Methoxy 13 C NMR Chemical Shift as a Molecular Descriptor in the Structural Analysis of Flavonoids and Other Phenolic Compounds

Abstract

The interpretation of nuclear magnetic resonance (NMR) chemical shifts plays an increasingly important role in the structure elucidation of flavonoids. Despite the fact that the 3H signal of the methoxy (OMe) group appears as a singlet in the ¹H NMR spectrum, its chemical shift is not reliably indicative of the OMe-substitution in the aromatic rings, unlike its ¹³C NMR chemical shift, which can be useful for the determination of the molecular structure of flavonoids. For example, the chemical shift of ortho-substitutions is at 56 ± 1 ppm for flavonoids having one OMe-substituent in either the ortho or orthó position and 61.5 ± 2 ppm for those having substituents in both ortho positions. The nuclear Overhauser effect spectroscopy (δ_H-OMe→ortho-aryl-H or adjacent substituent) and heteronuclear multiple-bond correlation (δ_H-OMe→ipso-aryl-C) data have been widely employed for the structure analysis of flavonoids and to identify the site of OMe-substitution. Thus, OMe chemical shifts can act as molecular descriptors. However, we have noticed that there are several examples that are not in accordance with this generalization, and a reinvestigation of such reported data seems to be relevant for reconsideration of assigned structures, and a few examples are cited herein.

Flavonoids possessing methoxy (OMe)/hydroxy/glycosyloxy groups are widespread throughout the plant kingdom and are important from the biological viewpoint presenting numerous benefits to human health, including antioxidant, antiinflammatory, antibacterial, antiproliferative, antihypertensive, antithrombotic, antiallergic, anticarcinogenic, antiviral, and anticoronoviral effects.^1–15 The structure elucidation of flavonoids from 1960 to the mid-1970s was facilitated by the combined application of ultraviolet (UV)–visible absorption spectroscopy in combination with various shift reagents, mass spectrometry (MS), and ¹H nuclear magnetic resonance (NMR) spectroscopic analysis.^16,17 The structures were proposed as the best fit based on infrared, UV, ¹H NMR, and mass spectral data. This was indispensable in the characterization of flavonoids presenting several positional isomers, but they cannot be unequivocally distinguished by their MS and ¹H NMR spectral data by direct comparison of chemical shifts and coupling constants.

In most cases, singlets integrating for 3H, appearing at around 3.9 ± 0.3 ppm in the ¹H NMR spectrum, correspond to an aromatic OMe group. Thus, a number of OMe groups can be determined from this spectral region, but this may be complex due to glycosides, the sugar proton resonances of which appear in this spectral region. In earlier studies, preparation of methyl ethers and/or acetyl derivatives was the preferred method to ascertain the number of hydroxyl groups in a phenolic compound under investigation as methylation/acetylation was accompanied by increased lipophilicity, which makes these derivatives easily soluble in deuterochloroform.^16,17 However, the precise locations of the OH/OMe, in most cases, pose a structural determination problem in characterizing these polyphenolic flavonoids. Sometimes, spectral complexity or negligence leads even experienced and skillful researchers to make mistakes, reporting erroneous structures for newly isolated compounds (vide infra).

The major advancement in flavonoid structural techniques was the application of ¹³C NMR spectroscopy and, to the best of our knowledge, Professor Joseph-Nathan's publication in 1974 was the first article to report the unequivocal assignment of all carbon resonances for flavone by comparison of the ¹³C NMR spectra of 5,6,7,8-tetradeuteroflavone, 2′,3′,4′,5′,6′-pentadeuteroflavone, and 4′-bromoflavone, and by the use of Eu(dpm)₃ shift reagent. ¹⁸ Owing to the intrinsic feature of the ¹³C chemical shift, which is spread over a wide spectral window (up to 220 ppm), its calculation has proved to be particularly valuable in the structural analysis of flavonoids. Joseph-Nathan's publication was followed by several others and the literature, and ¹³C NMR features were subsequently reviewed in 1981 ¹⁹ and 1982, ²⁰ including a monograph by Agrawal. ²¹

These publications provide detailed information about the descriptors that characterize the molecular skeleton of flavonoids, in addition to the use of ¹³C NMR data to describe the molecular structure. It can be noticed that these data yield rich information about the molecular structure and are sufficiently sensitive to detect small differences in the molecule. These differences are measured by the variation in chemical shifts, and their values can be used to correlate with the chemical structure due to their dependence on the chemical environment.

