the resonances that showed \nincreased intensity cluster to a patch on the ubiquitin molecule, encompassing the I44 \nhydrophobic surface, that overlaps with the region found to interact with SspH1 in our \nNMR results with free ubiquitin (Figure 6G and H).  \nTaken together, our NMR experiments show that ubiquitin non-covalently interacts with \nboth the NEL and LRR domains of SspH1. Our results with tethered ubiquitin, which \nmimics the activated E3~Ub intermediate, suggest that, in the SspH1~Ub conjugate, \nthe LRR domain restricts ubiquitin dynamic behaviour and substrate binding impacts \nubiquitin flexibility in the SspH1~Ub/PKN1 complex.  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 16 \n \nDISCUSSION \n \nBacterial NEL E3 ligases have an important role in Salmonella and Shigella evasion \nof host cell immune responses during infection, however the exact molecular \nmechanism of ubiquitin transfer and regulation of activity remains unclear. In this study, \nwe performed a cross -species comparison of NEL catalytic activity. We show that \nactive site cysteines of full length IpaH and SspH proteins are readily accessible for \nubiquitin loading but are immediately discharged, confir ming that autoinhibition is not \ndue to steric blockage of the catalytic cysteine, and supporting the model that water is \nthe favored nucleophile for ubiquitin discharge in the absence of substrate. 30 This \nactivity is advantageous to the bacteria, since it prevent s auto-ubiquitination and \nproteasomal degradation of the NELs themselves, and unanchored ubiquitin chain \nformation, which could trigger the host cell innate immune response. However, we also \nidentified differences in activity between Salmonella and Shigella NELs. To better \nunderstand these differences, we solved the crystal structure of a SspH1 construct \ncontaining both  LRR and NEL domains. When compared to previously published \nstructures for SspH2, IpaH9.8 and IpaH3, SspH1 is most similar in orientation to \nIpaH9.8 and IpaH3, with an ‘open’ conformation.  \nPrevious work had shown that NEL family members rely on significant flexibility within \nthe NEL domain to recruit ubiquitin.11 While available crystal structures of NEL proteins \nand substrate complexes provide snapshots of specific states, little is known about \nhow ubiquitin loading affects conformational dynamics  of the ligase , how substrate \nbinding affects the overall protein dynamics of NELs and how ubiquitin is passed onto \nthe substrate. Here, we have characterized the solution state behavio ur of SspH1 by \nSAXS and show that it exists in a conformation al continuum , which include the \norientations captured in the crystal structures of SspH2, IpaH9.8 and IpaH3. \nFurthermore, our SAXS analysis shows that apo SspH1 has a significant level of \nflexibility around the hinge between the LRR and NEL domains, which is not \nsubstantially affected by  substrate binding , though  it shifts the equilibrium towards \nmore globular conformations of SspH1. NMR experiments with free 15N-labelled \nubiquitin and a 15N-labelled ubiquitin probe loaded onto the catalytic cysteine of SspH1 \nenabled us to study the environment of ubiquitin in the presence of SspH1 and PKN1. \nThese experiments suggest that ubiquitin is partially restrained when non -covalently \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 17 \nor covalently bound to SspH1, forming transient interactions with the NEL domain and \npossibly weaker interactions with the LRR domain. The interactions