marine drugs

Article Model-Free Approach for the Configurational Analysis of Marine Natural Products †

Matthias Köck 1,* , Michael Reggelin 2 and Stefan Immel 2,*

1 Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany 2 Clemens-Schöpf-Institute for Organic Chemistry and Biochemistry, Technical University of Darmstadt, Alarich-Weiss-Straße 4, 64287 Darmstadt, Germany; [email protected] * Correspondence: [email protected] (M.K.); [email protected] (S.I.) † Dedicated to Prof. Dr. Horst Kessler on occasion of his 81st birthday.

Abstract: The NMR-based configurational analysis of complex marine natural products is still not a routine task. Different NMR parameters are used for the assignment of the relative configuration: NOE/ROE, homo- and heteronuclear J couplings as well as anisotropic parameters. The combined distance geometry (DG) and distance bounds driven dynamics (DDD) method allows a model-free approach for the determination of the relative configuration that is invariant to the choice of an initial starting structure and does not rely on comparisons with (DFT) calculated structures. Here, we will discuss the configurational analysis of five complex marine natural products or synthetic derivatives thereof: the cis-palau’amine derivatives 1a and 1b, tetrabromostyloguanidine (1c), plakilactone H (2), and manzamine A (3). The certainty of configurational assignments is evaluated in view of the accuracy of the NOE/ROE data available. These case studies will show the prospective breadth of


application of the DG/DDD method.


Citation: Köck, M.; Reggelin, M.; Keywords: configurational analysis; distance geometry; distance bounds driven dynamics; NMR Immel, S. Model-Free Approach for spectroscopy; NOE data the Configurational Analysis of Marine Natural Products . Mar. Drugs 2021, 19, 283. https://doi.org/ 10.3390/md19060283 1. Introduction The determination of the relative configuration of marine natural products usually Academic Editor: Bill J. Baker follows the constitutional structure elucidation. Despite numerous attempts, there is so far no standard or even automated protocol for the configurational and conformational Received: 8 April 2021 assignment of natural products or small organic molecules in general [1–14]. Although Accepted: 11 May 2021 recently, the application of anisotropic NMR data, such as residual dipolar couplings Published: 21 May 2021 (RDCs) or residual chemical shift anisotropy (RCSA), has become more and more popular, the cross-relaxation (NOE or ROE) from which interproton distances can be obtained is Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in still a major source of structural information. The use of NOE- or ROE-derived inter- published maps and institutional affil- proton distances for the configurational analysis is a mature technique [15,16] that can iations. be amended by including RDCs in restrained distance geometry (rDG) simulations in a protocol that we have discussed in detail very recently [17]. Herein, we will concentrate on the NOE/ROE-restrained distance geometry [18–20] and distance bounds driven dynamics method (rDG/DDD) [21,22], which is, from our point of view, the method of choice for the configurational analysis of small organic molecules with several stereogenic centers [23–28]. Copyright: © 2021 by the authors. The use of interproton distances from NOESY spectra as restraints in molecular dynamics Licensee MDPI, Basel, Switzerland. This article is an open access article simulations (rMD) was introduced about 35 years ago [29,30] and since then, it has been distributed under the terms and a successful method for the NMR-based conformational analysis of biomacromolecules. conditions of the Creative Commons The DDD approach was later applied as a method for “shaking up” the structures before Attribution (CC BY) license (https:// starting the rMD analysis [21,22]. The concept of floating chirality (fc) was added to aid the creativecommons.org/licenses/by/ assignment of diastereotopic protons or methyl groups in proteins, and was first applied in 4.0/). 1988 to distance geometry (DG) calculations [31] and in 1989 to rMD simulations [32]. The

Mar. Drugs 2021, 19, 283. https://doi.org/10.3390/md19060283 https://www.mdpi.com/journal/marinedrugs Mar. Drugs 2021, 19, x FOR PEER REVIEW 2 of 15

Mar. Drugs 2021, 19, 283 2 of 15

was first applied in 1988 to distance geometry (DG) calculations [31] and in 1989 to rMD simulationsmost important [32]. aspect The most of the important rDG/DDD aspect method of the is therDG/DDD possibility method to allow is the configurations possibility to to allowdynamically configurations change duringto dynamically the simulation change (floating during chirality, the simulation fc) and therefore(floating tochirality, determine fc) andthe conformationtherefore to determine and the relative the conformation configuration and of smallthe relative organic configuration molecules, simultaneously of small or- ganic(fc-rDG/DDD). molecules, simultaneously (fc-rDG/DDD). ThisThis approach approach combines combines distance distance geometry geometry (DG) (DG) [18–20] [18–20] and and dist distanceance bounds bounds driven driven dynamicsdynamics (DDD) (DDD) [21,22], [21,22], using NOE/ROE-derivedNOE/ROE-derived distances distances as as restraints restraints [23–28]. [23–28]. The The generalgeneral workflow workflow of of the the DG/DDD DG/DDD method method is is repr representedesented in in Figure Figure 11.. Similar to using aa molecularmolecular model model construction construction kit, kit, molecular molecular ge geometriesometries are are generated generated within within the the limits limits of holonomicof holonomic distance distance bounds bounds (relations (relations between between positional positional coordinates), coordinates), derived derived from from the constitutionthe constitution and experimental and experimental distance distance margins margins deduced deduced from NMR from NMRdata. As data. this As step this a usesstep distances a uses distances only, and only, all and implied all implied informatio informationn on angles on angles and torsions and torsions is actually is actually dis- carded,discarded, it is it ensured is ensured that that the the structures structures gene generatedrated are are not not biased biased by by the the input input structure; structure; modelsmodels of of all all possible possible configurations configurations and and conf conformationsormations are are generated generated with with purely purely statis- sta- ticaltistical weights. weights. In order In order to add to additional add additional degrees degrees of freedom of freedom to this to building this building process process (“me- trization”),(“metrization”), it is carried it is carried out in out four-dimensional in four-dimensional (4D) Cartesian (4D) Cartesian space. space.These Theseinitial guess initial modelsguess models (Figure (Figure 1b) are1 )then are thenrefined refined by a sequence by a sequence of 4D of and 4D 3D and simulated 3D simulated annealing annealing sim- ulationssimulations (DDD) (DDD) where where at each at each step, step, the squared the squared violations violations of experimental of experimental restraints restraints are computedare computed as a dimensionless as a dimensionless pseudo pseudo energy energy (total error) (total error)with empirically with empirically chosen chosen“force constants”.“force constants”. In all Indynamics all dynamics and andoptimization optimization calculatio calculations,ns, the the partial partial derivatives ∂E /∂rα (α ∈ x, y, z(, w) for all atoms) of this pseudo energy term, with respect to 4D ∂Etotaltotal/∂rα (α∈x,y,z(,w) for all atoms) of this pseudo energy term, with respect to 4D and 3Dand Cartesian 3D Cartesian atomic atomic coordinate coordinates,s, are interpreted are interpreted and used and as used forces as governing forces governing the evolu- the tionevolution of the of system; the system; a detailed a detailed description description of this of this procedure procedure is isgiven given in in Refs. Refs. [17,21,22]. [17,21,22]. Finally,Finally, all emergent structuresstructures areare rankedranked and and sorted sorted by by their their pseudo pseudo energies energies (total (total error; er- ror;sum sum of total of total violations), violations), and and the lowerthe lower this “energy”this “energy” becomes, becomes, the more the more likely likely the corre- the correspondingsponding molecular molecular geometry geometry represents represents the correct the configurationcorrect configuration (and conformation) (and confor- of mation)the compound of the compound under investigation under investigation (Figure1). (Figure Note that 1a). at Note any pointthat at of any any point rDG/DDD of any rDG/DDDsimulation, simulation, neither a conventionalneither a conventional parametrized parametrized force-field force-field is involved, is involved, nor are any nor pre-are anysumptions presumptions on conformational on conformational preferences preferences implied, butimplied, all molecular but all structuresmolecular and structures relative configurations of stereogenic centers evolve solely on the basis of experimental NMR data and relative configurations of stereogenic centers evolve solely on the basis of experi- (Figure1, Steps b and c). mental NMR data (Figure 1, Steps b and c).

FigureFigure 1. 1. WorkflowWorkflow of of fc-rDG/DDD fc-rDG/DDD calculations: calculations: (a ()a based) based on on the the known known molecular molecular cons constitutiontitution in inan an arbitrary arbitrary configu- config- ration (and conformation), a set of holonomic intramolecular distances including bond lengths (1,2-distances), angles (1,3- uration (and conformation), a set of holonomic intramolecular distances including bond lengths (1,2-distances), angles distances), and torsions (1,4-distances), excluded van der Waals volumes, and NOE/ROE-derived distance limits between (1,3-distances), and torsions (1,4-distances), excluded van der Waals volumes, and NOE/ROE-derived distance limits all atom–atom pairs 𝑖,𝑗 defines boundary conditions for computing molecular structures similar to using a model con- structionbetween allkit. atom–atom (b) Using this pairs basici, j defines scaffolding boundary information, conditions initial for computing guess structure molecular models structures of arbitrary similar configuration to using a model and conformationconstruction kit.are (generatedb) Using this by a basic “metrization scaffolding” procedure information, in 4D initial space. guess (c) These structure models models are subsequently of arbitrary configuration refined through and anconformation automated sequence are generated of simulated by a “metrization annealing” procedure steps in 4D in and 4D space.3D space (c) Theseunder modelsthe restrain are subsequentlyts of experimental refined NOE/ROE through distancean automated limits, sequence by which of finally simulated the correct annealing configuration steps in 4D evolves and 3D as space the underbest-fit the structures restraints of of the experimental lowest pseudo NOE/ROE energy (totaldistance error). limits, by which finally the correct configuration evolves as the best-fit structures of the lowest pseudo energy (total error). Although DDD calculations do not apply any angle- or torsion-dependent pseudo energy terms, the geometry of tetrahedral or planar centers can be maintained by so-called

