Assigning regioisomeric or diastereoisomeric relations of problematic trisubstituted double-bonds through heteronuclear 2D selective J-resolved NMR spectroscopy

Although one of the first 2D NMR methods, but so far neglected, selective J-resolved NMR spectroscopy offers a unique opportunity to help organic chemists in structure elucidation, avoiding natural and non-natural product misassignments. This NMR method indeed allowed us to unambiguously and simply assign (natural) product structures exhibiting trisubstituted unsaturations. For example, as demonstrated here, the isomeric aurone, flavone, coumarin and isocoumarin structures could easily be distinguished; regioisomer assignments in furans could be solved, as well as isomerisms in compounds containing trisubstituted double-bonds (aurones, lactones, divinyl ketones).


Introduction
In natural product research, structural and/or stereochemical assignments are obviously the key step, providing the foundation for a whole set of disciplines, from synthesis to biology, biochemistry or ecology, with oen the production of new therapeutic agents as the most practical consequences. Despite the current sophisticated techniques available, assignments oen prove delicate and sometimes cumbersome. The number of natural product structural misassignments reported each year is surprisingly high (Fig. 1). 1 These numbers are probably underestimated, since those encountered misassignments are frequent and only revealed through total synthesis enterprises. 2 Furthermore, the efforts and associated costs required to correct structures and appropriately reassign them could be quite signicant and higher than those associated with the initial structure, especially in total synthesis endeavors. 3 Besides the increased enrolments for total syntheses, wrong assignments might have dramatic effects due to the different biological activities the compounds could exhibit.
What was observed for natural products is also true in any synthetic efforts addressing compounds in which stereo-and/or regioisomers could be obtained, especially when only one is formed. In this area, the lack of re-synthesis minimizes the discovery of misassignments. Therefore, it is clear that, since the number of incorrectly determined structures of new products, natural or not, is quite large, reducing errors is clearly an important and useful challenge. Several routes are currently explored, among which are computer-assisted structure elucidation (CASE) method, 4 NMR shi calculations combined with probability analyses 5 or articial neural network pattern recognition, 6 combination of calculations with various spectroscopies, notably circular dichroism, 7 and even atomic force microscopy imaging at atomic resolution. 8 Despite these recent new developments, structure elucidation of new compounds, including natural products, heavily relies on two-dimensional NMR experiments. Correlation spectroscopy (COSY) and heteronuclear multiple and single bond correlation (HMBC/HSQC) detect interactions (couplings) of protons, respectively, with adjacent protons (from 2 to 4 bonds) or with neighbor carbon atoms (from 2 to 3 bonds). Both techniques provide connectivity between atoms. Stereochemistry and conformation of molecules are usually deduced from two-dimensional nuclear Overhauser enhancement spectroscopy (NOESY). Based on spin relaxation processes via dipolar coupling interacting through space, this method is able to detect protons that are in close proximity.
Facing various structural problems either in the total synthesis of natural products 9 or in the development of new methods, 10 we recently applied a neglected but simple NMR method, which successfully helped us to solve connectivity and/ or stereochemistry problems. Being very efficient in our hands, we share here our results gained by applying the selective Jresolved NMR method during the process of structure elucidation with some typical examples.

