A Chemical Biology Approach to Understanding Molecular Recognition of Lipid II by Nisin(1–12): Synthesis and NMR Ensemble Analysis of Nisin(1–12) and Analogues

Abstract Natural products that target lipid II, such as the lantibiotic nisin, are strategically important in the development of new antibacterial agents to combat the rise of antimicrobial resistance. Understanding the structural factors that govern the highly selective molecular recognition of lipid II by the N‐terminal region of nisin, nisin(1–12), is a crucial step in exploiting the potential of such compounds. In order to elucidate the relationships between amino acid sequence and conformation of this bicyclic peptide fragment, we have used solid‐phase peptide synthesis to prepare two novel analogues of nisin(1–12) in which the dehydro residues have been replaced. We have carried out an NMR ensemble analysis of one of these analogues and of the wild‐type nisin(1–12) peptide in order to compare the conformations of these two bicyclic peptides. Our analysis has shown the effects of residue mutation on ring conformation. We have also demonstrated that the individual rings of nisin(1–12) are pre‐organised to an extent for binding to the pyrophosphate group of lipid II, with a high degree of flexibility exhibited in the central amide bond joining the two rings.


Introduction
Antibiotic resistant infections are becominga ni ncreasing threat to global public health, [1] which has generated ar enewed interest in natural productsa sasourceo fp otent antimicrobial drugs. One such class of compounds is the lantibiotics:af amily of gene-encoded antimicrobial peptides which are extensively post-translationally modified. The lantibiotics have complexc yclic structuresg enerated by the thioether-bridged amino acids lanthionine (Lan) and methyllanthionine (MeLan), and frequently also contain the a,b-unsaturated amino acids dehydroalanine( Dha) and dehydrobutyrine (Dhb). [2,3] Nisin (Figure 1), the most commonly studied lantibiotic, is used commerciallya safood preservative, [4] andh as broad-spectrum activity against ar ange of Gram-positive organisms, including methicillin-resistant Staphylococcus aureus (MRSA). [5] The mechanism of action of the lantibiotics is mediated by high affinity binding to lipid II, ak ey intermediate in peptidoglycan biosynthesis. [6] In the case of nisin, this interaction results in the rapid formation of stable pores in the bacterial membrane, of which lipid II is an intrinsic component, at an 8:4n isin:lipid II ratio. [7] As econd effect of this binding is the inhibition of peptidoglycan biosynthesis, caused by the large-scale sequestration and aggregation of lipid II. [8,9] The importance of the interaction with lipid II in the antibacterial action of nisin has also been demonstrated in functional studies, in which it wass hown that nisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)i sa ble to antagonize the activity of WT nisin. [10] NMR studies have provedt ob eav aluable method in the study of lantibiotic conformation and lipid II binding. [11] Indeed, the solution conformation of nisin [12][13][14] and an umber of other lantibiotics, such as subtilin, [15] mutacin 1140 [16] and cinnamycin, [17] have been reported. The interaction between nisin and lipid II has also been investigated using NMR. Initially, studies were conducted with nisin in lipid II-doped micelles, revealing that the N-terminal of nisin is involved in target recognition and binding to the lipid. [18,19] An NMR study in DMSO with full length nisin and at runcated analogueo fl ipid II at a1 :1 ratio later revealed the natureo ft he interaction, which involves the binding of the nisin Aa nd Br ings (residues 1-12) in ac agelike formation around the pyrophosphate of lipid II (PDB ID 1WCO). [20] Isothermal calorimetry (ITC) andv esicle leakage studies have also confirmed the importance of the pyrophosphate group for nisin binding, and have demonstrated that the MurNAc moiety is required for high affinity interaction. [21] A similar binding mode has also been observed in NMR studies of the two-component