with the LRR \ndomain are relieved upon PKN1 binding. \nThe crystal structure of the SspH1 construct containing both the LRR and NEL  \ndomains allowed comparison with the previously reported substrate -bound LRR \ndomain structure, to identify potential changes induced upon substrate binding, as \npreviously described for IpaH9.8. Interestingly, no significant structural changes are \ninduced by the substrate, indicating that the mechanism of release of autoinhibition \nsuggested for IpaH9.8 must be protein specific and not generally applicable to this \nprotein family. Instead, we identified E487 in the NEL as important for activity. This \nresidue interacts with H498 from the adjacent helix which has previously been reported \nto be important for ubiquitin transfer from E3 onto the substrate and we speculate that \ninteraction between these two residues might be important for stabilizing the catalytic \nloop. \nTaken together with previously published studies, the work presented here suggests \nthat substrate binding to the LRR domain of NEL proteins is not sufficient to induce the \nmajor conformational changes required to  bring the substrate close to the E3~Ub \nthioester intermediate to enable ubiquitin transfer. At present it is unknown what \nadditional events are necessary to promote formation of a ubiquitin transfer competent \nconformation of a substrate-NEL~Ub complex and further studies are required to trap \nmembers of this enigmatic E3 ligase family in their active state. \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 18 \nREFERENCES \n1. Nemhauser, J. Travel-Associated Infections & Diseases. CDC Yellow Book 2024 \n243–548 (2023) doi:10.1093/OSO/9780197570944.003.0005. \n2. Pillay, T. D. et al. Speaking the host language: how Salmonella effector proteins \nmanipulate the host. Microbiology (United Kingdom) 169, 001342 (2023). \n3. Nasser, A., Mosadegh, M., Azimi, T. & Shariati, A. Molecular mechanisms of \nShigella effector proteins: a common pathogen among diarrheic pediatric \npopulation. Molecular and Cellular Pediatrics 2022 9:1 9, 1–21 (2022). \n4. Komander, D. & Rape, M. The Ubiquitin Code. Annu Rev Biochem 81, 203–229 \n(2012). \n5. Berglund, J., Gjondrekaj, R., Verney, E., Maupin -Furlow, J. A. & Edelmann, M. \nJ. Modification of the host ubiquitome by bacterial enzymes. Microbiol Res 235, \n(2020). \n6. Roberts, C. G., Franklin, T. G. & Pruneda, J. N. Ubiquitin -targeted bacterial \neffectors: rule breakers of the ubiquitin system. EMBO J 42, (2023). \n7. Maculins, T., Fiskin, E., Bhogaraju, S. & Dikic, I. Bacteria -host relationship: \nubiquitin ligases as weapons of invasion. Cell Res 26, 499–510 (2016). \n8. Singer, A. U. et al. Structure of the Shigella T3SS effector IpaH defines a new \nclass of E3 ubiquitin ligases. Nature Structural & Molecular Biology 2008 15:12 \n15, 1293–1301 (2008). \n9. Zhu, Y. et al. Structure of a Shigella effector reveals a new class of ubiquitin \nligases. Nature Structural & Molecular Biology 2008 15:12  15, 1302 –1308 \n(2008). \n10. Bernal-Bayard, A. ; et al.  