Results and discussion
The (selective) J-resolved NMR method Introduced in 1976 by Ernst et al., 11 the 2D J-resolved NMR experiments (JRES) are used to separate chemical shis and J-couplings into two independent dimensions. 12, 13 The spin coupling may be homocoupling, usually between protons, or heterocoupling, essentially between 1 H and 13 C, although 1 H and 31 P have also been reported. 14 The 2D spectra exhibit in one dimension the equivalent of a broad range decoupled proton or carbon spectrum and in the other the coupling, either 1 H-1 H or 13 C-1 H. Although one of the rst 2D NMR experiments, JRES has been supplanted by the more recent COSY and HMBC methods in chemistry. These NMR techniques could also provide information about such coupling constants, but in a less practical way (see below).
Interestingly, the JRES method is currently re-gaining interest for the study of complex mixtures, 15 mostly in biology and medicine as a tool for metabolic screening 16 and metabolomics 17 as well as for biomarker identication, especially for cancer monitoring. 18 Stereochemical double-bond assignment is still a problem for the trisubstituted ones, 19 unfortunately quite common in natural products. Classically solved for disubstituted doublebonds with the E or Z 3 J coupling constants ( 1 H NMR), the case of double-bonds carrying a single hydrogen is far from obvious. Coupling between vinylic and allylic hydrogens ( 4 J) can be measured, but the corresponding E and Z values being low and close to each other do not allow unambiguous assignments. 20 Due to the parallelism between 1 H-1 H and 13 C-1 H coupling constants, the E or Z geometry could also be reected by the 3 J or 2 J ( 13 C-1 H) coupling constants. 21 The usual way to measure them is to record 13 C coupled 1 H NMR. 22,23 However, this technique leads to very complex spectra, due to overlapping of signals and to the complexity of multiplets, which may require computational method to solve out. Furthermore, rather long measurement times are needed and typically, signal-to-noise ratio is poor for such spectra.
Some of these aspects could also occur for the classic JRES method, but to a lower extent as all n J ( 13 C-1 H) coupling constants are expressed in 2D. However, the heteronuclear selective J-resolved NMR experiment (SELJRES) 12b elegantly solves this problem by inverting the resonance of the considered proton allowing to remove from the 2D map all other coupling constants excepted those of the selectively inverted proton: 24 the required coupling can thus be easily available in one of the two dimensions.
Such stereochemical double-bond problems are oen encountered in various families of natural products or in synthesis of unsaturated compounds, but only four typical examples will be detailed here.

Regioisomer assignment of avonoids
Among avonoids, 25 coumarins, isocoumarins, aurones, and avones constitute a sub-family contributing to the pigmentation of owers and fruits, but also exhibiting interesting pharmaceutical activities. 26 These four subclasses are structurally related, each one being isomeric to the others around a trisubstituted double-bond (Scheme 1). Due to their similarity, the structural elucidation of these natural products remains a challenge, and already led to numerous misassignments, from which only a few were detected and induced structural revision. 9a,27 Facing such problem, weas othersrelied on total synthesis of each possible isomer, 9a but this gave us the opportunity to apply heteronuclear SELJRES and to nd a useful and quick distinction between these isomers. Indeed, the couplings between the vinylic proton and the carbonyl carbon (red C in Scheme 1), the vinylic carbon (blue C) and the ortho carbon on the phenyl substituent (green C) in each series proved typical. 28 Coumarin 1 showed a weak 2 J Cvinyl-H coupling constant of 1.1 Hz compared to the three other isomers 2-4 ( 2 J Cvinyl-H ¼ 5.1 AE 0. 4 Hz), probably reecting the lack of oxygenated substituent on the double-bond. Isocoumarin 2 could easily be distinguished from the three other possible isomers by the lack of coupling between the vinylic proton and the carbonyl carbon. Flavone 3 and aurone 4 usually cannot be distinguished between themselves by their 1 H NMR spectra due to a very similar shi of their vinylic proton (6.8-6.9 ppm) and other chemical shis. However, they could be unambiguously distinguished by the presence or not of a coupling between the vinylic proton and the ortho carbon of the aromatic substituent: aurone 4 exhibits a 3 J coupling (6.1 Hz) while avone 3 can only have a 4 J coupling, almost not detectable (Scheme 1).

Regioisomer assignment of trisubstituted furans
Furans constitute an important class of aromatic compounds, found in many natural products. 29 Therefore, a large number of synthetic methods have been developed to build such ve membered rings. 30 Our group recently contributed to this area, achieving mild and convenient access to trisubstituted furans. 31 During this study, as for furan natural product structural assignments, the regioselectivity has to be determined.
With a single proton, such isomeric furans may be difficult to distinguish. When substituents in positions 2 and 5 do not carry hydrogen, there is no possibility to measure the 3 J ( 1 H-1 H) coupling with the furanic proton. Since the situation of the furanic proton can be compared to those described in the previous section, we envisaged to solve this question by applying the SELJRES method. Indeed, the single proton should exhibit 2 J coupling with the furanic carbon (red or blue C in Scheme 2), reecting the proximity between the two atoms.
To look at this hypothesis, ve different trisubstituted furans 5a-e were submitted to JRES by selecting the frequency of the furanic proton for acquisition. The analysis showed a 2 J coupling constant around 7 to 8 Hz between the furanic proton and its adjacent b-carbon (red or blue C on Scheme 2). Interestingly, no 3 J coupling was expressed with the g-furanic carbon. This combination thus allowed for the unambiguous determination of the substitution pattern of furans. Furthermore, the analysis can be done on a mixture of two isomers as it was shown with 5a and 5b.