lantibiotic lacticin3 147, in whicht he Cterminus of the A1 peptideforms acage around the lipid II pyrophosphate. [22] Recently,W eingarth et al. reported the solid state NMR of lipid II-bound nisin as part of the pore complex in DOPC liposomes and in native Micrococcusf lavus membranes. [23] Although this study confirmed the broad features of the 8:4p ore complex, the authors observed that the spectra under theseconditions differed drastically from that of the previously reported spectrao fl ipid II-bound nisin in DMSO, suggesting that nisin adopts ad ifferent conformation in the native pore. [20] In additiont os tructures of full-length lantibiotics, the conformations of the Aa nd Br ings of nisin have also been investigated. For example, Palmer et al. reported the solution structures of analogues of nisin ring Aa nd ring B, [24] which adopt conformations remarkablys imilart ot hose found in the full length wild-type (WT) peptide in aqueous solution. [14] Recently, we have also studied [25] the conformations of the individual A and Br ings of nisin and anotherc lass Il antibiotic, mutacin I. [26] Although potentially useful for determiningt ow hat extent the isolated lantibiotic rings are pre-organized for lipid II binding, one disadvantage of such studies is that they provide no insight into how each ring may affect the conformationo ft he other,o ro nt he relative orientation of the two rings. Given that flexibility in the nisin hinge region is essential for bioactivity, [6] andt hat largec hanges of torsion angle between lantibiotic rings are importanttoe nable mersacidin-lipid II binding, [27] it is perhaps surprising that ac onformational study of the entire Ring A-Ring Bs tructure, nisin(1-12), has not yet been conducted.
Another factor to consider,e specially in the interesto fd evelopingm ore stable antibiotics based on the structureo f nisin, [28][29][30][31] is the effect of residue mutation within lantibiotic binding rings on either solutionc onformation or antibacterial activity.S ignificant efforts have been directed towards understanding the effects of dehydro residue replacement,a st hese residues contribute to the metabolic instability of these pep-tides, however no clear picture has yet emerged. Palmer et al. have shown that substitution of Dha5 for Ala in nisin ring A leads to significant conformational change of the isolated ring A, [32] conversely,o ur NMR studies [25] comparingi solated ring A structures of nisina nd mutacin Iw ith saturated analogues of mutacin Ir ing Ai ndicated that the replacement of Dha5 by either Ser or Ala did not significantly affectt he overall conformation of the Leu4-Xaa5-Leu6 portion of ring A. This is supported by mutation studies. The observation that full length nisin bearing aD ha5Alam utation retains bioactivity against Micrococcus luteus [33] suggests that any conformational change causedb yd ehydro replacement does not affect the activity of nisin, and therefore that it may not interfere with lipid II binding. Similarly,W iedemann et al. have shown that the replacement of Dhb2 in full length nisin with either Ser,A la or Valh as little effecto nt he MIC. [6] Other groups have investigated the effect of (Me)Lan replacement. Slootweg et al. synthesized dicarba bridged analogues of nisin(1-12)b yR CM, finding that replacement of (Me)Lan with longer dicarba bridges was reasonably well tolerated. [34] As parto ft his work the authors also investigated the effect of replacing both Dha and Dhb with Ala, and found that the presence of the dehydro residues increased the affinity of the dicarbab ridged peptides for lipid II. Introduction of at hirdc yclic constraint in dicarbab ridged analogueso fn isin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12), by creating al actam bridge between the N-terminus and the Br ing, has also been investigated by Harmsen et al. [35] The resulting reduction in flexibility increased the affinity of the peptide for lipid II over the bicyclic dicarba analogue, but was still five-fold less active than WT nisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12).