The NEL Family of Bacterial E3 Ubiquitin Ligases. \nInternational Journal of Molecular Sciences 2022, Vol. 23, Page 7725 23, 7725 \n(2022). \n11. Cook, M. et al. The ubiquitin ligase SspH1 from Salmonella uses a modular and \ndynamic E3 domain to catalyze substrate ubiquitylation. Journal of Biological \nChemistry 294, 783–793 (2019). \n12. Piro, A. S. et al. Detection of cytosolic shigella flexneri via a C -terminal triple-\narginine motif of GBP1 inhibits actin-based motility. mBio 8, (2017). \n13. Wandel, M. P. et al. GBPs Inhibit Motility of Shigella flexneri but Are Targeted for \nDegradation by the Bacterial Ubiquitin Ligase IpaH9.8. Cell Host Microbe  22, \n507-518.e5 (2017). \n14. Li, P. et al. Ubiquitination and degradation of GBPs by a Shigella effector to \nsuppress host defence. Nature 2017 551:7680 551, 378–383 (2017). \n15. Rohde, J. R., Breitkreutz, A., Chenal, A., Sansonetti, P. J. & Parsot, C. Type III \nSecretion Effectors of the IpaH Family Are E3 Ubiquitin Ligases. Cell Host \nMicrobe 1, 77–83 (2007). \n16. Ashida, H. et al. A bacterial E3 ubiquitin ligase IpaH9.8 targets NEMO/IKK γ to \ndampen the host NF -κB-mediated inflammatory response. Nature Cell Biology \n2010 12:1 12, 66–73 (2009). \n17. De Jong, M. F., Liu, Z., Chen, D. & Alto, N. M. Shigella flexneri suppresses NF-\nκB activation by inhibiting linear ubiquitin chain ligation. Nature Microbiology \n2016 1:7 1, 1–11 (2016). \n18. Hansen, J. M. et al.  Pathogenic ubiquitination of GSDMB inhibits NK cell \nbactericidal functions. Cell 184, 3178-3191.e18 (2021). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 19 \n19. Haraga, A. & Miller, S. I. A Salmonella enterica serovar Typhimurium \ntranslocated leucine-rich repeat effector protein inhibits NF-κB-dependent gene \nexpression. Infect Immun 71, 4052–4058 (2003). \n20. Haraga, A. & Miller, S. I. A Salmonella type III secretion effector interacts with \nthe mammalian serine/threonine protein kinase PKN1. Cell Microbiol 8, 837–\n846 (2006). \n21. Bhavsar, A. P. et al. The Salmonella Type III Effector SspH2 Specifically Exploits \nthe NLR Co-chaperone Activity of SGT1 to Subvert Immunity. PLoS Pathog 9, \n(2013). \n22. Bernal-Bayard, J. & Ramos -Morales, F. Salmonella Type III Secretion Effector \nSlrP Is an E3 Ubiquitin Ligase for Mammalian Thioredoxin. Journal of Biological \nChemistry 284, 27587–27595 (2009). \n23. Rahighi, S. et al. Specific Recognition of Linear Ubiquitin Chains by NEMO Is \nImportant for NF-κB Activation. Cell 136, 1098–1109 (2009). \n24. Xia, Z. P. et al. Direct activation of protein kinases by unanchored polyubiquitin \nchains. Nature 2009 461:7260 461, 114–119 (2009). \n25. Skaug, B. et al. Direct, Noncatalytic Mechanism of IKK Inhibition by A20. Mol \nCell 44, 559–571 (2011). \n26. Catici, D. A. M., Horne, J. E., Cooper, G. E. & Pudney, C. R. Polyubiquitin Drives \nthe Molecular Interactions of the NF -κB Essential Modulator (NEMO) by \nAllosteric Regulation. Journal of Biological Chemistry 290, 14130–14139 (2015). \n27. Keszei, A. F. A. et al. Structure of an SspH1-PKN1 Complex Reveals the Basis \nfor Host Substrate Recognition and Mechanism of Activation for a Bacterial E3 \nUbiquitin Ligase. Mol Cell Biol 34, 362–373 (2014). \n28. Quezada, C. M., Hicks, S. W., Galán, J. E. & Erec Stebbins, C. A family of \nSalmonella virulence factors functions as a distinct class of autoregulated E3 \nubiquitin ligases. Proc Natl Acad Sci U S A 106, 4864 (2009). \n29. Ye, Y., Xiong, Y. & Huang, H. Substrate -binding destabilizes the hydrophobic \ncluster to relieve the autoinhibition of bacterial ubiquitin ligase IpaH9.8. \nCommunications Biology 2020 3:1 3, 1–14 (2020). \n30. Keszei, A. F. A. & Sicheri, F. Mechanism of catalysis, E2 recognition,& \nautoinhibition for the IpaH family of bacterial E3 ubiquitin ligases. Proc Natl Acad \nSci U S A 114, 1311–1316 (2017). \n31. Madeira, F. et al. Search and sequence analysis tools services from EMBL-EBI \nin 2022. Nucleic Acids Res 50, W276–W279 (2022). \n32. Jumper, J. et al.  