Mar. Drugs 2021, 19, 283 3 of 15

Mar. Drugs 2021, 19, x FOR PEER REVIEW 3 of 15 Although DDD calculations do not apply any angle- or torsion-dependent pseudo energy terms, the geometry of tetrahedral or planar centers can be maintained by so-called chiral volume restraints (Figure2). These are defined as pseudo scalar triple products chiral volume restraints→ (Figure→ → 2). These are defined as pseudo scalar triple products ⃗ Vchir = a · b × c of three vectors spanned by the substituents on each center, respec- Vchir=a⃗· b×c⃗ of three vectors spanned by the substituents on each center, respectively. Figure 2 illustratestively. the Figure definition2 illustrates of chiral the volumes definition in DG/DDD of chiral volumesfor sp3- and in DG/DDDfor sp2-type for sp 3- and for atomic centers, spwhere2-type the atomic absolute centers, configurat whereion the absoluteof stereogenic configuration centers is of encoded stereogenic by the centers is encoded sign of Vchir, andby theVchir sign=0 of forVchir planar, and fragments.Vchir = 0 for Though planar these fragments. parameters Though can these be used parameters can be in principle as restraintsused in principle to fix the as configuration restraints to of fix certain the configuration stereogenic ofcenters certain in stereogenicthe course centers in the of DDD simulations,course floating of DDD chir simulations,ality is achieved floating by chirality not using is achieved chiral volume by not restraints using chiral volume re- for unknown prochiralstraints for and unknown stereogenic prochiral centers. and In stereogenic general, we centers. apply In chiral general, volume we apply re- chiral volume straints only torestraints methyl groups only to to methyl keep groupsthem tetrahedral, to keep them and tetrahedral, to aryl rings and and to aryldouble rings and double bonds to boundbonds them toto boundplanarity, them respectively to planarity,; only respectively; one stereogenic only one center stereogenic of choice center is of choice is fixed in an arbitraryfixed inselected an arbitrary absolute selected config absoluteuration in configuration order to avoid in orderindistinguishable to avoid indistinguishable enantiomeric structures.enantiomeric structures.

Figure 2. Definition of chiral volumes for sp3-type (left) and sp2-type (right) atomic centers in Figure 2. Definition of chiral volumes for sp3-type (left) and sp2-type (right) atomic centers in DG/DDD simulations   ⃗ DG/DDD simulations as pseudo→ scalar→ → triple products Vchir=a⃗· b×c⃗. The sign of Vchir encodes as pseudo scalar triple products Vchir= a · b × c . The sign of Vchir encodes the absolute configuration of tetrahedral the absolute configuration of tetrahedral centers, whereas 𝑉 vanishes for planar structure frag- centers, whereasments.V Althoughchir vanishes Vchir for may planar serve structure as a chiral fragments. restraint Although in DG/DDDVchir calculations,may serve as omitting a chiral restraintthese in DG/DDD calculations,chiral omitting restraints these for chiral certain restraints prochiral for certain or stereogeni prochiralc centers or stereogenic by deleting centers them by from deleting the list them of fromthe the list of the chiral volumeschiral allows volumes these allows centers these to “float” centers (i.e., to “float” change (i.e their., change configuration) their configuration) within the calculationwithin the calcula- (floating chirality). tion (floating chirality). In contrast to rMD simulations, DG relies solely on experimental data, and all stere- In contrastogenic to rMD centers simulations, are allowed DG relies to adopt solely their on experimental relative configuration data, and all in accordancestere- with the ogenic centers experimentalare allowed to data adopt and their thus, rela removetive configuration any intrinsic biasin accordance imposed onwith the the results by not im- experimental dataplementing and thus, physical remove force-fieldsany intrinsic of bias any imposed type. As on already the results mentioned, by not anotherim- important plementing physicalaspect force-fields of the DG/DDD of any approach type. As isalready that all mentioned, structure models another are important first generated in four- aspect of the DG/DDDdimensional approach (4D) spaceis that before all structure these are models transferred are firs intot generated “real” 3D in space four- (Figure1, Step dimensional (4D)b). space The extra before dimension these are nottransferred only provides into “real” additional 3D space degrees (Figure of 1, freedom Step to assemble b). The extra dimensionstructures ofnot different only provides configurations additional and degrees conformations of freedom within to theassemble limits of the distance structures of differentbounds, configurations but also efficiently and conf eliminatesormations “energy” within barriers the limits for of configurational the distance inversions in bounds, but alsothe efficiently DDD step eliminates [33,34]. By “energy” analogy barriers with 2D for chiral configurational objects that can inversions be transformed in into each Mar. Drugs 2021, 19, x FOR PEER REVIEW 4 of 15 the DDD step [33,34].other through By analogy a rotation with 2D in chiral higher objects dimensions that can (3D), be thetransformed configuration into each of 3D chiral objects other through acan rotation be inverted in higher by rotationsdimensions in 4D(3D), (Figure the configuration3). of 3D chiral objects can be inverted by rotations in 4D (Figure 3). However, there are fundamental differences in treating large biomacromolecules, such as proteins or small (marine) natural products, using the rDG/DDD approach. For the former, usually very large datasets of intramolecular distances can be obtained, and the “configurational problem” can generally be ignored, as the configurations of the re- peating units (e.g., amino acids for proteins) are known. Thus, only qualitative distances defined by general categories (e.g., “weak”, “medium”, and “strong”) are sufficient to ob- tain reasonable 3D models for biomacromolecules. Quite another matter are small mole- FigureFigure 3. By 3. By analogy analogy with with 2D 2D chiral chiral objects objects (left (left) that) that can can be transformedbe transformed into into each each other other through through a rotation a rotation in higherin higher dimensionscules (3D), with the multiple configuration stereogenic of 3D chiral centers, objects where can thebe inverted relative just configuration by rotations ofin the4D (constituentright). dimensionsunits (3D), is the unknown configuration a priori of 3D and chiral only objects a limited can be invertednumber justof NOE by rotations contacts in 4Dcan (right be identified.). Therefore, more accurateGenerally, intramolecular the interproton distances distances and tighter are upperderived and by lowervolume distance integration of the cross- limits are requiredpeaks to unequivocally in the 2D NOESY deduce or the 2D relative ROESY configuration spectrum [15,16]. of complex The conversion natural of volume inte- products. grals into distances is carried out using reference peaks related to known intramolecular distances (“fixed distances”). Typical examples are diastereotopic methylene groups or vicinal CH–CH contacts in aromatic systems. All other distances are then derived from the r−6-weighted ratios of the corresponding peak volumes with respect to the volume of the reference cross peak. An important point when dealing with NOESY and ROESY spec- tra is the mixing time of the experiment. The dependence of the NOE or ROE effect with respect to the mixing time is of an exponential nature, which does require measuring build-up rates with several different mixing times. This can be circumvented by using the linear approximation approach (initial rate approximation), which would require meas- urement with relative short mixing times (linear region of the exponential function). This assumption is valid when the distances extracted for two different mixing times are iden- tical. Especially for small molecules in the extreme narrowing limit, the accumulation of quantitative NOESY/ROESY spectra also requires carefully degassing the NMR samples by freeze–pump–thaw techniques in order to remove dissolved paramagnetic oxygen. These experimental intricacies and the mixing-time dependency of the effect frequently lead to using only qualitative NOE intensities (“weak”, “medium”, and “strong”) and ra- ther large empirical error estimates on internuclear distances for the configurational anal- ysis of natural compounds. The choice for a NOESY or a ROESY spectrum depends mainly on the magnetic field strength, the size of the molecule, and the solvent viscosity. NOESY spectra have the disadvantage that for medium-sized molecules, the NOE becomes 0 if (ω0·τC) ≈ 1.12. This is approximately the case for a correlation time 𝜏 of 0.3 ns at 400 MHz. ROESY spectra do not have this disadvantage, but because of the necessary spinlock to achieve transverse cross relaxation, other effects (e.g., offset dependence or TOCSY arti- facts) have to be considered. In general, ROESY spectra are plagued by more artifacts than NOESY spectra [16]. For the conversion of the NOE/ROE-derived distances into distance restraints (upper/lower bounds), we generally use ±10% (+pseudo-atom [35] correction for positions with more than one isochronous H) in order to account for experimental errors. For compounds 1a–c, we also tested ±5%, ±20%, and ±30% error margins on distances in order to evaluate the certainty of configurational assignments with respect to the accuracy of the experimental data (more details are given in Chapter 2). In the literature, it is a common standard to assume an uncertainty of ±10% in NOE/ROE-derived distances due to experimental errors (without the consideration of the pseudo-atom correction), though it must be noted that the error-prone quantities meas- ured are actually the volume integrals of the cross peaks. Due to the r−6 scaling of the cross-peak volumes, the uncertainties in both quantities peak volumes and distances scale differently, and—given a proper setup of the NMR experiment and correct evaluation of the build-up rates and mixing times—an assumed error of ±10% of the former leads to a drop in the accuracy of the distance information to well below ±5%.