Flexible molecules,s uch as nisin(1-12)a nd the analogues described above,e xist in an umber of rapidlye quilibrating conformations in solution. Experimental NMR variables, such as NOEs and J-couplings, are therefore averaged over the whole population of solution conformations,a nd as ingle average structure can be an inadequate representation of the true conformationsp resenti ns olution. [36] Ag ood methodf or the study of the conformations of flexible molecules is NAMFIS (NMR analysiso fm olecular flexibility in solution). [37] In the NAMFIS technique, the averaged NMR variables are deconvoluted by varying the molarf ractions of ac omputational theoretical ensemble, calculated by unrestrained Monte Carlo molecular me-  , until the best possible fit of the experimental NMR data is obtained. The result of this is an ensembleo fall conformations which are present in solution, and their probabilities, hence providing am ore completep ictureo ft he shape of the compound in solution than is possible by average structure calculation. The utility of NAMFIS analysisi nd etermining the solution conformation of small cyclic peptides has been demonstrated, [38][39][40] but has never previously been applied to lantibiotic systems.

Peptides ynthesis
Severald ifferenta pproaches to the chemical synthesis of lantibiotics have been reported over the past 20 years. [41,42] We have developed very effective solid-phase peptides ynthesis methodology which we and others have appliedtothe synthesis of individual rings of lantibiotics, [43,44] overlapping rings [45] and to the total synthesis of complete lantibiotics. [46][47][48][49] We have previously reported the solid-phase synthesis of the individual Aa nd Br ings of nisin and of the related lantibiotic, mutacin I, [25] and have investigated the conformationalp roperties of these isolated rings ands ynthetic analogues by NMR. We have now extended this work to prepares ynthetic analogues of WT nisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12), which can be compared with WT nisin(1-12) itself (1)( Figure 2). (Thr2, Ser5) analogue 2 was designed using the amino acids in the 2-and5 -positions that would be present in the biosynthetic precursor peptide, and that would undergo dehydration by the enzyme NisB in the producing organism. [50] Similarly,t oi nvestigate the effects of dehydro amino acidso nt he conformation of the bicyclic structure, (Abu2, Ala5) analogue 3 was designed with the saturated analogues of Dhb and Dha in positions 2a nd 5r espectively.W e envisagedt hat these chemically modified nisin analogues would be more stable than the parent nisin structure, and for analogue 2 we also expected that the Thr and Ser residues would also offer some improvement in aqueous solubility of the resulting peptide. In both analogues 2 and 3,w ea lso sub-stitutedL an for the naturally-occurring MeLani nr ing B, as our previouss tudies had indicatedt hat replacement of Lan for MeLan did not significantly change the backbonec onformation. [25] Our route to (Thr2, Ser5) analogue (2)startedfrom the previously reported [25] resin-bound cyclic ring Bp eptide 4.W ei nitially attempted to couple the next (Teoc, TMSE/Fmoc) Lan monomer 5 [45] to the free -NH 2 group of 4 to give resin-bound intermediate 6,a nd then to build up the linear precursor to ring A, using standard Fmoc SPPS coupling conditions (HOAt and PyAOP) (Scheme 1). However, only trace amounts of product with extensive impurities were obtained. Mass spectrometry analysis indicated that the initial couplingo ft he (Teoc, TMSE/Fmoc) Lan monomer 5 was unsuccessful under these conditions. We have previously found [51] that microwave conditions improve the coupling of orthogonallyp rotected lanthionines to resin-bound intermediates, and repeating the synthesis using microwavec oupling for the incorporation of 5 was successful. Subsequent addition of the remaininga mino acids in the sequence of ring A, removal of the Fmoc group and se-lectiveT eoc and TMSE deprotection gave the resin-bound intermediate 7a.H owever,a ll attempts to cyclise 7a on-resin, using standard coupling conditions, were unsuccessful. Mass spectrometry of intermediate peptides cleaved from the resin showedo nly trace amounts of peptide 8a had been formed (Supporting Information, Figure S1).