Diastereomer assignment of aurones and alkylidenelactones
As for the previous examples, the E or Z stereochemical assignment of trisubstituted double-bonds may not be obvious. This could also be true for methylenic systems. Aurones are typical of the former case, and our synthetic endeavor 31 again provided us the opportunity to look at the JRES potential. Naturally occurring aurones usually exhibit a Z conguration of the double-bond, but such type of compounds can isomerize under irradiation or on silica gel. 32 Alkylidene lactones also offer useful examples, 9a,10b with the methylene lactones being typical of the latter case. Both allowed interesting comparisons.
As before, SELJRES allowed us to get 2 J or 3 J ( 13 C-1 H) coupling constants, from which those between the vinylic proton and the carbonyl or vinylic carbons simply proved distinctive for each stereoisomer. The data collected for E or Z aurones 4 and 4a-d and related compound 4e clearly showed a correlation between such coupling constants and the stereochemistry of the double-bond (Scheme 3). The Z always exhibited 3 J and/or 2 J ( 13 C-1 H) values of 3-4 Hz, while the E isomer exhibited values of 7-8 Hz. Such clear-cut differences could obviously be fruitful in assigning stereochemistry and would have predictive value. This was conrmed with the alkylidene lactone series, the stereochemistry of which being already assigned by other techniques. Indeed, even in the simple methylene lactone 6, the same set of values was observed for the E and Z protons for the 3 J ( 13 C-1 H) coupling (3.3 and 7.8 Hz, respectively) (Scheme 3). In other alkylidene lactones (Z)-7 and 8, similar trends could be observed with 3 J ( 13 C-1 H) values of 3.5 AE 0.4 Hz noticed in Z isomers and 6.2 Hz for the E isomer (E)-7. In some specic cases, no 3 J value was detected as for the bromomethylene furanone 9, structurally very close to the mbrolide natural product family. 33 However, a 2 J coupling constant of 4.3 Hz was expressed in SELJRES, which could be attributed to the Z isomer by comparison with lactones (E)-7, (Z)-7 and 8.
It is worth mentioning that correlations from 1 H-coupled 13 C spectra with double-bond diastereoisomery have already been observed and compiled from around forty examples in the seventies (Table 1). 21 The listed values are surprisingly large and similar to classical 1 H-1 H coupling constants, the E 3 J ( 13 C-1 H) coupling constants being always larger than the Z ones, but with a strong inuence upon hybridization of the selected carbon. Although the general tendency is the same, with larger coupling constants for E relative to Z proton, the 3 J SELJRES values we collected are far different from those already reported: between 3-4 Hz vs. 4-10 Hz for Z and 7-8 Hz vs. 9-17 Hz for E (Scheme 3 vs. Table 1) and, in our study, no inuence of the hybridization has been noticed (see compound 4 vs. 4e). More surprisingly, all our measured SELJRES values should correspond to cis proton/Z systems according to Table 1 data.