It is worthwhile to mention that the assignment of the substitution position in OH/OMe-substituted flavonoids has been commonly made by ¹H and/or ¹³C NMR spectroscopy. However, such determination is not always trivial when no reference spectra are available. For instance, we isolated taxifolin (3,5,7,3´,4´-pentahydroxyflavanone, 1), dihydromyricetin (3,5,7,3´,4´,5´-hexahydroxyflavanone, 2), and the related mono-C-methylated compounds: cedeodarin (6-C-methyl-3,5,7,3´, 4´-pentahydroxyflavanone or-6-C-methyl-taxifolin, 3), and cedrin (6-C-methyl-3, 5,7,3´,4´,5´-hexahydroxyflavanone or 6-C-methyl-dihydromyricetin, 4) from Cedrus deodara, ²² and, based on UV, mass spectral and ¹H NMR data, it was evident that ring-A was mono-C-methylated, but failed to establish which of the two sites (ie, C-6 or C-8) had the C-methyl substitution as both would result in identical retro-Diels–Alder fragments in the mass spectra and proton multiplicity. To acquire relevant information for the differentiation between these two isomeric structures, we sought ¹³C NMR analysis. We first unambiguously assigned the ¹³C NMR data for 1, employing substituent-induced chemical shifts, and analysis of ¹H-coupled ¹³C NMR spectral data, as well as after deuterium exchange of chelated 5-OH, which leads to loss of coupling between C-6 and 5-OH. ²³ The ¹³C NMR data for 1 then acted as a reference in analyzing the ¹³C NMR data for 3. An analogous detailed ¹³C NMR analysis based on deuterium exchange long-range ³J_CH coupling between C-6- and 5-OH identified C-6 as the site for C-methyl substitution in 3. Similar ¹³C NMR studies for 4 elucidated 6-C-methyl substitution and thus the isolated flavonoids were having a common 5,7-dihydroxy-6-C-methylated dihydroflavonol skeleton with ring B either 3´,4´-dihydroxylated, as in 3, or 3´,4´,5´-trihydroxylated, as in 4.

To gain further insight, ¹³C NMR spectral studies were performed for taxifolin-tetramethyl ether ((3-hydroxy-5,7,3´,4´-tetramethoxyflavanone, 5), which exhibited all 4 aromatic OMe resonances at around 56 ppm (specifically at 55.72, 56.02 × 2, and 56.21 ppm). However, the equivalent signals were observed at 56.0 × 3 and 61.4 ppm in the ¹³C NMR spectrum of cedeodarin-tetramethyl ether ((6-C-methyl-3-hydroxy-5,7,3´,4´-tetramethoxyflavanone, 6). We rationalize the anomalous shift at 61.4 ppm for 1 OMe to structural aspects of 6 versus 5. The only structural feature which differentiates these 2 compounds is that 5-OMe is di-ortho-substituted (quaternary C-6 and quaternary C-10) in 6, whereas 5-OMe has 1 free ortho-H (H-6) in 5. Therefore, the signal at δ_C 61.4 corresponds to 5-OMe in 6. This would not be the case in an alternative structure with C-Me at C-8 as both 5-OMe and 7-OMe would have a free H-6 in the ortho position and thus would be absorbing at around 56 ppm, similar to 5. Thus, a 5.4 ppm downfield shift of the 5-OMe signal in 6 versus 5 is associated with the lack of free ortho positions. Interestingly, our observation was in accordance with studies related to substituted anisoles^24,25 and supported by a publication almost 20 years later in 2003 incorporating the ¹³C NMR data for isomeric 2,5-dihydroxylated-7-methoxy-dihydroflavonols having Me at C-6/C-8, and the signal of 7-OMe was observed at 56.0 and 56.1 ppm. ²⁶ The 5-OMe was noticed at δ 61.6 in 5-methoxy, 6-hydroxyflavanone from Helichrysum forskahlii. ²⁷

Based on a detailed literature search, aryl-OMe groups can be grouped into 2 well-defined ¹³C NMR chemical shift ranges: 56 ± 1 and 61.5 ± 2 ppm.^{19–21,23–28} The former is considered normal for OMe having at least 1 free ortho-H, whereas the latter is typical for compounds missing free ortho and orthó-H. The chemical shift difference [Δc = δ_COMe(ortho-di-substituted)-δcOMe (having at least one free ortho position) = 4.5 ± 1.5 ppm] has been rationalized as in-plane in the case of the former versus out-of-plane orientation of the OMe group with respect to the aromatic ring. Hence, the ¹³C NMR chemical shifts of OMe carbons are strongly dependent on the atomic environment (substituents in ortho positions) and their topology plays a dominant role in determining their chemical shifts.