Highly accurate protein structure prediction with AlphaFold. \nNature 2021 596:7873 596, 583–589 (2021). \n33. Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the \nstructural coverage of protein -sequence space with high -accuracy models. \nNucleic Acids Res 50, D439–D444 (2022). \n34. Varadi, M. et al.  AlphaFold Protein Structure Database in 2024: providing \nstructure coverage for over 214 million protein sequences. Nucleic Acids Res \n52, D368–D375 (2024). \n35. Zouhir, S. et al.  The structure of the Slrp –Trx1 complex sheds light on the \nautoinhibition mechanism of the type  III secretion system effectors of the NEL \nfamily. Biochemical Journal 464, 135–144 (2014). \n36. Chou, Y. C., Keszei, A. F. A., Rohde, J. R., Tyers, M. & Sicheri, F. Conserved \nStructural Mechanisms for Autoinhibition in IpaH Ubiquitin Ligases. Journal of \nBiological Chemistry 287, 268–275 (2012). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 20 \n37. Schwieters, C. D. & Clore, G. M. Using small angle solution scattering data in \nXplor-NIH structure calculations. Prog Nucl Magn Reson Spectrosc  80, 1–11 \n(2014). \n38. Evans, R. et al.  Protein complex prediction with AlphaFold -Multimer. bioRxiv \n2021.10.04.463034 (2022) doi:10.1101/2021.10.04.463034. \n39. Winter, G. et al. DIALS as a toolkit. Protein Science 31, 232–250 (2022). \n40. Evans, P. R. & Murshudov, G. N. How good are my data and what is the \nresolution? Acta Crystallogr D Biol Crystallogr 69, 1204–1214 (2013). \n41. Adams, P. D. et al.  PHENIX: a comprehensive Python -based system for \nmacromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66, 213–\n221 (2010). \n42. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta \nCrystallogr D Biol Crystallogr 60, 2126–2132 (2004). \n43. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular \nstructures by the maximum -likelihood method. Acta Crystallogr D Biol \nCrystallogr 53, 240–255 (1997). \n44. Franke, D. et al. ATSAS 2.8: a comprehensive data analysis suite for small-angle \nscattering from macromolecular solutions. J Appl Crystallogr  50, 1212 –1225 \n(2017). \n45. Rambo, R. P. & Tainer, J. A. Characterizing flexible and intrinsically unstructured \nbiological macromolecules by SAS using the Porod-Debye law. Biopolymers 95, \n559–571 (2011). \n46. Schwieters, C. D. & Clore, G. M. Using Small Angle Solution Scattering Data in \nXplor-NIH Structure Calculations. Prog Nucl Magn Reson Spectrosc 0, 1 (2014). \n47. Delaglio, F. et al.  NMRPipe: A multidimensional spectral processing system \nbased on UNIX pipes. J Biomol NMR 6, 277–293 (1995). \n48. Vranken, W. F. et al. The CCPN data model for NMR spectroscopy: development \nof a software pipeline. Proteins 59, 687–696 (2005). \n49. Wilkinson, K. D., Gan -Erdene, T. & Kolli, N. Derivitization of the C -Terminus of \nUbiquitin and Ubiquitin-like Proteins Using Intein Chemistry: Methods and Uses. \nMethods Enzymol 399, 37–51 (2005). \n50. Burger, V. M., Arenas, D. J. & Stultz, C. M. A Structure -free Method for \nQuantifying Conformational Flexibility in proteins. Scientific Reports 2016 6:1 6, \n1–9 (2016). \n  \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 21 \nAcknowledgements \nThe authors would like to thank Andrew Purkiss for crystallography data collection , \nAurelien Thureau for SAXS beamline support, and Ian Taylor for help with SEC-MALLS \nexperiments. X-ray crystallography data collection was performed on the i24 beamline \nat Diamond; SAXS data collection on the SWING beamline at Soleil synchrotrons. This \nwork