Mar. Drugs 2021, 19, 283 4 of 15

However, there are fundamental differences in treating large biomacromolecules, such as proteins or small (marine) natural products, using the rDG/DDD approach. For the former, usually very large datasets of intramolecular distances can be obtained, and the “configurational problem” can generally be ignored, as the configurations of the repeating units (e.g., amino acids for proteins) are known. Thus, only qualitative distances defined by general categories (e.g., “weak”, “medium”, and “strong”) are sufficient to obtain reasonable 3D models for biomacromolecules. Quite another matter are small molecules with multiple stereogenic centers, where the relative configuration of the constituent units is unknown a priori and only a limited number of NOE contacts can be identified. Therefore, more accurate intramolecular distances and tighter upper and lower distance limits are required to unequivocally deduce the relative configuration of complex natural products. Generally, the interproton distances are derived by volume integration of the cross- peaks in the 2D NOESY or 2D ROESY spectrum [15,16]. The conversion of volume integrals into distances is carried out using reference peaks related to known intramolecular distances (“fixed distances”). Typical examples are diastereotopic methylene groups or vicinal CH–CH contacts in aromatic systems. All other distances are then derived from the r−6- weighted ratios of the corresponding peak volumes with respect to the volume of the reference cross peak. An important point when dealing with NOESY and ROESY spectra is the mixing time of the experiment. The dependence of the NOE or ROE effect with respect to the mixing time is of an exponential nature, which does require measuring build-up rates with several different mixing times. This can be circumvented by using the linear approximation approach (initial rate approximation), which would require measurement with relative short mixing times (linear region of the exponential function). This assumption is valid when the distances extracted for two different mixing times are identical. Especially for small molecules in the extreme narrowing limit, the accumulation of quantitative NOESY/ROESY spectra also requires carefully degassing the NMR samples by freeze–pump–thaw techniques in order to remove dissolved paramagnetic oxygen. These experimental intricacies and the mixing-time dependency of the effect frequently lead to using only qualitative NOE intensities (“weak”, “medium”, and “strong”) and rather large empirical error estimates on internuclear distances for the configurational analysis of natural compounds. The choice for a NOESY or a ROESY spectrum depends mainly on the magnetic field strength, the size of the molecule, and the solvent viscosity. NOESY spectra have the disadvantage that for medium-sized molecules, the NOE becomes 0 if (ω0·τC) ≈ 1.12. This is approximately the case for a correlation time τc of 0.3 ns at 400 MHz. ROESY spectra do not have this disadvantage, but because of the necessary spinlock to achieve transverse cross relaxation, other effects (e.g., offset dependence or TOCSY artifacts) have to be considered. In general, ROESY spectra are plagued by more artifacts than NOESY spectra [16]. For the conversion of the NOE/ROE-derived distances into distance restraints (upper/lower bounds), we generally use ±10% (+pseudo-atom [35] correction for positions with more than one isochronous H) in order to account for experimental errors. For compounds 1a–c, we also tested ±5%, ±20%, and ±30% error margins on distances in order to evaluate the certainty of configurational assignments with respect to the accuracy of the experimental data (more details are given in Chapter 2). In the literature, it is a common standard to assume an uncertainty of ±10% in NOE/ROE-derived distances due to experimental errors (without the consideration of the pseudo-atom correction), though it must be noted that the error-prone quantities measured are actually the volume integrals of the cross peaks. Due to the r−6 scaling of the cross-peak volumes, the uncertainties in both quantities peak volumes and distances scale differently, and—given a proper setup of the NMR experiment and correct evaluation of the build-up rates and mixing times—an assumed error of ±10% of the former leads to a drop in the accuracy of the distance information to well below ±5%. Mar. Drugs 2021, 19, x FOR PEER REVIEW 5 of 15

2. Results and Discussion

+ Mar. Drugs 2021, 19, 283 Here, the application of the fc-rDG/DDD method as implemented in ConArch 5 [36– of 15 38] will be demonstrated on five complex natural products (Scheme 1). Compounds 1a to 1c are synthetic derivatives or congeners of palau’amine, which is probably the most prominent cyclic dimeric pyrrole-imidazole alkaloid (PIA). Palau’amine, a PIA with a hex- acyclic2. Results core, and was Discussion originally described in 1993 from the marine sponge Stylotella agminata (in 1998Here, changed the application to Stylotella of the aurantium fc-rDG/DDD) from method the Western as implemented Caroline in Islands ConArch (1d+ [36) [39].–38] Moreover,will be demonstrated the successful on fiveapplication complex of natural the method products will (Scheme be shown1). for Compounds two other 1amarineto 1c naturalare synthetic products, derivatives using data or congeners from the literature. of palau’amine, The oxygenated which is probably polyketide the plakilactone most promi- Hnent (2) cyclicwas isolated dimeric from pyrrole-imidazole the marine sponge alkaloid Plakinastrella (PIA). Palau’amine, mamillaris a from PIAwith Fiji in a hexacyclic2013 [40]. Thecore, β was-carboline originally alkaloid described manzamine in 1993 A from (3) was the isolated marine spongefrom theStylotella marine agminatasponge Haliclona(in 1998 spp.changed from to Okinawa,Stylotella Japan aurantium in 1986) from [41]. the Western Caroline Islands (1d)[39]. Moreover, the successfulAs NMR application can determine of the methodrelative willconfigurations be shown for only, two in other all rDG marine simulations, natural products, a single stereogenicusing data fromcenter the of literature. compounds The 1 oxygenated to 3 each was polyketide fixed by plakilactone applying a Hchiral (2) was volume isolated re- straintfrom the in marineorder to sponge avoid Plakinastrellaenantiomeric mamillaris structures.from In the Fiji insequel, 2013 [the40]. number The β-carboline of the gener- alka- atedloid manzaminestructures in A the (3) wasfc-rDG isolated/DDD fromcalculations the marine was sponge set to Haliclona1000 to allowspp. fromfor reasonable Okinawa, samplingJapan in 1986 of the [41 configurational]. and conformational space.

Scheme 1. StructuralStructural formulae formulae of of the the investigated investigated molecule moleculess with with atom atom numbering, numbering, respectively: respectively: synthetic synthetic ciscis-palau’amine-palau’amine derivatives 1a and 1b, tetrabromostyloguanidine ( 1c), plakilactone H ( 2), and manzamine A ( 3). Important Important positions with respect to the configuration configuration of 1 are highlighted in red.

2.1. Palau’amineAs NMR can Derivatives determine (1) relative configurations only, in all rDG simulations, a single stereogenicPalau’amine center itself of compounds was a very1 to prominent3 each was target fixed in by synthetic applying organic a chiral volumechemistry restraint from 1995in order to 2010 to avoid [42]. The enantiomeric original structure structures. of palau’amine In the sequel, (1d the), proposed number in of 1993 the generated [39], was firststructures revised in in the 1998 fc-rDG/DDD [43], correcting calculations the configuration was set to 1000 of C-20 to allow (1e).for Asreasonable discovered sampling in 2007, upof theto that configurational time, all groups and followed conformational the wrong space. target with cis-fused rings D and E [44]. In that year, the relative configuration of C-12 and C-17 of the palau’amine congeners was 2.1. Palau’amine Derivatives (1) revised and the unusual trans-fusion of the five-membered rings was established [45–47]. The totalPalau’amine synthesis itselfof rac was-1f (Scheme a very prominent 2) was completed target in by synthetic the Baran organic groupchemistry in 2010 [48,49], from and1995 the to 2010asymmetric [42]. The variant original in 2011 structure [50]. of palau’amine (1d), proposed in 1993 [39], was firstCompounds revised in 1998 1a [and43], 1b correcting are synthetic the configuration intermediates of [51] C-20 on (1e the). Asroute discovered to the “old” in 2007, rel- cis ativeup to configuration that time, all groupsof palau’amine followed ( the1d wrongand 1e). target From with our point-fused of ringsview, Dthese and Etwo [44 syn-]. In that year, the relative configuration of C-12 and C-17 of the palau’amine congeners was thetic derivatives and compound 1c are perfect candidates for which we have illustrated revised and the unusual trans-fusion of the five-membered rings was established [45–47]. the “starting structure problem”. The cis- and trans-configured palau’amine congeners are The total synthesis of rac-1f (Scheme2) was completed by the Baran group in 2010 [ 48,49], very well-known examples from the literature (as mentioned before) and therefore, an and the asymmetric variant in 2011 [50]. Compounds 1a and 1b are synthetic intermediates [51] on the route to the “old” relative configuration of palau’amine (1d and 1e). From our point of view, these two synthetic derivatives and compound 1c are perfect candidates for which we have illustrated the “starting structure problem”. The cis- and trans-configured palau’amine congeners are very well-known examples from the literature (as mentioned before) and therefore, an ideal Mar. Drugs 2021, 19, 283 6 of 15 Mar. Drugs 2021, 19, x FOR PEER REVIEW 6 of 15

ideal test case fortest the case evaluation for the evaluation of the fc-rDG/DDD of the fc-rDG/DDD method. method.Additionally, Additionally, 1a and 1b1a dif-and 1b differ from fer from palau’aminepalau’amine in that in there that there is no is ch nolorine chlorine at C-17, at C-17, which which reduces reduces the thenumber number of of contiguous contiguous stereogenicstereogenic centers centers from from eight eight to seven, to seven, resulting resulting in 64 in possible 64 possible relative relative con- configurations figurations (diastereomers),(diastereomers), respectively. respectively. Furthermore, Furthermore, the the shortshort sideside chainchain of of palau’amine pa- at C-18 (–CH –NH in 1d–f) is changed to –OH (1a and 1b). Compounds 1a and 1b are C-20 lau’amine at C-18 (–CH2 2–NH2 2 in 1d–f) is changed to –OH (1a and 1b). Compounds 1a and 1b are C-20 epimers,epimers, which which are are the the stereogeni stereogenicc centers centers whose whose configuration configuration was was revised revised from 1993 from 1993 (1d)( 1dto )1998 to 1998 (1e). (1e Tetrabromostyloguanidine). Tetrabromostyloguanidine (1c ()1c has) has eight eight contiguous contiguous stereo- stereogenic centers, resulting in 128 possible relative configurations (diastereomers). genic centers, resulting in 128 possible relative configurations (diastereomers).