It has been suggested that normal OMe groups possess a coplanar aryl–O bond in which the π orbitals of the aromatic ring tend to overlap with the lone pair orbitals of the OMe oxygen, leading to a delocalization of the nonbonding oxygen electrons and a strengthening of the Ar–O bond.^28,29 In the out-of-plane conformation, which is expected for a di-ortho-substituted OMe group, the conjugation of the nonbonding oxygen electrons with bonds of the aromatic ring is interrupted, leading to a decreased electron density on the OMe carbons.^28–30 These observations were supported by the T1 relaxation times for OMe resonances. ³¹ Using theoretical calculations, Carter et al ³¹ suggested that the deshielding contributed by the C–H bonds was responsible for the out-of-plane orientation of the di-ortho-substituted OMe group. ³¹ Jardon et al ³² suggested that the OMe carbon is more shielded (∼56 ppm) in the absence of a neighboring substituent because the in-plane orientation of the OMe group enabled better conjugation of the lone pair of electrons (LP) of the oxygen atom with the aromatic ring. The OMe group is rotated out of the plane of the aromatic ring, causing weaker conjugation of the oxygen LP with the aromatic system in the presence of 2 ortho substituents, accompanied by an accumulation of negative charge on the oxygen atom, resulting in a reduced shielding of the OMe carbon (∼62 ppm). In a recent study by Toušek et al, ³³ it was suggested that the atypical ¹³C NMR chemical shifts observed are not directly related to a different conjugation of the LP of the OMe oxygen with the aromatic ring, but to changes in the virtual molecular orbital space caused by the rotation of the OMe group, which, in turn, correlate with the predominant part of the contribution of the paramagnetic deshielding connected with the magnetic interactions, resulting in the experimentally observed deshielding of the ¹³C NMR resonance of the out-of-plane OMe group. From all of these theoretical as well as experimental observations, it is well evident that the OMe chemical shift depends predominantly on either the presence or lack of ortho substitution, and thus (Δδ values) can have significant implications in the determination of the molecular structure of methoxylated phenolic compounds.

Recently, it has been shown that the widths at half height (W_1/2) of the ¹H NMR OMe signal are dependent on the presence of free ortho positions due to ⁵J_Hortho,OMe and can also be of importance in deducing the site of OMe-substitution. ³⁴ In a detailed ¹H NMR study related to the structure elucidation of 7,2′,5′-trimethoxy-3′, 4′- methylenedioxy-isoflavone, Joseph-Nathan and coworkers noticed that 3 OMe signals were observed at 3.85, 3.87, and 3.92 ppm, and the first peak corresponds to a nonlong-range coupled 2′-OMe group, while the other 2 signals are long-range coupled to aromatic hydrogen atoms corresponding to 5′-OMe and 7-OMe, respectively. ³⁵ The W_1/2 values of these signals were 0.26, 0.53, and 0.61 Hz, in increasing chemical shift order, thus the OMe signal lacking the ortho-H appeared sharper as compared to those having hydrogens in ortho locations. This is confirmed by careful irradiation of the sole aromatic-H signals and/or by homonuclear spin decoupling. ³⁵ Based on several publications related to polymethoxylated flavonoids, it can be observed that OMe signals can be observed as independent and/or overlapping signals appearing in the 3.85 ± 0.25 ppm range in ¹H NMR spectra and their chemical shifts seem not to be very clearly correlated with the oxygenation pattern on the flavonoid nuclei.^8,22,34–44

The next major advancement in flavonoid structural techniques was the application of homonuclear (¹H–¹H) and heteronuclear (¹H–¹³C) correlation 2-dimensional-NMR spectroscopy. In ¹H–¹H correlation (COSY) spectroscopy, correlations are observed among chemical shifts of ¹H NMR resonances which are coupled with each other via J-couplings. Thus, cross-peaks among ortho-coupled aromatic protons are common to all flavonoid types,^43,45–47 in addition to ¹H–¹H COSY cross-peaks among vicinal and geminal protons, which can only be observed in flavanone, dihydroflavonols, flavans, flavanols, isoflavanones, and isoflavanols due to ¹H-coupling among heterocyclic ring-C protons.^47–50 Thus, ¹H–¹H COSY spectra provide very limited structural information.