We hypothesized that the resin-bound intermediate 7a was folded on-resin in ac onformationw here the amino group of Ile4 was remote from the carboxylic acid group of Lan3. Te chniques to improve the synthesis of "difficult peptides" have been extensively researched. [52] Such peptides contain sequences which show ah igh tendency to foldo ra ggregate on-resin, maskingt he nascent amino group and resulting in low or failed peptidec oupling steps. Many of the approaches used to overcomet hesep roblems focus on preventing the formation of inter-or intra-chain hydrogen bonds by maskingt he amide NH. We reasoned that similara pproaches could be used to overcomet he failure of 7a to cyclise, by diminishing its ability to fold into an on-productive conformation.W efirst attempted to improvet he on-resin cyclisation by incorporation of aH mbprotected Leu residue (Scheme 1). Hmb-protected amino acids have previouslyb een used to improve the cyclisationofp entapeptides [53] and of larger lanthionine-containing rings. [51] Althoughw ew ere ablet os uccessfully incorporate Fmoc-(Hmb)Leu-OH to give resin-bound intermediate 7b,c hain extension and on-resin cyclisation did not give the required 8b.
Another approacht ot his problem is the use of pseudoprolines, in which threonine, serine (or cysteine)-derived dipeptides are protected with proline-like oxazolidines (or thioazolidines).M any of these dipeptides are commerciallya vailable and can be incorporated directly into peptides ynthesis protocols. The incorporationo fp seudoproline residues into linear sequences has been reported to improvet he head-to-tail cyclisation of both short [54] and longer [55] peptides. This effect was attributed to the observation that such residues induce ap redominantly cisoid conformation about the amide bond adjacent to the modified amino acid, resulting in the temporary inductiono fab-turn.T he presence of aS er residue in ring A made this an attractive approach to attempt to pre-organise the peptide for ring closure.
We therefore synthesised the resin-bound intermediate 9,incorporating the commercially available dipeptide Fmoc-Ile-Ser[y(Me,Me)pro]-OH 10 using standard coupling conditions (Scheme 2). Pleasingly,i tw as then possible to cyclise 9 to give 11,u sing microwave couplingc onditions. The peptides equencew as completed by the coupling of the two N-terminal residues,F moc-Thr(tBu)-OH and Fmoc-Ile-OH. Unexpectedly, cleavage from the resin with TFAd id not result in acidolysis of the pseudoproline, and the partially deprotected 12 was recovered. Other groups [56][57][58] have also reported that these prolinelike oxazolidines and thioazolidines were resistantt od eprotection with TFA. Subsequentt reatment of 12 with TMFSA [56] at 0 8Cg ave the desired (Thr2, Ser5) analogue 2 in 3% overall yield after purification.
The synthesis of (Abu2, Ala5) analogue 3 was also carried out from the resin-bound intermediate 6 (Scheme 3). The three ring Ar esidues, Leu, Ala and Ile were added by standard SPPS methods and the Te oc, TMSE andF moc groups removed to give 13.C yclisation was carried out on-resin to give the resinboundb icyclic peptide 14.T his was followed by chain extension with Fmoc-Abu-OH and Fmoc-Ile-OH, and cleavage from the resin, to give analogue 3.U sing this synthetic protocol, the purity of the bicyclic analogue was poor and after extensive purification the peptide was isolated in 0.5 %y ield, insufficient to allow full structural assignment. As with analogue 2,t he poor yield and purity may also be attributable to the failure of the resin-bound intermediate 13 to fold into ac onformation where cyclisation is possible. Unfortunately,apseudo-proline approachisn ot possible with the three ring Ar esidues present in analogue 3.W ea ttempted to improvet he yield of 3 by incorporation of either Fmoc(Hmb)Leu-OH or Fmoc(Hmb)Ala-OH, as appropriate, but none of the desired bicyclic peptidec ould be isolated using this approach.