was supported by the Francis Crick Institute, which receives its core funding from \nCancer Research United Kingdom (CC 2075), the United Kingdom Medical Research \nCouncil (CC 2075), and the Wellcome Trust (CC 2075); by the Biotechnology and \nBiological Research Council, BB/T014547/1 to K.R. and D.H.; and by the Engineering \nand Physical Sciences Research Council, EP/V038028/1 to D.H. and K.R. For the \npurpose of open access, the author has applied a CC BY public copyright license to \nany author-accepted manuscript version arising from this submission. Figures were \ncreated in BioRender.com under the institutional license belonging to the Francis Crick \nInstitute. \nConflict of interest \nThe authors declare that they have no conflicts of interest with the contents of this \narticle. \nAuthor contributions \nC.R.K and D.E.: experiments, data analysis, visualisation, writing original draft, review \n& editing. D.H. and K.R.: conceptualisation, data analysis, funding acquisition, \nsupervision, writing – review & editing. \nMethods \nProtein Expression and Purification \nAll proteins were expressed in pET49b vectors containing a 3C cleavage site and His-\ntag in BL21 Gold E.Coli in LB media. Cells were grown at 37 C until OD600 reached \n0.6-0.8, and protein expression induced with 0.5 mM IPTG overnight at 20 C. Cells \nwere lysed by sonication in 50 mM HEPES pH 7.5, 150 mM NaCl, 20 mM imidazole, \n0.5 mM TCEP in the presence of proteas e inhibitor. Proteins were purified on nickel \nNTA resin ( Thermo Scientific, #88222 ), followed by elution and treatment with 3C \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 22 \nprotease overnight. Proteins were then purified by gel filtration into 50 mM HEPES pH \n7.5, 150 mM NaCl, 0.5 mM TCEP. The following constructs were used in activity \nassays:  \nFull length constructs: IpaH1.4 1-575, IpaH9.8 1-545, SspH1 1-700, SspH2 1-788. \n∆N constructs: IpaH1.4 31-575, IpaH9.8 21-545, SspH1 161-700, SspH2 166-788. \nNEL domain constructs: IpaH1.4 283-575, IpaH9.8 252-545, SspH1 405-700, SspH2 \n493-788. \n \nSubstrate ubiquitination assays \nNEL E3 ligase constructs of IpaH1.4, IpaH9.8 and SspH1 (1 µM) were incubated at 22 \ndegrees with shaking with their respective substrates LUBAC (C82A, 1 µM), GBP1 (1 \nµM) and PKN1 (residues 122-199, 2 µM) in the presence of 0.1 µM E1, 2 µM UbcH5A, \n20 µM ubiquitin and 10 mM ATP in 25 mM HEPES pH 7.5, 150 mM NaCl, 10 mM \nMgCl2 for 0-30 minutes. Samples were quenched with LDS loading dye containing \nDTT and run by SDS-PAGE on 4-12% gels. \nAuto-ubiquitination Assays \nNEL E3 ligase  constructs of IpaH1.4, IpaH9.8 , SspH1  and SspH 2 (1 µM) were \nincubated at 22 degrees with shaking in the presence of 0.1 µM E1, 2 µM UbcH5A, 20 \nµM ubiquitin and 10 mM ATP in 25 mM HEPES pH 7.5, 150 mM NaCl, 10 mM MgCl2 \nfor 0-30 minutes. Samples were quenched with LDS loading dye containing DTT and \nrun by SDS-PAGE on 4-12% gels. \nDischarge Assays \nNEL E3 ligase constructs of IpaH1.4, IpaH9.8 , SspH1  and SspH 2 (100 nM) were \nincubated at 22 degrees with shaking in the presence of 1 µM  precharged UbcH5A-\nUb-cy3 for 0-20 minutes. Samples were quenched with LDS loading dye and run by \nSDS-PAGE on 4-12% gels. \nUbVS loading assays \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 23 \nNEL E3 ligase constructs of IpaH1.4, IpaH9.8 , SspH1  and SspH 2 (5 µM) were \nincubated with 20 µM ubiquitin vinyl sulfone (UbVS, U-202B-050, BioTechne) at 22 \ndegrees with shaking for 0 -120 minutes. Samples were quenched with LDS