Scheme 2. OriginalScheme 2. (1993: Original1d [ 39(1993:]) and 1d revised [39]) and (1998: revised1e [ 43(1998:] and 1e 2007: [43] and1f [45 2007:]) structural 1f [45]) structural formulae formulae of palau’amine. The structures areof allpalau’amine. shown in the The correct structures absolute are all configuration shown in the [50 correct,52,53] abso and therefore,lute configuration1d and 1e [50,52,53]are drawn and as enantiomers, compared totherefore, the original 1d and publications 1e are drawn [39,43 as]. enantiomers, compared to the original publications [39,43].

For our investigationFor our on investigation compounds on 1a compounds to 1c, we used1a to the1c ,ROE-derived we used the distances ROE-derived distances from Overman’sfrom work Overman’s in 2007 (from work the in ROESY 2007 (from spectra the with ROESY a mixing spectra time with of 100 a mixing ms; 1a time: of 100 ms; 16 ROEs, 1b: 171a :ROEs, 16 ROEs, and 1b1c:: 9 17 ROEs) ROEs, [51]. and The1c: experimental 9 ROEs) [51]. setup The experimental for 1a and 1b setup was for 1a and 1b identical to ourwas own identical setup for to tetrabromostyloguanidine our own setup for tetrabromostyloguanidine (1c) in 2007 [45], which (1c )is in a 2007per- [45], which is fect preconditiona perfect for a direct precondition comparison for aof direct the results. comparison The complete of the results. lists of The ROEs complete of 1a, lists of ROEs 1b, and 1c are givenof 1a, in1b the, and Supplementary1c are given in Materials the Supplementary (1a: Table S1a, Materials 1b: Table (1a S1b,: Table and S1a, 1c: 1b: Table S1b, Table S1c). As andmentioned1c: Table above, S1c). one As stereoge mentionednic above,center of one each stereogenic compound center was fixed of each and compound was set as a referencefixed (C-10 and for set all as three a reference compounds). (C-10 forIn a all traditional three compounds). approach of In pre-calcu- a traditional approach lating structures,of pre-calculatingthis would entail structures, the necessit thisy wouldto evaluate entail a total the necessity of 64 or 128 to evaluate diastere- a total of 64 or omers, respectively.128 diastereomers, In the fc-rDG/DDD-calculations respectively. In the described fc-rDG/DDD-calculations here this is not necessary described here this is because these configurationsnot necessary evolve because under these the configurations influence of the evolve distance under restraints the influence without of the distance any action of therestraints experimentalist. without anyThe actionROE interproton of the experimentalist. distances for Thecompounds ROE interproton 1a to 1c distances for were used withcompounds varying lower1a to and1c upperwere useddistance with bounds varying (±5%, lower ±10%, and ±20%, upper and distance ±30%) bounds (±5%, in order to demonstrate±10%, ± 20%,the importance and ±30%) of in accurate order to distance demonstrate limits for the the importance configurational of accurate distance analysis of smalllimits molecules for the configurational in floating-chirality-restrained analysis of small moleculesDG/DDD incalculations floating-chirality-restrained (fc- rDG/DDD). DG/DDD calculations (fc-rDG/DDD). The total errorThe (dimensionless) total error (dimensionless) of the 500 lowest of thepseudo 500lowest energy pseudo structures energy for 1a structures is for 1a is plotted for eachplotted structure, for each ordered structure, accordin orderedg to ascending according total to ascending errors. The total first errors. wrong The first wrong ± configuration ofconfiguration 1a (for the standard of 1a (for DG/ theDDD standard calculations DG/DDD using ROEs calculations ±10%) with using respect ROEs 10%) with to the seven stereogenicrespect to centers the seven is No. stereogenic 93 (green centers square is in No. Figure 93 (green 4a). This square structure in Figure differs4a). This structure differs from structures #1 to #92 by a change in all configurations, besides C-6. These two from structures #1 to #92 by a change in all configurations, besides C-6. These two relative relative configurations have a ∆E value of 0.40. The result of this calculation (±10%) is configurations have a 𝛥𝐸 value of 0.40. The result of this calculation (±10%) is not de- not dependent on the starting structure. Irrespective of the cis- or trans-configuration as pendent on the starting structure. Irrespective of the cis- or trans-configuration as input input coordinates for the calculation, the ROE-derived distances of 1a lead to a cis-fused coordinates for the calculation, the ROE-derived distances of 1a lead to a cis-fused final final structure. Softening up the ROE distance bounds to ±20% and ±30% dramatically structure. Softening up the ROE distance bounds to ±20% and ±30% dramatically down- downgrades the certainty with which the correct configuration of 1a is identified by the grades the certainty with which the correct configuration of 1a is identified by the DG/DDD simulations. DG/DDD simulations.

Mar. Drugs 2021, 19, 283 7 of 15

Mar. Drugs 2021, 19, x FOR PEER REVIEW 7 of 15

(a) (b)

(c) (d)

(e) (f)

Figure 4. (aFigure) Plot of 4.the( atotal) Plot “pseudo of the energy” total of “pseudo ranked rDG energy” structures of rankedfor 1a, showing rDG structures the first 500 forout of1a 1000, showing structures the first 500 −2 generated (KNOE = 50.0 Å ). The plot shows the results for four fc-rDG/DDD−2 calculations with different ranges for the ROE- derived distanceout of restraints: 1000 structures ±5% (orange), generated ±10% (green), (KNOE ±20%= 50.0 (blue), Å and). The±30% plot(black). shows The bold the printed results symbols for four in fc-rDG/DDDthe correspondingcalculations colors highlight with the different first wrong ranges structure, for respectively the ROE-derived (±5%: #138, ±10%: distance #93, ±20%: restraints: #21, and ±±30%:5% #3). (orange), The ±10% dashed lines(green), indicate± the20% corresponding (blue), and Δ𝐸± 30% values, (black). which areThe given bold on printed the rightsymbols side of the in plot. the The corresponding inset plot shows colors the highlight the first wrong structure, respectively (±5%: #138, ±10%: #93, ±20%: #21, and ±30%: #3). The dashed lines indicate the corresponding ∆E values, which are given on the right side of the plot. The inset plot shows the “energy” steps in detail (first 250 structures). (b) Molecular structures for 1a showing the superposition of all DG structures identified up to the first wrong configuration (92 structures) of the ±10% calculation; the DG best-fit (lowest pseudo energy) structure is plotted below, displaying the correct configuration of 1a.(c,d) Similar plot as above for compound 1b and (e,f) compound 1c (first wrong structures of 1b: ±5%: #290, ±10%: #174, ±20%: #84, ±30%: #53, and 1c: ±5%, #196, ±10%: #175, ±20%: #114, and ±30%: #66; number of superimposed structure: 1b: 173 and 1c: 174). Mar. Drugs 2021, 19, 283 8 of 15

The results of the fc-rDG/DD calculations for the synthetic palau’amine derivative 1b are represented in Figure4c,d by plots equivalent to those given above for 1a. Although compounds 1a and 1b only differ in the configuration of a single stereogenic center (C-20) and in one ROE (16 versus 17 experimental restraints), the differentiability of the diastere- omers of 1b is much more pronounced, compared to 1a. The first different configuration of 1b (for the standard DG/DDD calculations using ROEs ±10%) with respect to the seven stereogenic centers is No. 174 (green square in Figure4c). This structure differs from struc- tures #1 to #173 by the configuration of C-20. As already observed for 1a, the certainty with which the correct configuration of 1b can be identified becomes significantly attenuated the more broadly the restraining boundaries are chosen (Figure4c). On the contrary, setting tighter distance limits (±5%) to the experimental restraining dataset again improves the reliability of the configurational assignment dramatically, and the first wrong configuration is observed for structure No. 290 (orange circle in Figure4c) whilst the ∆E value increases from 1.69 to 2.05. The simulation results for tetrabromostyloguanidine (1c) are depicted in Figure4e in the same way as described above. The first wrong configuration of 1c (for the standard DG/DDD calculations using ROEs ±10%) with respect to the seven stereogenic centers is No. 175 (green square in Figure4e), differing from the preceding structures by the configuration of C-16. Similar to the results obtained above for 1a and 1b, tighter NOE restraints lead to an increase in the differentiability of the correct configuration of 1c, and looser restraints lead to a significant loss of information. It must be stressed that all rDG simulations yielding quickly and with high reliability the correct configurations of the palau’amine derivatives (1a–c) are actually single, fully automated calculations (one for each compound) and not 64 or 128 individual calculations on alternate diastereomers. Due to the fact that at no point of the simulations a physical force-field is involved, the method finds cis-(1a and 1b) or trans-fused (1c) five-membered rings (bicyclo [3.3.0] octane ring systems, Schemes1 and2, and orange colored structure fragments in the plots of Figure4b,d,f) without bias toward the energetically strong-favored cis-fusion. Unlike simulations employing physical force-fields, which inevitably would tend to favor the low-energy cis-connectivity of the fused rings, the DG simulations do not suffer from this energetic bias, but only follow the restraints imposed by the experimental data. In particular, the initial 4D “metrization” step that antedates any rDG/DDD simulation ensures that the DG results do not depend on an assumed initial configuration, as any biased information in this respect is actually discarded at the earliest possible point of each simulation.