Digestion of commercially availablen isin to give WT nisin(1-12) 1 has been extensively described in the literature, and has been used by an umber of groups to produce fragments of the WT peptidef or variousa pplications. As commerciallya vailable nisin from L. lactis containsonly approximately 2.5 %n isin, prior enrichment and removal of salts and denatured milk solids is required. We initially followed the enrichmentm ethod reported by Slootweg et al., [59] giving pure 1 for structural analysis. However,t he modified digestion procedure described by Koopmans et al., [28] in which higher concentrationso fn isin and trypsin were used, was found to requires horter reaction times, hence decreasing the risk of lanthionine oxidation caused by extended periodsofi ncubation in buffer.
Theoretical ensembles for both peptides were generated using Monte Carlo conformational searches, using two different force fields, followed by molecular mechanics minimization (MCMM) ( Ta bleS 3). The lipid II-bound conformation available in the PDB (PDB ID 1WCO) [20,61] was added to the ensemble for nisin(1-12)( 1). By deconvolution of the population averaged experimentalc onstraintsi nto the probability-weighted sum of the back-calculated constraints from the computational predicted ensembles, we estimated the molar fraction of each theoreticalc onformer presenti ns olution using the NAMFIS algorithm (Tables S5, S6). The ensemble analyses were validated, followingt he previous literature, [39,40] through evaluation of the reliability of the conformational restraints by the addition of 10 %r andom noise to the experimental data as well as by the random removal of individual restraints.
These conformationsa re also the highest molar fraction, with ac ombined population of 49 %. Among the conformations of each peptides elected by NAMFIS analysis, there is a high degree of similarity within each of the rings, that is, either ring Ao rr ing Bc an be aligned with low RMSD. However,there is ah igh degree of flexibility around the centrala mide bond between the two rings, leadingt oaset of structures with diverse overall backbone conformation and high globalR MSD (Table 1).
In the case of nisin(1-12) (1)t here is al arge range of rotation, enabling the peptidet oa dopt the lipid II-bound conformation, whereas in the (Thr2, Ser5) analogue (2), each conformation hast he same overall fold (relativet or ing A, the Br ings all rotate in the same direction) andd on ot adopt lipid II-bound conformations( Figure 5). Examination of the (Thr2, Ser5) (2)a naloguec onformations revealed that this difference    Figure 6A). This hydrogen bond appears to fix the first two residues over one face of ring A, possibly hindering ring Bf rom adopting the lipid II-binding conformation. This hypothesis is supported by the observation that as imilar' blocking" of one face of ring Ai sa lso observed in half of the nisin(1-12)( 1)c onformations (conformations 1, 2, and 4), caused by ah ydrogen bond between Ile1 CO and Ile4 NH, resulting in an overall fold similar to that adopted by the (Thr2, Ser5) analogue (2)c onformations( Figure 6B). An alternative, or perhaps complementary,e xplanation for the tendency of the rings to fold as observed in the (Thr2, Ser5) analogue (2)c onformations is that one or more hydro-gen bond(s) can be formed between ring Aa nd ring Bw hich stabiliset his orientation. All of the conformationso fb oth nisin(1-12) (1)a nd (Thr2, Ser5) analogue (2)w hich do not exhibit al ipid II-binding fold contain at least one such hydrogen bond, with the most commonlyf ormed bond being between the carbonyl of residue5andt he NH of either residue 8o r1 0 ( Figure 7).
This indicates that the (Thr2, Ser5) analogue (2)m ay be less likely to adopt the lipid II-bound conformation in solution, thought he fact that full length nisin Dha5Ser mutants have been shown to maintain MIC values comparable to the native peptidei ndicates that 2 should still be an effective lipid II binder. [6] Interestingly,four of the (Thr2, Ser5) analogue (2)conformations contain a cisPro (conformations1 ,2 ,3 ,a nd 5, total 71 %), compared to only one conformation in nisin(1-12) (1) (conformation1 ,1 1%). Indeed, the short experimental distance between ProH a and Lan8 Ha in the (Thr2, Ser5) analogue ( 2) suggested that a cisPro was likely to be present in at least one of the selected conformations.