loading \ndye containing DTT and run by SDS-PAGE on 4-12% gels. \nCrystallization, data collection, phasing and refinement \nΔNSspH1 (161-700) at a concentration of 12 mg/mL was crystallized in 50 mM HEPES \npH 7.5, 150 mM NaCl and 0.5 mM TCEP at 20 °C. Crystals grew from a sitting drop \nmade of 0.1 µl of protein solution mixed with 0.1 µl of 0.225 M sodium tartrate, 22% \nPEG 3350, 11 mM sarcosine . The crystals were cryo -protected with 0.225M sodium \ntartrate, 35% PEG 3350, 10% glycerol  and data were collected at Diamond Light \nSource at the i24 beamline. Data processing was done using DIALS and merged and \nscaled using AIMLESS.39,40 The initial model was obtained by molecular replacement \nusing the available coordinates of the SspH1 LRR domain (4NKH) and the coordinates \nof the NEL domain from the SspH1 AlphaFold model as templates in Phenix Phaser.41  \nModels were iteratively improved by manual building in Coot and refined using both \nREFMAC and Phenix.41–43 Coordinates and structure factors are deposited in the PDB \nwith the code 9H6W and the final refinement statistics are reported in Supplementary \nTable 1. \nSmall-Angle Xray Scattering and SspH1 ensemble modelling. \nSAXS data were collected at the SWING beamline at SOLEIL (GIF -sur-YVETTE \nCEDEX, France). Samples of ΔNSspH1 and FL SspH1 at concentrations 168 µM and \n104 µM in 50 mM HEPES pH 7.5, 150 mM NaCl and 0.5 mM TCEP in the absence \nand presence of 1:2 molar equivalents of PKN1, were injected onto a Bio SEC-3 100 \nÅ Agilent column and eluted at a flow rate of 0.3 ml/min at 15 °C (Figure S8A). Frames \nwere collected continuously during the fractionation of the proteins (1 frame/sec). \nFrames collected before the void volume (0.6 – 1.5 ml) were averaged and subtracted \nfrom the signal of the elution profile to account for background scattering. Da ta \nreduction, subtraction, and averaging were performed using the software FOXTROT \n(SOLEIL). The scattering curves were analyzed using the package ATSAS ,44 and \nreported as function of the angular momentum transfer q = 4π/λ sinθ, where 2θ is the \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 24 \nscattering angle and λ the wavelength of the incident beam. Values of the cross -\nsectional radius of gyration were obtained with SCATTER.45  \nTo model the dynamic behavior of SspH1 and explore its conformational space, we \nemployed an Xplor-NIH protocol.46 Initially, the E2-UbBD (residues 623-697) and the \nLRR (residues 161-393) domains were randomly oriented relative to a fixed E3 domain \n(residues 405-622) by randomizing linker torsion angles (note: the LRR-E3 linker has \na high B-factor). Subsequently, a high-temperature torsion-angle dynamic at 3000 K \nwas followed by simulated annealing from 3000 K to 25 K in 12.5 K increments. The \nfinal step involved gradient minimization in torsion -angle space to generate the \nstructure ensembles. Our calculations incorpor ated the experimental SAXS -derived \nforce field, knowledge -based energy terms, such as torsion angle potential from \nconformational databases, and standard Xplor -NIH covalent and nonbonded energy \nterms. Each ensemble member was assigned a weight corresponding to 1/n, where n \nis the number of structures in the ensemble. Fifty equidistant  points were used to \nsimulate the scattering curve in a q interval of 0-0.3. \nTo avoid overfitting and determine the minimum number of conformers needed to \naccurately replicate the experimental SAXS data, we initially computed 100 ensembles \nwith 