2.2. Plakilactone H (2) The polyketide plakilactone H (2), which was isolated from the marine sponge Plak- inastrella mamillaris, was chosen as the next example. We used the distance information from the original publication on the structural elucidation of 2 (25 interproton distances extracted from 1D NOESY spectra with mixing times of 500 ms; the complete list of NOEs of 2 is given in the Supplementary Materials, Table S2) [40]. Using the same methodology as in case of the palau’amine derivatives (1), the results for the 1000 generated possible structures of plakilactone H (2) are shown in Figure5a (“best 400”) as a graphical representation of the total error (dimensionless) for each structure, ordered according to ascending total errors; the C-4 quaternary stereogenic center of 2 was set as reference and fixed by the application of a chiral volume restraint. In the low-pseudo energy region of structures identified for 2 (“best 100” inset plot of Figure5a) six small energy steps are present, which in total correspond to two con- figurational families A (#1 to #42, in total 42 structures) and B (#43 to #72, in total 30 structures) separated by a very small energy difference with ∆E = 0.10. Both families are subclassified by even smaller energy jumps due to different rotamers of the ethyl groups, which arise from the impossibility to assign the diastereotopic methylene protons a priori and individually in this example. All structures of groups A and B feature the same relative Mar. Drugs 2021, 19, x FOR PEER REVIEW 9 of 15

as any biased information in this respect is actually discarded at the earliest possible point of each simulation. Mar. Drugs 2021, 19, 283 9 of 15 2.2. Plakilactone H (2) The polyketide plakilactone H (2), which was isolated from the marine sponge Plaki- nastrella mamillaris, was chosen as the next example. We used the distance information configuration of the stereogenic centers C-6, C-7, and C-8, but differ in their configuration from the original publication on the structural elucidation of 2 (25 interproton distances of C-4 only (see structures shown in Figure5b). Noteworthy, as a side note, is the fact extracted from 1D NOESY spectra with mixing times of 500 ms; the complete list of NOEs that the structures in group B actually emerged as mirror images, because we fixed the of 2 is given in the Supplementary Materials, Table S2) [40]. configuration of C-4 in the calculations presented here, and were inverted for plotting Using the same methodology as in case of the palau’amine derivatives (1), the results in Figure5b for clarity only. In Figure5a, additional ethyl rotamers and C-10 epimers for the 1000 generated possible structures of plakilactone H (2) are shown in Figure 5a intermixed with each other are found along the first larger step of pseudo energy ranked (“best 400”) as a graphical representation of the total error (dimensionless) for each struc- structures (#73 to #280), followed by another step that is associated with the first wrong ture, ordered according to ascending total errors; the C-4 quaternary stereogenic center of side chain configuration (structure No. #281, bold green circle in Figure5a). 2 was set as reference and fixed by the application of a chiral volume restraint.

(a) (b)

FigureFigure 5.5. ((a)) PlotPlot ofof thethe totaltotal “pseudo“pseudo energy”energy” ofof rankedranked rDG structures of plakilactone H, (22)) showingshowing thethe firstfirst 400400 outout ofof −−22 10001000 structuresstructures generated generated (K (NOEKNOE= =10.0 10.0 Å Å ). The minimumminimum energyenergy level level is is indicated indicated by by the the horizontal horizontal dashed dashed green green line, line, the dashedthe dashed lines lines at higher at higher energies energies indicate indicate the differentiabilitythe differentiabili ofty the of best-fitthe best-fit solution solution with with respect respect to alternate to alternate configurations. configura- Thetions. inset The plot inset shows plot shows the first the “energy” first “energy” stepsin steps detail in (firstdetail 100 (first structures). 100 structures). (b) Molecular (b) Molecular structures structures of plakilactone of plakilactone H, (2) H, (2) showing the superposition of all DG structures of configurational family A (42 structures, top left) and configura- showing the superposition of all DG structures of configurational family A (42 structures, top left) and configurational tional family B (30 structures, top right). Molecular structures of plakilactone H, (2) showing the superposition of the best family B (30 structures, top right). Molecular structures of plakilactone H, (2) showing the superposition of the best 72 DG 72 DG structures (configurational families A and B, bottom). structures (configurational families A and B, bottom). In the low-pseudo energy region of structures identified for 2 (“best 100” inset plot In summary, the DG/DDD method unequivocally identifies the correct side chain of Figure 5a) six small energy steps are present, which in total correspond to two config- configuration of 2 (stereogenic centers C-6, C-7, and C-8) on the basis of the NOE data. As urational families A (#1 to #42, in total 42 structures) and B (#43 to #72, in total 30 struc- displayed by the structures plotted in Figure6, this approach simultaneously also detects tures) separated by a very small energy difference with Δ𝐸 = 0.10. Both families are sub- ambiguities in the configurational assignment of C-4 (marked in orange). Indeed, due to classified by even smaller energy jumps due to different rotamers of the ethyl groups, the flexibility of the ethyl groups, the experimental data are commensurable with either thewhich 4R arise(epimer from A) the or 4impossibilityS configuration to assign (epimer the B) diastereotopic with about the methylene same error protons margins. a priori and individually in this example. All structures of groups A and B feature the same rela- 2.3.tive Manzamineconfiguration A ( 3of) the stereogenic centers C-6, C-7, and C-8, but differ in their configu- rationThe of marineC-4 only natural (see struct productures manzamine shown in Figure A (3) was 5b). chosenNoteworthy, as a very as complexa side note, structure is the withfact that a relative the structures small number in group of stereogenicB actually emerged elements as (five mirror stereogenic images, because centers pluswe fixed two doublethe configuration bonds) [41 of]. C-4 For in manzamine the calculations A (3), presented we used a here, hybrid and dataset were inverted based on for only plotting nine experimentalin Figure 5b NOEfor clarity contacts only. listed In Figure in [41]. 5a, Since additional the authors ethyl only rotamers have given and the C-10 interactions epimers ofintermixed the protons with and each not other the actual are found distances along between the first them, larger we step used of pseudo these nine energy interactions ranked andstructures added (#73 the distancesto #280), fromfollowed the crystal by another structure step of that3 [41 is] associated to obtain the with distance the first restraints wrong forside the chain DG/DDD configuration calculations (structure (the complete No. #281, list bold of NOEs green of circle3 is givenin Figure in the 5a). Supplementary Materials, Table S3). The results for the 500 low-energy structures out of 1000 generated structures of manzamine A (3) are shown in Figure7a (in the same representation as for 1 and 2; C- 12 fixed and set as reference). However, the limited NOE dataset and the pronounced conformational flexibility of the 8- and 13-membered rings in 3 make an unambiguous Mar. Drugs 2021, 19, x FOR PEER REVIEW 10 of 15 Mar. Drugs 2021, 19, 283 10 of 15

In summary, the DG/DDD method unequivocally identifies the correct side chain configurationconfigurational of assignment 2 (stereogenic unfeasible, centers C-6, as the C-7, pseudo and C- energy8) on stepsthe basis and of differences the NOE data. between As the best-fit (correct) and first alternate (wrong, No. #267) configuration vanish (Figure7a , displayed by the structures plotted in Figure 6, this approach simultaneously also detects black line, ∆E = 0.10). This means that even with this sparse dataset, the correct config- ambiguities in the configurational assignment of C-4 (marked in orange). Indeed, due to uration was found in the lowest energy structures but due to the missing energy jumps the flexibility of the ethyl groups, the experimental data are commensurable with either between these structures and the wrong ones, no reliable decision could be made. the 4R (epimer A) or 4S configuration (epimer B) with about the same error margins.

Mar. Drugs 2021, 19, x FOR PEER REVIEW 11 of 15

Figure 6. PlotPlot of the best-fit best-fit (minimum pseudo energy) DG structuresstructures ofof bothboth configurationsconfigurations ((((AA,B,B),), C-4 C-4 epimers, epimers, atom atom C-4 C- 4 is presented in orange) of plakilactone H (2) with color-coded representation of all NOE contacts used in the configura- is presented in orange) of plakilactone H (2) with color-coded representation of all NOE contacts used in the configurational tional and conformational analysis. The color scale was adapted from calculated final NOE violations, ranging from −0.25 and conformational analysis.#185 The (conformational color scale was adapted family fromA) of calculated manzamine final A NOE (3) and violations, also the ranging structures from −#1860.25 to Å #327 Å (blue) to +0.25 Å (red). (blue) to +0.25 Å (red). (top, right; conformational family B). All structures with the correct configuration of 3 (420 structures) are shown in Figure 7b (bottom). 2.3. Manzamine A (3) The marine natural product manzamine A (3) was chosen as a very complex structure with a relative small number of stereogenic elements (five stereogenic centers plus two double bonds) [41]. For manzamine A (3), we used a hybrid dataset based on only nine experimental NOE contacts listed in [41]. Since the authors only have given the interac- tions of the protons and not the actual distances between them, we used these nine inter- actions and added the distances from the crystal structure of 3 [41] to obtain the distance restraints for the DG/DDD calculations (the complete list of NOEs of 3 is given in the Sup- plementary Materials, Table S3). The results for the 500 low-energy structures out of 1000 generated structures of man- zamine A (3) are shown in Figure 7a (in the same representation as for 1 and 2; C-12 fixed and set as reference). However, the limited NOE dataset and the pronounced conforma- tional flexibility of the 8- and 13-membered rings in 3 make an unambiguous configura- tional assignment unfeasible, as the pseudo energy steps and differences between the best- fit (correct) and first alternate (wrong, No. #267) configuration vanish (Figure 7a, black line, ΔE=0.10). This means that even with this sparse dataset, the correct configuration was found in the lowest energy structures but due to the missing energy jumps between these structures and the wrong ones, no reliable decision could be made. (a) For demonstration purposes, we have, therefore,( bdecided) to add artificial residual dipolar couplings (RDCs) to the dataset of 3. The RDCs were calculated from the crystal FigureFigure 7.7. ((aa)) PlotPlot ofof thethe totaltotal “pseudo“pseudo energy”energy” ofof rankedranked rDGrDG structuresstructures ofof manzaminemanzamine AA ((33),), showingshowing thethe firstfirst 500500 outout ofof structure of −32 −[41]2 in conjunction with a randomly generated alignment tensor [36] in order 10001000 structuresstructures generated ( KNOE = =10.0 10.0 Å Å). The). The green green line line shows shows the thecalculations calculations with with NOEs NOEs and andRDCs. RDCs. The dashed The dashed lines to obtain a total of 28 RDCs (the complete list of RDCs of 3 is given in the Supplementary linesat higher at higher energies energies indicate indicate the differenti the differentiabilityability of the of best-fit the best-fit solution solution with with resp respectect to alternate to alternate configurations. configurations. Three Three dif- Materials, Table S4; the dataset included all methine and methylene protons, where the differentferent fc-rDG/DDD fc-rDG/DDD calcul calculationsations were were run run for for3: (1)3: (1)only only NOEs NOEs (black (black line), line), (2) (2)only only RDCs RDCs (inset (inset plot, plot, blue blue line) line) and and 3) latter were used only as the corresponding sums of two C-H dipolar couplings without (3)NOEs NOEs and and RDCs RDCs (green (green line). line). (b ()b Molecular) Molecular structures structures of of manzamine manzamine A A ( (33)) showing showing the superposition conformationalconformational family A (185 structures, topindividual left) and diastereotopicconformational familyassignments). B (142 structures, The green top line right). in Figure The superposition 7a displays ofthe all result DG for family A (185 structures, top left) and conformational family B (142 structures, top right). The superposition of all DG structures identified up to the firstcombined wrong configurationNOE and RDC (421 dataset. structures) Here, is shown the first on wrongthe bottom configuration (right). The ofcentral 3 is structurefrag- structures identified up to the first wrong configuration (421 structures) is shown on the bottom (right). The central fragment ment (green circle) of the DGNo. best-fit 422 (green (lowest circle pseudo in Figureenergy) 7a), structure which is differsplotted fromseparate structuresly without #1 hydrogen to #421 by (superim- the relative (green circle) of the DG best-fit (lowest pseudo energy) structure is plotted separately without hydrogen (superimposed posed partial structures in theconfiguration green circle). of C-12. By analyzing the structures with correct relative configuration, two partial structures in the green circle). conformational families A (185 structures) and B (142 structures) of 3 can easily be identi- The results obtained by the RDCs-only calculation are shown in the inset plot of Fig- fied (Figure 7a). The plot in Figure 7b (top, left) shows the superimposed structures #1 to ure 7a (“best 100”, blue line). The first wrong configuration of 3 is structure No. 58 (blue triangle in Figure 7a, inset plot). This structure differs from structures #1 to #57 by the configuration of C-34. This result, together with the one of the NOE-only calculation