Another variant of COSY is total correlation spectroscopy or homonuclear Hartmann-Hahn correlation spectroscopy. These are quite useful for identifying spin-systems of individual monosaccharide residues in glycosides,^51–54 but exhibit limited application in structural analysis of flavonoid aglycones due to OH/OMe and other substituents on rings A, B, and C.

The proximity between the OMe group and neighboring-H can be observed in nuclear Overhauser effect spectroscopy (NOESY). In general, an ortho-aryl-H exhibits an intense NOESY cross-peak with OMe protons.^43,55–58 Irradiation of the methoxyl group protons usually causes nuclear Overhauser effect enhancement of the adjacent ortho-aryl-H, thus providing some evidence about its location on the aromatic ring. This method has been used to elucidate structures of isoflavonoids from Dalbergia parviflora. ⁴⁹

Heteronuclear multiple-quantum coherence (HMQC) and heteronuclear single-quantum coherence correlation (HSQC) spectroscopy allow 1-bond correlations between the ¹³C signal and ¹H NMR chemical shifts, thus providing chemical shift correlation data for protonated carbons. ⁵⁹ Thus, if ¹³C NMR assignments are known, then directly bonded-H resonances can be assigned by correlations observed in these spectra and/or vice versa. However, unambiguous assignments of ¹³C NMR spectra are absolutely needed to acquire further structural information. Heteronuclear multiple-bond correlation (HMBC) spectroscopic data allow the assignment of a quaternary carbon linked usually via 3 bonds to ¹H resonances. Thus, the ¹H NMR chemical shift of an OMe signal exhibits a cross-peak to an OMe-substituted aryl-carbon and sometimes also with an aryl-carbon occupying an ortho position. This feature (δ_H-OMe → Aryl-C HMBC correlation) has been found to be applicable to assign the site of OMe-substitution in several flavonoids.^{27,43,60–64}

In any case, the number of OMe groups can be ascertained by observing 1-bond correlations between methyl-C resonances resonating between δ_C 55–63 and their correlations with 3H singlets appearing in the δ_H 3.6–4.2 region in HMQC/HSQC spectra.

There are some exceptions to the above-mentioned chemical shift ranges, for example, the OMe signal was reported to appear at 65.0 ppm in (2S)-8-formyl-5-hydroxy-7-methoxy-6-methylflavanone and at 67.0 in 3′-formyl-4′,6′-dihydroxy-2′-methoxy-5′-methylchalcone isolated from Cleistocalyx operculatus. ⁶⁵ An unusually upfield position, 48.33 ppm, was reported for hesperetin-5′-O-β-rhamnoglucoside, isolated from Taraxacum mongolicum. ⁶⁶

We noticed that several authors have reported the characterization of methoxyflavonoids, including ¹³C NMR data, but did not mention the ¹³C chemical shift for the OMe signal(s), despite the potential value of their shifts (e.g., Yadava and Kumar ⁶⁷ and Srivastava and Srivastava ⁶⁸ ). It can also be mentioned that NMR data for several compounds have been wrongly interpreted with respect to OMe assignments. For example, 2′-OMe and 6′-OMe were assigned to signals at 56.1 and 63.2 ppm, respectively, in the case of praecansone B (1) ⁶⁹ and we think these values should be reversed. Mir et al ⁷⁰ reported the characterization of pinnataflavonol, a new flavonol from Pongamia pinnata, as 5,6-furano-7,8- furano-4ʹ-methoxyflavonol exhibiting a 4′-OMe signal (having unsubstituted ortho and orthó positions) at 60.24 ppm. Inconsistent chemical shifts for OMe signals were reported for flavonoids of Casimiroa edulis identified as 6,7-dimethoxyflavone [61.9 (7-OMe), 57.2 (6-OMe)] and 56,2´-trimethoxyflavone [61.9 (2´-OMe), 57.3 (5-OMe), 55.7 (6-OMe)]. ⁷¹

Based on the above discussion, it is evident that the OMe chemical shift plays an important part in the understanding of the molecular structure of methoxylated/oxygenated flavonoids. Although the use of OMe shifts in elucidating structure has a long history, it is surprising that many authors have not taken this as a criterion for structure elucidation. We consider that careful assignments of carbon resonances are of utmost importance as carbon chemical shifts are very sensitive to substitution and oxy-substituted carbons can appear in a narrow range depending on their positions on the flavonoid nucleus. Intense NOESY coupling (δ_H-OMe→ortho-aryl-H or adjacent substituent) and HMBC (δ_H-OMe→ δ_C) data, together with methoxyl carbon shifts, should be considered for the deduction of the site of substitution in flavonoids, as well as related phenolic compounds.