1-6 conformers each (Figure S8B). The average χ2 rapidly decreased, plateauing \nfor ensembles with three or more conformations. Subsequently, we calculated 2000 3-\nconformer ensembles and selected those with χ2 ≤ 1.5 for analysis (significance level \nα = 0.05, ensuring statistical significance in the association between the experimental \ndata and our models). Distances between domain centers of mass and intra-domain \nangles were calculated using custom Python scripts. Envelope diameters and radii of \ngyration for each conformer were evaluated with the program Crysol (same reference \nof ATSAS).  \nNuclear Magnetic Resonance \nNMR experiments were carried out using a Bruker AVANCE spectrometer operating \nat a proton nominal frequency of 800 MHz. Data were acquired with Topspin (Bruker), \nprocessed with NMRPipe,47 and analyzed by CCPNMR.48  \nFor conjugated ubiquitin NMR, 15N isotope enriched UbMESNa was expressed and \npurified in M9 minimal medium using 1g/L of 15N-ammonium chloride as sole source \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 25 \nof nitrogen. 15N UbMESNa  was further reacted with glycine vinyl sulfo ne (EN300 -\n1264982, En amine) as described previously ,49 to give 15N-ubiquitin vinyl sulfone. \nSspH1 constructs (10 µM) were reacted with 50 µM 15N UbVS for 2 hours before \npurification by gel filtration into 50 mM HEPES pH 7.5, 75 mM NaCl, 0.5 mM TCEP. \nFor ubiquitin titration NMR experiments, 15N isotope enriched ubiquitin was prepared \nby growing the bacteria in M9 minimal medium using 1g/L of 15N-ammonium chloride \nas sole source of nitrogen. 15N-ubiquitin was titrated with SspH1 constructs and \nmeasurements taken as described above. Chemical shifts changes for the backbone \namide proton and nitrogen nuclei ( ΔδNH) were calculated according to a procedure \nimplemented in CCPNMR analysis.48  \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 26 \nFIGURES \n \nFigure 1 ∆N NEL activity assays \nA) Diagram of constructs of NEL bacterial E3 ligases IpaH1.4, IpaH9.8, SspH1 and SspH2; B) \nSubstrate ubiquitination assay with ∆N constructs of IpaH1.4 (substrate LUBAC), IpaH9.8 \n(substrate GBP1) and SspH1 (substrate HR1b PKN1). Performed with 0.1 µM UBA1 (E1), 2 \nµM UbcH5A (E2), 1.0 µM ∆N NEL (E3), 20 µM ubiquitin, 10 mM ATP at RT for 30 minutes with \neither 1 µM (LUBAC, GBP1) or 2 µM  (HR1b PKN1) substrate. C) Auto -ubiquitination assay \nwith ∆N constructs of IpaH1.4, IpaH9.8, SspH1 and SspH2. Performed with 0.1 µM UBA1 \n(E1), 2 µM UbcH5A (E2), 1 µM ∆N NEL (E3), 20 µM ubiquitin, 10 mM ATP at RT for 30 minutes.  \nD) E2~Ub discharge assay with ∆N IpaH1.4 and IpaH9.8. Performed with 1 µM UbcH5A~Ub-\ncy3 (E2~Ubcy3), 5 nM NEL (E3) at RT for 0 -30 minutes. E) E2~Ub discharge assay with ∆N \nSspH1 and SspH2. Performed with 1 µM UbcH5A~Ub-cy3 (E2~Ubcy3), 50 nM NEL (E3) at RT \nfor 0-30 minutes. Assays for full length constructs are in Supplementary Figure 2. F) Ubiquitin-\nloading assay with ∆N constructs of IpaH1.4, IpaH9.8, SspH1 and SspH2. Performed with 20 \nµM UbVS, 5 µM ∆N NEL (E3) at RT for 2 hours. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 27 \n \nFigure 2  NEL domain activity assays  \nA) Auto-ubiquitination assay with NEL domains of IpaH1.4, IpaH9.8, SspH1 and SspH2. \nPerformed with 0.1 µM UBA1 (E1), 2 µM UbcH5A (E2), 1 µM NEL domains (E3), 20 µM \nubiquitin, 10 mM ATP at RT for 30 minutes. B) E2~Ub discharge assay with FL constructs of \nIpaH1.4, IpaH9.8, SspH1 and SspH2. Performed with 1 µM UbcH5A~Ub-cy3 (E2~Ubcy3), 0.2 \nµM NEL domain (E3) at RT for 0-30 minutes. C) Ubiquitin-loading assay with NEL domains of \nIpaH1.4, IpaH9.8, SspH1 and SspH2. Performed with 20 µM UbVS, 5 µM NEL domain (E3) at \nRT for 2 hours. \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 28 \n \nFigure 3 Crystal structure of ∆NSspH1 \nA) Overall structure from side and top view; B) Loop containing catalytic cysteine, with the \nCysteine side chain shown as spheres ; C) Overlay of ∆N NEL structures aligned to the NEL \ndomain showing SspH1, SlrP (PDB 3PUF), IpaH9.8 (PDB 6LOL) and IpaH3 (PDB 3CVR) in \nthe ‘open’ conformation, and SspH2 (PDB 3G06) in the ‘closed’ conformation; D) Aligned E2-\nUbBD thumbs for SSpH1 and SspH2, and SlrP, IpaH9.8 and IpaH1.4 (PDB 3CKD). \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 29 \n \n \n \nFigure 4 Interface between the LRR and NEL domains \nA) Interface between the LRR domain in green and the NEL domain in blue highlighting the \ninteractions made between R351 (LRR) and E493 and D522 (NEL). The side chain of the \ncatalytic cysteine is shown. \nB) Top: Auto ubiquitination assay with ∆NSspH1 mutants R351A, R375A, R351A R375A. \nPerformed with 0.1 µM UBA1 (E1), 2 µM UbcH5A (E2), 1.0 µM SspH1 (E3), 20 µM ubiquitin, \n10 mM ATP at RT for 0 -30 minutes. Bottom: Substrate ubiquitination assay with ∆NSspH1 \nmutants R351A, R375A, R351A R375A. Performed with 0.1 µM UBA1 (E1), 2 µM UbcH5A \n(E2), 1.0 µM SspH1 (E3), 2 µM HR1b PKN1, 20 µM ubiquitin, 10 mM ATP at RT for 0 -30 \nminutes.  \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 30 \n \n \nFigure 5 Small-Angle X-ray scattering analysis \nX-ray scattering intensities profiles, normalized pair distance distribution P(R) and Kratky plot \nof ΔSspH1 (A) of full-length SspH1 (B) in the absence and presence of PKN1. The dotted lines \nin the Kratky plot are drawn at qR g = Ö3 and (qR g)2I(q)/I(0) = 1.104. Folded and globular \nproteins have a maximum where the two lines intersect. 50 (C) LRR-NEL intra-domain angles \nand distances distribution for the structural ensembles calculated by Xplor-NIH that represent \nthe experimental Xray scattering intensities ( χ2 ≤ 1.5). ( D) The three -conformers structural \nensemble that best agrees with the experimental SAXS data (lowest χ2). \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 31 \n \nFigure 6 NMR analysis of the SspH1-ubiquitin interaction  \nDetails of the titration of 15N-labeled ubiquitin titrated with ∆N SspH1 (A) SspH1 NEL domain \n(B), and SspH1 LRR domain (C). Spectra at different ligand concentrations are plotted at the \nsame contour level. Chemical shifts perturbations versus residue number in the NMR \nspectrum of 15N-labeled ubiquitin in the presence of (D) ∆N SspH1 or (E) NEL domain with \nCSPs mapped onto ubiquitin structure (G & H). Ubiquitin 1H-15N HSQC line broadening versus \nresidue number for the titration with the LRR domain with the rate of broadening mapped onto \nubiquitin structure (F&I). \n \n \n \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint \n\n 32 \n \n \n \n \n \n \n \n \n \n \n \n \n \nFigure 7. NMR of conjugated15N-labelled ubiquitin. \nSpectra of 15N-Ub tethered to ∆NSspH1 C492 in the absence (A) and presence (B) of PKN1. \nResidues in the ubiquitin 1H-15N HSQC spectrum reappear upon addition of the substrate. The \nreappearing residues are plotted on the ubiquitin structure in (C).  \n \n \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted May 15, 2025. ; https://doi.org/10.1101/2025.05.12.653404doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}