Assessingthe conformational effect of separating the Aa nd Brings
Previously,w eh ave reported the synthesis and average solution phase conformations( calculated in XPLOR-NIH [62,63] )o ft he individual lipid II-binding rings of nisin and mutacin I. [25] We therefore sought to compare the NAMFIS solutions to these previously calculated structurest od etermine whether the presenceo fasecond ring significantly affects conformation. Firstly,i solated nisin ring B( 15)a nd aL an analogue ( 16)w ere compared to the corresponding regionso ft he nisin(1-12) (1) and (Thr2, Ser5) analogue (2)N AMFIS solutions( Figure 9A and B). As both 15 and 16 were found to bear a cisPro, backbone RMSD was low between these and most of the (Thr2, Ser5) analogue (2)c onformations, as well as for nisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)( 1)c onformation 1. Backbone RMSD was not as low when comparing the NAMFIS solutionst on isin ring A( 17)o rm utacin Ir ing A (Ser2, Ala5, Ala8) analogue ( 18) ( Figure 9C and D). Thisi sp resumably due to the increased flexibility made possible by the larger ring size, particularlyw hen Dha5 is replaced for am ore flexible residue such as in mutacin Ir ing A( Ser2, Ala5, Ala8) analogue (18). In all cases the largest divergence between the NAMFISs olutionsa nd the isolated rings appearst ob ei nt he positiono ft he lanthionine bridge, thought he higher flexibility of the lanthioninei nn isin ring Ai sw ell known in the literature. [14] Taken together, these results indicatet hat the structure of each individual lipid II-binding ring is not particularly affected by the presence of as econd ring, thought his could be attributed to the relative inflexibility of as mall cyclic peptide. However,t he results of the NAMFISa nalysisp resented here indicate that the nisin lipid II-binding region does exhibit flexibility,t hough it is predominantly aroundt he centrala mide bond between the two rings, and that the overall conformation is affected by the nature of the aminoacids within each ring.

Conclusions
Detailed studies of the interactions between the lantibiotic nisin and its biological target, lipid II, require the synthesis of both wild-type and chemically modified analogues of the key structuralm oieties. In this paper,w er eport the first syntheses of two analogues of the bicyclic N-terminus of nisin, rings A and B, which form ac age-like structure around the pyrophosphate group of lipid II. Cyclisation of ring As tructures from resin-bound intermediates with ring Bi ns itu provedc hallenging, probably due to the conformationalc onstraints and partial folding imposed on the key intermediates by the ring Bs tructure. These problems were overcome by the use of pseudoproline residues to induce at urn structure in the linear precursor to ring A, thus facilitatingc yclisation and allowing the previously unknown (Thr2, Ser5) nisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)analogue to be successfully prepared.
However,f or bicyclic peptides such as the (Abu2, Ala5) analogue 3,p seudoproline residues cannotb ei ncorporated into these sequences, and the yield of 3 wast hus disappointingly low.T he effects of substitution of l-Ala (and indeed D-Ala) for Dha at position5on the conformation of isolated ring As tructures are unclear [25,32] andt he ability to study both 3 and the d-Abu2, d-Ala5) analogue would have further confirmed whethers uch simplified analogues could effectively bind lipid II. This highlightst he need for further developmento fg enerally applicable methodology for the efficient cyclisationo fc onstrained or polycyclic peptides.
Our motivation for this study was to understand the degree to which the peptide sequence, and the conformational constraints imposed by the two thioether bridges, lead to cage structures that are pre-organised to bind to the pyrophosphate moiety of lipid II. Previous NMR studies hadi ndicated that individuall anthionine-bridged ring Bs tructures existed as mixtures of different conformers, [24,64] however NMR structures of fulllength nisin, either alone or bound to lipid II, show only a single conformer.W eh ave previously shown that isolated ring As tructures generally adopt as imilar conformation to that observedb yN MR for wt nisin in a1 :1 complex with lipid II in solution (1WCO), suggesting ad egree of pre-organisationo fr ing A. [25] However,w ealso demonstrated that isolated ring Bs tructures do not always adopt the conformation observed in the 1WCO NMR structure, althought his may be af unctiono ft he synthetic methodology used.