Diastereomer assignment of divinyl ketones
For a program devoted to gold catalysis and to the synthesis of cyclopentenone derivatives via the Nazarov reaction, 34 we obtained various divinyl ketones, 35 for which we had to assign the double-bond stereochemistry. Indeed, the reaction we developed led to divinyl ketones in which one doublebond stereochemistry was not dened by the synthetic pathway. In a rst attempt, 2D NOESY was used, leading to some conclusions. However, based on the absence of correlation and due to the exibility of such structures, assignments were suspicious. Fortunately, we were able to obtain crystal structures allowing us to unambiguously determine the stereochemistry. This helped us to look again at the correlations between 13 C-1 H SELJRES values and doublebond stereochemistry. Among the divinyl ketones we prepared, different examples were selected and submitted to SELJRES NMR experiments.
A series of O-substituted divinyl ketones (10a-c and 11) was rst evaluated, including one (10a) with X-ray crystallography structure (see ESI †) and one ((E)-10b and (Z)-10b) with both diastereomers. Their 3 J and 2 J ( 13 C-1 H) coupling constants, respectively, between the vinylic proton and the carbonyl carbon or the adjacent sp 2 carbon were extracted (Scheme 4). Values similar to those already gained (see preceding section and Scheme 3) were obtained, with 3 J and 2 J ( 13 C-1 H) coupling constants of 8 AE 0.2 Hz for E isomers (10a and (E)-10b) and 3.6 AE 0.2 Hz for Z compounds ((Z)-10b, 10c and 11). These two sets of constants clearly depend on the geometry of the system and can easily be exploited to unambiguously assign double-bond geometry.
We also looked at the coupling constants of the doublebonds having C-substituents (10a-c, 11, and 12a). Once again, the geometry of each diastereomer could be easily determined by SELJRES with 3 J ( 13 C(O)-1 H) values around 13 Hz for trans isomers (H E ) and between 6.4 and 7.3 Hz for cis isomers (H Z ). However, in the case of C-substituted divinyl ketones, the 2 J ( 13 C-1 H) coupling constants did not follow the tendency of 3 J ( 13 C-1 H) values: only weak values were measured (1.8 AE 0. 8 Hz) without direct correlation with the E and Z nature of olens. Scheme 3 Simple E/Z assignment of aurones and lactone derivatives by SELJRES spectral analysis.  Being well represented in the literature due to their good reactivity in the Nazarov reaction, 36 divinyl ketones carrying an ester group (12a-d) were also investigated. In our case, only E isomers were available. SELJRES spectra showed that their 3 J ( 13 C(O)-1 H) (trans values) and 3 J ( 13 CO 2 Me-1 H) (cis values) coupling constants lay, respectively, between 9 to 10 Hz and around 8 Hz for compounds 12a and 12c-d. Working on the Nazarov reaction, Frontier et al. have also been interested in assigning the stereochemistry of divinyl ketones, and they have reported the 3 J ( 13 C-1 H) extracted from 13 C non-decoupled 1 H NMR spectra, for Z and E isomers analogs. 36e In this report, the coupling constants for E isomers correlated perfectly with the range of values obtained for 12a and 12c-d (9- 10 Hz). However, the 3 J ( 13 CO-1 H) (cis values, 4.6-7.2 Hz) and 3 J ( 13 CO 2 Me-1 H) (trans values, 10.7-12. 8 Hz) for Z isomers are far different, such as in 12b, which conrmed the Z nature of the considered olen.

Conclusion
In this work, we have demonstrated that SELJRES NMR spectra gave direct and unambiguous access to the regioisomerism of various types of molecules, such as avonoids and furans. SELJRES NMR thus offers a simple and easy method to rapidly determine the E and Z diastereochemistry of trisubstituted doublebonds in aurone, alkylidene lactone or divinyl ketone derivatives. Indeed, the 3 J ( 13 C-1 H) coupling constants extracted from SELJRES are always larger for the E isomers, between 7-13 Hz, than for the Z isomers (3-8 Hz) depending on the substrate nature.