demonstrates the usefulness of the combined use of NOEs and RDCs for the configura- tional analysis of manzamine A (3). It is obvious that this holds for most other configura- tional assignment problems.

3. Methods 3.1. NMR Data The ROE-derived distance restraints for compounds 1a–c were taken from Ref. [51]. The interproton distances were used from the ROESY spectra with a mixing time of 100 ms. For all compounds, three different mixing times (100, 150, and 200 ms) were meas- ured. The total numbers of ROEs are 16 for 1a, 17 for 1b, and 9 for 1c. The complete lists of ROEs for compounds 1a–c are given in the Supplementary Materials (Table S1a–c). In the case of 2 only, 1D NOESY spectra with a mixing time of 500 ms were recorded, which led to 25 NOEs [40]. The complete list of NOEs of 2 is given in the Supplementary Mate- rials (Table S2). The base of the used NOEs for 3 was Figure 2 (“selected NOE correla- tions”) from Ref. [41]. Since the authors only have given the interactions of the protons and not the actual distances between them, we used these nine interactions and added the distances from the crystal structure of 3 [41] to obtain the distance restraints. The complete list of NOEs of 3 is given in the Supplementary Materials (Table S3).

Mar. Drugs 2021, 19, 283 11 of 15

For demonstration purposes, we have, therefore, decided to add artificial residual dipolar couplings (RDCs) to the dataset of 3. The RDCs were calculated from the crystal structure of 3 [41] in conjunction with a randomly generated alignment tensor [36] in order to obtain a total of 28 RDCs (the complete list of RDCs of 3 is given in the Supplementary Materials, Table S4; the dataset included all methine and methylene protons, where the latter were used only as the corresponding sums of two C-H dipolar couplings without individual diastereotopic assignments). The green line in Figure7a displays the result for the combined NOE and RDC dataset. Here, the first wrong configuration of 3 is structure No. 422 (green circle in Figure7a), which differs from structures #1 to #421 by the relative configuration of C-12. By analyzing the structures with correct relative configuration, two conformational families A (185 structures) and B (142 structures) of 3 can easily be identified (Figure7a). The plot in Figure7b (top, left) shows the superimposed structures #1 to #185 (conformational family A) of manzamine A (3) and also the structures #186 to #327 (top, right; conformational family B). All structures with the correct configuration of 3 (420 structures) are shown in Figure7b (bottom). The results obtained by the RDCs-only calculation are shown in the inset plot of Figure7a (“best 100”, blue line). The first wrong configuration of 3 is structure No. 58 (blue triangle in Figure7a, inset plot). This structure differs from structures #1 to #57 by the configuration of C-34. This result, together with the one of the NOE-only calculation demonstrates the usefulness of the combined use of NOEs and RDCs for the configurational analysis of manzamine A (3). It is obvious that this holds for most other configurational assignment problems.

3. Methods 3.1. NMR Data The ROE-derived distance restraints for compounds 1a–c were taken from Ref. [51]. The interproton distances were used from the ROESY spectra with a mixing time of 100 ms. For all compounds, three different mixing times (100, 150, and 200 ms) were measured. The total numbers of ROEs are 16 for 1a, 17 for 1b, and 9 for 1c. The complete lists of ROEs for compounds 1a–c are given in the Supplementary Materials (Table S1a–c). In the case of 2 only, 1D NOESY spectra with a mixing time of 500 ms were recorded, which led to 25 NOEs [40]. The complete list of NOEs of 2 is given in the Supplementary Materials (Table S2). The base of the used NOEs for 3 was Figure2 (“selected NOE correlations”) from Ref. [41]. Since the authors only have given the interactions of the protons and not the actual distances between them, we used these nine interactions and added the distances from the crystal structure of 3 [41] to obtain the distance restraints. The complete list of NOEs of 3 is given in the Supplementary Materials (Table S3).

3.2. DG/DDD The distance geometry calculations were carried out, using a modified version of the DG-II program (kindly provided by Prof. Ruud Scheek, University of Groningen, the Netherlands), linked and interfaced to our ConArch+ a program package. ConArch+ is developed by us for the determination of configurations, using arbitrary combinations of all kinds of NMR data (isotropic and anisotropic NMR parameters) as input [36]. The lists of distance restraints for the DG calculations were constructed from the initial guess primary structures, using bond-lengths-derived upper and lower distance bounds (taken as ±1%), respectively; a detailed description of the DG methodology is given in Ref. [17]. The experimental interproton distances derived from NOESY or ROESY spectra were utilized with default error margins of ±10%, respectively. For each DG simulation, 1000 structures were embedded in 4D space used to initiate the distance bounds driven dynamics (DDD) calculation, beginning at 300 K in 4D for 5 ps (5000 steps, step size 10 fs) and then with a gradual reduction in temperature to 0 K over the next 5 ps (5000 steps, step size 10 fs). This complete procedure was repeated in 3D after reduction of the dimension (4D to 3D). More details on the setup of DG/DDD calculation are given in Refs. [17,36]. Mar. Drugs 2021, 19, x FOR PEER REVIEW 12 of 15

3.2. DG/DDD The distance geometry calculations were carried out, using a modified version of the DG-II program (kindly provided by Prof. Ruud Scheek, University of Groningen, the Netherlands), linked and interfaced to our ConArch+ a program package. ConArch+ is de- veloped by us for the determination of configurations, using arbitrary combinations of all kinds of NMR data (isotropic and anisotropic NMR parameters) as input [36]. The lists of distance restraints for the DG calculations were constructed from the initial guess primary structures, using bond-lengths-derived upper and lower distance bounds (taken as ±1%), respectively; a detailed description of the DG methodology is given in Ref. [17]. The ex- perimental interproton distances derived from NOESY or ROESY spectra were utilized with default error margins of ±10%, respectively. For each DG simulation, 1000 structures were embedded in 4D space used to initiate the distance bounds driven dynamics (DDD) calculation, beginning at 300 K in 4D for 5 ps (5000 steps, step size 10 fs) and then with a Mar. Drugs 2021, 19, 283 gradual reduction in temperature to 0 K over the next 5 ps (5000 steps, step size 1210 of fs). 15 This complete procedure was repeated in 3D after reduction of the dimension (4D to 3D). More details on the setup of DG/DDD calculation are given in Refs. [17,36].

4. Conclusions As demonstrateddemonstrated in the firstfirst partpart ofof thisthis work,work, aa reliablereliable determinationdetermination ofof relativerelative configurationsconfigurations depends strongly strongly on on careful careful quantification quantification of of the the NOE/ROE NOE/ROE data. data. The Thefre- frequentlyquently used used coarse coarse classification classification of of NOEs NOEs/ROEs/ROEs [54] [54 ]as as being being “strong “strong (2.00–2.49 (2.00–2.49 Å)”, “medium“medium (2.50–2.99(2.50–2.99 Å)”,Å)”, oror “weak “weak (3.00–4.00 (3.00–4.00 Å)” Å)” does does not not fulfil fulfil this this demand. demand. This This classifi- classi- cationfication corresponds corresponds to to the the large large NOE-derived NOE-derived distance distance ranges ranges (± (±20%20% and and± ±30%)30%) wewe usedused withwith the palau’amine derivatives derivatives ( (11).). The The results results clearly clearly show show that that the the differentiability differentiability of ± ofdiastereomers diastereomers dramatically dramatically drops drops if the if thedistance distance ranges ranges exceed exceed ±10% 10%considerably. considerably. An- Anotherother problem problem related related with with the thecoarse coarse classifica classificationtion of distances of distances is that is thatvery very similar similar dis- distances (e.g., 2.45 Å and 2.55 Å) could be assigned to different ranges (Figure8), which tances (e.g., 2.45 Å and 2.55 Å) could be assigned to different ranges (Figure 8), which will will adversely affect the search for the correct configuration. adversely affect the search for the correct configuration.