In this paper,w eh ave shown that the conformations of each of ring Aa nd ring Ba re hardly affected by the presence of the other ring, and do not appear to adopt two conformers in the bicyclic structure. However,t he nisin-lipid II binding region exhibits considerable flexibility around the (Lan7, MeLan8) amide bond between the two rings. For the wt nisin(1-12)s equence, this flexibility allows the two rings to fold into the pyrophosphate-binding cage observed in the 1WCO structure. Conversely,t he (Thr2, Ser5) analogue does not appear to form the lipid II-binding conformationi ns olution. Perhapsd ue to additional hydrogen bonding with Ser-OH, the (Thr2, Ser5) analogue adopts mostly cis conformation in solution. These resultsw ill inform the rational design of further lipid II-binding cage structures [35] whichc ould in turn representn ovel lead structures for the development of new antimicrobial peptides.
Our NAMFIS results must also be viewed in the light of recently reporteds olid state NMR studies [23] of the 8:4n isin:lipid II pore in both modell iposomes and in membrane vesiclesd erived from Micrococcus flavus. These have suggested that the 1:1n isin:lipid II structure [20] may not report on ap hysiologically relevant state. In particular, there were major perturbationso f the chemicals hifts of protons in the nisin(1-12) region between the solution state and solid state NMR structures,i ndi-cating that rings Aa nd Ba dopt different conformations in the 1:1c omplex solution structure compared with the 8:4p ore structure in the solid state. The observed discrepancies may stem from the different stoichiometrieso ft he complexes,additional constraintsi mposed by the pore structure, the differences in environmentb etween DMSO and am embrane-bound complex, or differences between the solutiona nd solid states. In addition, NMR studies on theb inding of the structurally un- Figure 9. Overlay of averaged solution phase structures [25] calculated inX PLOR-NIH [62,63] (green)with the corresponding regionso fnisin (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) or (Thr2,S er5) analogue from NAMFIS analysis (grey).
Chem.E ur.J.2019, 25,14572 -14582 www.chemeurj.org 2019 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim related lantibiotic nukacin ISK-1 to lipid II [65] have shown that this lantibiotic exists in two conformational states, but with only one conformation capable of bindingt ol ipid II. Intriguingly,t he two conformational states also differ in terms of the relative orientation of the two lanthionine-bridged rings that coordinate to the pyrophosphate moiety.T he NAMFIS deconvolutionm ethod has previously been used to analyse the ensemble of conformations of epothilones, [66] and of geldanamycin and radicicol. [67] In both cases, the presence (or absence) of the receptor-bound conformation in the NAMFIS ensemble was used to assess the plausibilityo fc onflicting solid-stateand solution structures,a nd the conformations could then be used as dockingc andidates for predicting the experimental binding poses in ligand-receptor complexes.I no ur work, 26 %o ft he conformations (conformations5and6 )d oc orrespond to the 1WCO 1:1c omplex structure, but the most populated set of conformations (conformations 3a nd 4, FigureS4: 52 %) do not correspond to the lipid II-bound conformation observed in the 1WCO structure. Our results suggest that the 1:1c omplex solution structure represents one plausible binding mode for the nisin:lipid II interaction. However,t here is another favourable set of conformationsa vailablet ot his bicyclic ring structure. These might correspond to an energetically favourableu nbound state, as in the nukacin ISK-1 structure, or to the conformations present in the nisin:lipid II 8:4s olid state pore structure. Our analysis of the conformational states present in the solution ensemble may enable these two possibilities to be distinguished, and will lead to ad eeper understandingo ft he complexf actorsgoverning the nisin:lipid II interaction in different environments.