General informations
Proton ( 1 H NMR) and carbon ( 13 C NMR) nuclear magnetic resonance spectra were recorded on a 500-or 600 MHz Bruker Avance II spectrometer equipped with a 5 mm DCH dual cryoprobe, with a z-gradient and operating at 500.130 MHz for 1 H and 125.758 MHz for 13 C. The solvent peak was used as reference value, for 1 H NMR: CHCl 3 ¼ 7.26 ppm or benzene-d ¼ 7.16 ppm and for 13 C NMR: CHCl 3 ¼ 77.16 ppm or benzene-d ¼ 128.06 ppm. For 1 H NMR, data are presented as follows: chemical shi (ppm), multiplicity (s ¼ singlet, d ¼ doublet, t ¼ triplet, q ¼ quartet, m ¼ multiplet, b ¼ broad), coupling constants (J/Hz), integration, and attribution. For 13 C NMR, data are presented as follows: chemical shi (ppm) and attribution. For 13 C-{ 1 H} selective J-resolved NMR, data are presented as follows: (chosen proton), 13 C chemical shi (ppm) and coupling constant (J/Hz) or folded 1 J ( 13 C-1 H), actually the acquisition method prevented the direct detection of the 1 J ( 13 C-1 H) constant that is folded. Remark: despite all discussions about trisubstituted double-bonds, during the characterization it was observed that, for a chosen proton, sometimes the 2 J ( 13 C-1 H) is smaller than its 3 J ( 13 C-1 H). 37 Assignments were determined either on the basis of unambiguous chemical shis or coupling patterns, COSY, HSQC, HMBC, and NOESY experiments, to fully interprete spectra for related compounds. The pulse sequence used for measuring n J ( 13 C-1 H) coupling constants is similar to the seljresqfsp sequence of the Bruker Scheme 4 3 J and 2 J ( 13 C-1 H) SELJRES values of variously substituted divinyl ketones 10-12. library. Experiments were acquired at 293 K with a relaxation delay of 3 s. Acquisition parameters were adjusted when necessary, but typically spectral windows were set to 25 kHz (200 ppm) for 13 C and 50 Hz for n J ( 13 C-1 H). Data size was at least 16k points in the direct dimension and 32 points in the indirect dimension. Broadband decoupling during acquisition was achieved by a composite pulse sequence Waltz16. Refocusing of 13 C magnetization during evolution period was performed with an adiabatic shape pulse Crp60comp.4 of 2 ms. Simultaneously, a selective 180 pulse focused on the chosen proton frequency was achieved with a refocusing band-selective pulse (REBURP) of 80 ms. Melting points were recorded on a Melting Point Apparatus SMP3 and are uncorrected. Infrared spectra were recorded neat with Hewlett-Packard spectrometer. Wavelengths of maximum absorbance (n max ) are quoted in wave numbers (cm À1 ). High resolution mass spectra (HR-MS) data were recorded on a microTOF spectrometer equipped with orthogonal electrospray interface (ESI). The parent ions [M + Na] + [M + H] + is quoted. Analytical thin-layer chromatography (TLC) was carried out over silica gel 60 F254 plates or basic alumina (63-200 mm) with visualization by ultraviolet light, cerium ammonium molybdate (CAM) or potassium permanganate dip (KMnO 4 ). Flash-column chromatography was carried out using silica gel 60 (40-63 mm) and the procedure included the subsequent evaporation of solvents in vacuo. Solvents were puried using standard means. 1,2-Dichloroethane (DCE), benzene and triethylamine (NEt 3 ) were distilled from CaH 2 ; pyridine (py) was distilled from potassium carbonate; diisopropylamine (DIPA) and N,N-diisopropylethylamine (DIPEA) were distilled from KOH; all under an argon atmosphere. Tetrahydrofuran (THF), acetonitrile (CH 3 CN) and dichloromethane (DCM) were dried using Glasstechnology GT S100 device. AuCl (Premion grade, 99.99%) and NaAuCl 4 $2H 2 O (Premion grade, 99.99%) were purchased from Alfa Aesar. AgSbF 6 (98%) and PdCl 2 (dppf)$DCM were purchased from STREM Chemicals. IPrAuCl, PdCl 2 (PPh 3 ) 2 (98%), HNTf 2 and CuI (98%) were purchased from Aldrich. AgNTf 2 was prepared from HNTf 2 (Aldrich) and Ag 2 CO 3 . 38 All phosphinegold(I) chloride precatalysts were prepared by reduction of NaAuCl 4 $2H 2 O with thiodiethanol and subsequent addition of the appropriate phosphine. 39 PPh 3 AuNTf 2 catalyst was prepared from the corresponding triphenylphosphinegold(I) chloride and AgNTf 2 . 40 All other chemicals were used as received. (1) 4-Phenyl-2H-chromen-2-one (1) was obtained in 91% yield over 2 steps from chromane-2,4-dione, using reported procedures (see ESI †). 41