Figure 8. Schematic representation of of the the division division of of NOEs/ROEs NOEs/ROEs intensities intensities into into “rough” “rough” ranges ranges (“weak”,(“weak”, “medium”,“medium”, and and “strong”; “strong”; adapted adapted from from a graphica graphic by by Horst Horst Kessler). Kessler). Although Although the the distances dis- tances B and C are almost identical, they are assigned to two different regions. B and C are almost identical, they are assigned to two different regions.

The applicationapplication ofof thethe fc-rDG/DDD fc-rDG/DDD methodmethod to to three three structural structural types types of of marine marine natural natu- productsral products gave gave insights insights into theinto prospective the prospective breadth breadth of this approach.of this approach. Contrasting Contrasting methods relyingmethods on relying physical on force-fields physical force-fields calculations calculations on the palau’amine on the palau’amine derivatives ( 1derivatives) have proven (1) thathave fc-rDG/DDD proven that fc-rDG/DDD can identify high-energy can identifytrans high-energy-fused five-membered trans-fused five-membered rings when the rings data demandswhen the this data configuration. demands this The configuration. relative configurations The relative of all configurations three palau’amine of all derivatives three pa- werelau’amine unambiguously derivatives determined,were unambiguously irrespective dete ofrmined, the starting irrespective structure of the with starting the method struc- discussed.ture with the In particular,method discussed. the results In for particul tetrabromostyloguanidinear, the results for tetrabromostyloguanidine (1c) are truly astonishing since(1c) are it hastruly eight astonishing stereogenic since centers it has andeight only stereogenic nine ROEs centers were and sufficient only nine for ROEs the correct were configurationalsufficient for the analysis. correct Comparingconfigurational the sizeanalysis. and complexity Comparing of the the size two and molecules complexity (1c and of 2the) as two well molecules as the number (1c and of 2) experimental as well as the restraints, number of one experimental would probably restraints, argue one that would the resultsprobably for argue plakilactone that the H results (2) should for plakilactone even be better H than (2) should those for even tetrabromostyloguanidine be better than those for (tetrabromostyloguanidine1c). In spite of the comparable (1c). In large spite number of the comparable of NOEs (25) large in relation number to of the NOEs size (25) of the in molecule,relation to thethe NOEssize of were the molecule, not able to the unambiguously NOEs were not solve able theto unambiguously relative configuration solve the of 2relative. Three configuration of the four stereogenic of 2. Three centersof the four could stereogenic be determined centers butcould it wasbe determined impossible but to assign the quaternary stereogenic center C-4, which means that the analysis proposes two diastereomers (C-4 epimers) as equivalent solutions. The main difference between 1c and 2 is the conformational flexibility. The obtained results are in accordance with the expectation

that flexibility requires more distance restraints. Opposite to 2, manzamine A (3) is a very complex structure for which only a rather limited number of NOE-derived distances (9) were available. When using the same approach as that for 1 and 2, the configuration of 3 cannot be solved by NOEs only. Only the consideration of RDCs in combination with the NOEs allows for an unambiguous assignment of the five stereogenic centers and the two double bonds. NMR parameters as NOEs/ROEs and RDCs depend on the measurement condi- tions (temperature, type of a solvent, concentration, and pH), and the results of the rDG simulations are valid for the conditions under which the NMR data have been recorded. Nevertheless, with changing NMR restraints, the rDG approach can be used to track conformational changes of the analyte.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/md19060283/s1, Table S1a: NOE data used for 1a, Table S1b: NOE data used for 1b, Table S1c: Mar. Drugs 2021, 19, 283 13 of 15

NOE data used for 1c, Figure S1: rDG structures of compounds 1a–c, Table S2: NOE data used for 2, Figure S2: rDG structure of compound 2, Table S3: NOE data used for 3, Figure S3: rDG structure of compound 3, Table S4: RDC data used for 3. Table S5. rDG best-fit structures (atomic coordinates) for compounds 1a–c, 2, and 3. Author Contributions: Conceptualization, M.K. and S.I.; methodology, M.K. and S.I.; writing— original draft preparation, M.K. and S.I.; writing—review and editing, M.K., S.I., and M.R.; visualiza- tion, M.K. and S.I. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Deutsche Forschungsgemeinschaft (DFG), grant number Re1007/9-1. Acknowledgments: We would like to thank R.M. Scheek (University of Groningen) for providing his modified version of the DG-II program package. We also thank the Center for Scientific Computing (CSC) of the Goethe University, Frankfurt, for granting access to the high-performance computing cluster and providing the CPU time required for the DFT calculations. Conflicts of Interest: The authors declare no conflict of interest.

References and Note 1. Köck, M.; Junker, J. Marine Natural Products—New Ways in the Constitutional Assignment. In New Aspects in Bioorganic Chemistry; Diederichsen, U., Lindhorst, T.K., Wessjohann, L., Westermann, B., Eds.; Wiley-VCH: Weinheim, Germany, 1999; pp. 365–378. 2. Lindel, T.; Junker, J.; Köck, M. COCON: From NMR correlation data to molecular constitutions. J. Mol. Model. 1997, 3, 364–368. [CrossRef] 3. Meiler, J.; Sanli, E.; Junker, J.; Meusinger, R.; Lindel, T.; Will, M.; Maier, W.; Köck, M. Validation of structural proposals by substructure analysis and C-13 NMR chemical shift prediction. J. Chem. Inf. Comput. Sci. 2002, 42, 241–248. [CrossRef] 4. Meiler, J.; Köck, M. Novel methods of automated structure elucidation based on C-13 NMR spectroscopy. Magn. Reson. Chem. 2004, 42, 1042–1045. [CrossRef] 5. Smurnyy, Y.D.; Elyashberg, M.E.; Blinov, K.A.; Lefebvre, B.A.; Martin, G.E.; Williams, A.J. Computer-aided determination of relative stereochemistry and 3D models of complex organic molecules from 2D NMR spectra. Tetrahedron 2005, 61, 9980–9989. [CrossRef] 6. Smurnyy, Y.D.; Blinov, K.A.; Churanova, T.S.; Elyashberg, M.E.; Williams, A.J. Toward more reliable C-13 and H-1 chemical shift prediction: A systematic comparison of neural-network and least-squares regression based approaches. J. Chem. Inf. Model. 2008, 48, 128–134. [CrossRef] 7. Elyashberg, M.; Blinov, K.; Molodtsov, S.; Smurnyy, Y.; Williams, A.J.; Churanova, T. Computer-assisted methods for molecular structure elucidation: Realizing a spectroscopist’s dream. J. Cheminformatics 2009, 1, 3. [CrossRef][PubMed] 8. Elyashberg, M. Identification and structure elucidation by NMR spectroscopy. Trends Anal. Chem. 2015, 69, 88–97. [CrossRef] 9. Buevich, A.V.; Elyashberg, M.E. Synergistic Combination of CASE Algorithms and DFT Chemical Shift Predictions: A Powerful Approach for Structure Elucidation, Verification, and Revision. J. Nat. Prod. 2016, 79, 3105–3116. [CrossRef][PubMed] 10. Troche-Pesqueira, E.; Anklin, C.; Gil, R.R.; Navarro-Vázquez, A. Computer-Assisted 3D Structure Elucidation of Natural Products using Residual Dipolar Couplings. Angew. Chem. Int. Ed. 2017, 56, 3660–3664. [CrossRef][PubMed] 11. Navarro-Vázquez, A.; Gil, R.R.; Blinov, K. Computer-Assisted 3D Structure Elucidation (CASE-3D) of Natural Products Combin- ing Isotropic and Anisotropic NMR Parameters. J. Nat. Prod. 2018, 81, 203–210. [CrossRef] 12. Milanowski, D.J.; Oku, N.; Cartner, L.K.; Bokesch, H.R.; Williamson, R.T.; Sauri, J.; Liu, Y.Z.; Blinov, K.A.; Ding, Y.Q.; Li, X.C.; et al. Unequivocal determination of caulamidines A and B: Application and validation of new tools in the structure elucidation tool box. Chem. Sci. 2018, 9, 307–314. [CrossRef][PubMed] 13. Castro, S.J.; Garcia, M.E.; Padron, J.M.; Navarro-Vázquez, A.; Gil, R.R.; Nicotra, V.E. Phytochemical Study of Senecio volck- mannii Assisted by CASE-3D with Residual Dipolar Couplings and Isotropic H-1/C-13 NMR Chemical Shifts. J. Nat. Prod. 2018, 81, 2329–2337. [CrossRef][PubMed] 14. Menna, M.; Imperatore, C.; Mangoni, A.; Della Sala, G.; Taglialatela-Scafati, O. Challenges in the configuration assignment of natural products. A case-selective perspective. Nat. Prod. Rep. 2019, 36, 476–489. [CrossRef][PubMed] 15. Vögeli, B. The nuclear Overhauser effect from a quantitative perspective. Prog. Nucl. Magn. Reson. Spectrosc. 2014, 78, 1–46. [CrossRef] 16. Neuhaus, D.; Williamson, M.P. The Nuclear Overhauser Effect in Structural and Conformational Analysis; Wiley-VCH: Weinheim, Ger- many, 2000. 17. Köck, M.; Reggelin, M.; Immel, S. The Advanced Floating Chirality Distance Geometry Approach—How Anisotropic NMR Parameters Can Support the Determination of the Relative Configuration of Natural Products. Mar. Drugs 2020, 18, 330. [CrossRef] 18. Crippen, G.M.; Havel, T.F. Distance Geometry and Molecular Conformation; John Wiley & Sons: Hoboken, NJ, USA, 1988. Mar. Drugs 2021, 19, 283 14 of 15

19. de Vlieg, J.; Scheek, R.M.; van Gunsteren, W.F.; Berendsen, H.J.C.; Kaptein, R.; Thomason, J. Combined procedure of distance geometry and restrained molecular dynamics techniques for protein structure determination from nuclear magnetic resonance data: Application to the DNA binding domain of lac repressor from Escherichia coli. Proteins Struct. Funct. Genet. 1988, 3, 209–218. [CrossRef] 20. Havel, T.F.; Kuntz, I.D.; Crippen, G.M. The theory and practice of distance geometry. Bull. Math. Biol. 1983, 45, 665–720. [CrossRef] 21. Kaptein, R.; Boelens, R.; Scheek, R.M.; van Gunsteren, W.F. Protein structures from NMR. Biochemistry 1988, 27, 5389–5395. [CrossRef] 22. Scheek, R.M.; van Gunsteren, W.F.; Kaptein, R. Molecular dynamics simulation techniques for determination of molecular structures from nuclear magnetic resonance data. Methods Enzymol. 1989, 177, 204–218. 23. Köck, M.; Junker, J. Determination of the relative configuration of organic compounds using NMR and DG: A systematic approach for a model system. J. Mol. Model. 1997, 3, 403–407. [CrossRef] 24. Köck, M.; Schmidt, G.; Seiple, I.B.; Baran, P.S. Configurational Analysis of Tetracyclic Dimeric Pyrrole-Imidazole Alkaloids Using a Floating Chirality Approach. J. Nat. Prod. 2012, 75, 127–130. [CrossRef] 25. Reggelin, M.; Hoffmann, H.; Köck, M.; Mierke, D.F. Determination of Conformation and Relative Configuration of a Small, Rapidly Tumbling Molecule in Solution by Combined Application of NOESY and Restrained MD Calculations. J. Am. Chem. Soc. 1992, 114, 3272–3277. [CrossRef] 26. Mierke, D.F.; Reggelin, M. Simultaneous determination of conformation and configuration using distance geometry. J. Org. Chem. 1992, 57, 6365–6367. [CrossRef] 27. Reggelin, M.; Köck, M.; Conde-Frieboes, K.; Mierke, D. Determination of the Relative Configuration by Distance Geometry Calculations with Proton-Proton Distances from NOESY Spectra. Angew. Chem. Int. Ed. 1994, 33, 753–755. [CrossRef] 28. Köck, M.; Junker, J. How many NOE derived restraints are necessary for a reliable determination of the relative configuration of an organic compound? Application to a model system. J. Org. Chem. 1997, 62, 8614–8615, Erratum in 1998, 63, 2409–2409. [CrossRef] 29. Kaptein, R.; Zuiderweg, E.R.P.; Scheek, R.M.; Boelens, R.; van Gunsteren, W.F. A protein structure from nuclear magnetic resonance data. The lac repressor headpiece. J. Mol. Biol. 1985, 182, 179–182. [CrossRef] 30. Clore, G.M.; Gronenborn, A.M.; Brünger, A.T.; Karplus, M. Solution Conformation of a Heptadecapeptide Comprising the DNA-Binding Helix F of the Cyclic AMP Receptor Protein of Escherichia coli—Combined Use of H-1 Nuclear Magnetic-Resonance and Restrained Molecular Dynamics. J. Mol. Biol. 1985, 186, 435–455. [CrossRef] 31. Weber, P.L.; Morrison, R.; Hare, D. Determining stereo-specific H-1 nuclear magnetic-resonance assignments from distance geometry calculations. J. Mol. Biol. 1988, 204, 483–487. [CrossRef] 32. Holak, T.A.; Gondol, D.; Otlewski, J.; Wilusz, T. Determination of the complete three-dimensional structure of the trypsin inhibitor from squash seeds in aqueous solution by nuclear magnetic resonance and a combination of distance geometry and dynamical simulated annealing. J. Mol. Biol. 1989, 210, 635–648. [CrossRef] 33. Havel, T.F. An evaluation of computational strategies for use in the determination of protein-structure from distance contraints obtained by nuclear magnetic resonance Prog. Biophys. Mol. Biol. 1991, 56, 43–78. [CrossRef] 34. van Schaik, R.C.; Berendsen, H.J.C.; Torda, A.E.; van Gunsteren, W.F. A Structure Refinement Method Based On Molecular- Dynamics In Four Spatial Dimensions. J. Mol. Biol. 1993, 234, 751–762. [CrossRef] 35. A pseudo-atom is a virtual atom whose position is the geometrical average of two (isochronous methylene groups or ortho- /meta-CHs in aromatic systems), three (methyl group), or more hydrogen atoms. A pseudo-atom is applied if an assignment to one proton is not possible and therefore its position is used in order to consider the distance information. 36. Immel, S.; Köck, M.; Reggelin, M. Configurational Analysis by Residual Dipolar Coupling Driven Floating Chirality Distance Geometry Calculations. Chem. Eur. J. 2018, 24, 13918–13930. [CrossRef] 37. Immel, S.; Köck, M.; Reggelin, M. Configurational analysis by residual dipolar couplings: A critical assessment of diastereomeric differentiabilities. Chirality 2019, 31, 384–400. [CrossRef] 38. Reggelin, M.; Immel, S. Configurational Analysis by Residual Dipolar Couplings: Critical Assessment of “Structural Noise” from Thermal Vibrations. Angew. Chem. Int. Ed. 2021, 60, 3412–3416. [CrossRef] 39. Kinnel, R.B.; Gehrken, H.P.; Scheuer, P.J. Palauamine—A Cytotoxic and Immunosuppressive Hexacyclic Bisguanidine Antibiotic from the Sponge Stylotella-Agminata. J. Am. Chem. Soc. 1993, 115, 3376–3377. [CrossRef] 40. Di Micco, S.; Zampella, A.; D’Auria, M.V.; Festa, C.; De Marino, S.; Riccio, R.; Butts, C.P.; Bifulco, G. Plakilactones G and H from a marine sponge. Stereochemical determination of highly flexible systems by quantitative NMR-derived interproton distances combined with quantum mechanical calculations of 13C chemical shifts. Beilstein J. Org. Chem. 2013, 9, 2940–2949. [CrossRef] 41. Sakai, R.; Higa, T. Manzamine A, a Novel Antitumor Alkaloid from a Sponge. J. Am. Chem. Soc. 1986, 108, 6404–6405. [CrossRef] 42. Köck, M.; Grube, A.; Seiple, I.B.; Baran, P.S. The pursuit of Palau’amine. Angew. Chem. Int. Ed. 2007, 46, 6586–6594. [CrossRef] 43. Kinnel, R.B.; Gehrken, H.P.; Swali, R.; Skoropowski, G.; Scheuer, P.J. Palau’amine and its congeners: A family of bioactive bisguanidines from the marine sponge Stylotella aurantium. J. Org. Chem. 1998, 63, 3281–3286. [CrossRef] 44. Halford, B. Palau’amine reconsidered. Chem. Eng. News 2007, 85, 12–12. [CrossRef] 45. Grube, A.; Köck, M. Structural Assignment of Tetrabromostyloguanidine: Does the Relative Configuration of the Palau’amines Need Revision? Angew. Chem. Int. Ed. 2007, 46, 2320–2324. [CrossRef][PubMed] Mar. Drugs 2021, 19, 283 15 of 15

46. Buchanan, M.S.; Carroll, A.R.; Addepalli, R.; Avery, V.M.; Hooper, J.N.A.; Quinn, R.J. Natural products, stylissadines A and B, specific antagonists of the P2X(7) receptor, an important inflammatory target. J. Org. Chem. 2007, 72, 2309–2317. [CrossRef] 47. Kobayashi, H.; Kitamura, K.; Nagai, K.; Nakao, Y.; Fusetani, N.; van Soest, R.W.M.; Matsunaga, S. Carteramine A, an inhibitor of neutrophil chemotaxis, from the marine sponge Stylissa carteri. Tetrahedron Lett. 2007, 48, 2127–2129. [CrossRef] 48. Peplow, M. Chemists crack complex compound. Nature 2010, 463, 14–15. [CrossRef][PubMed] 49. Seiple, I.B.; Su, S.; Young, I.S.; Lewis, C.A.; Yamaguchi, J.; Baran, P.S. Total Synthesis of Palau’amine. Angew. Chem. Int. Ed. 2010, 49, 1095–1098. [CrossRef][PubMed] 50. Seiple, I.B.; Su, S.; Young, I.S.; Nakamura, A.; Yamaguchi, J.; Jørgensen, L.; Rodriguez, R.A.; O’Malley, D.P.; Gaich, T.; Köck, M.; et al. Enantioselective Total Syntheses of (−)-Palau’amine, (−)-Axinellamines, and (−)-Massadines. J. Am. Chem. Soc. 2011, 133, 14710–14726. [CrossRef] 51. Lanman, B.A.; Overman, L.E.; Paulini, R.; White, N.S. On the structure of Palau’amine: Evidence for the revised relative configuration from chemical synthesis. J. Am. Chem. Soc. 2007, 129, 12896–12900. [CrossRef] 52. Reinscheid, U.M.; Köck, M.; Cychon, C.; Schmidts, V.; Thiele, C.M.; Griesinger, C. The Absolute Configuration of Dibro- mopalau’amine. Eur. J. Org. Chem. 2010, 6900–6903. [CrossRef] 53. Lindel, T.; Jacquot, D.E.N.; Zöllinger, M.; Kinnel, R.B.; McHugh, S.; Köck, M. Study on the absolute configuration of (−)- palau’amine. Tetrahedron Lett. 2010, 51, 6353–6355. [CrossRef] 54. Wüthrich, K. NMR of Proteins and Nucleic Acids; Wiley: New York, NY, USA, 1986.