Unprecedented Benzoquinone Motifs Reveal Post-Oligomerizational Modification of Proanthocyanidins

Abstract

Proanthocyanidins (PACs) are complex flavan-3-ol polymers with stunning chemical complexity due to oxygenation patterns, oxidative phenolic ring linkages, and intricate stereochemistry of their heterocycles and interflavan linkages. Being promising candidates for dental restorative biomaterials, trace analysis of dentin-bioactive cinnamon PACs now yielded novel trimeric (1 and 2) and tetrameric (3) PACs with unprecedented o- and p-benzoquinone motifs (benzoquinonoid PACs). Challenges in structural characterization, especially their absolute configuration, prompted the development of a new synthetic-analytical approach involving comprehensive spectroscopy including NMR with quantum mechanics-driven ¹H iterative functionalized spin analysis (HifSA) plus experimental and computational electronic circular dichroism (ECD). Vital stereochemical information was garnered from synthesizing 4-(2,5-benzoquinone)flavan-3-ols and a truncated analogue of trimer 2 as ECD models. Discovery of the first natural benzoquinonoid PACs provides new evidence to the experimentally elusive PAC biosynthesis as their formation requires two oxidative post-oligomerizational modifications (POMs) that are distinct and occur downstream from both quinone-methide driven oligomerization and A-type linkage formation. While Nature is known to achieve structural diversity of many major compound classes by POMs, this is the first indication of PACs also following this common theme.

Graphical Abstract

Introduction

Chemically, proanthocyanidins (PACs) are oligomeric to polymeric flavan-3-ols, which constitute a ubiquitous group of shikimate/malonate-derived natural products essential for plant physiology.^1,2 Biologically, PACs play significant roles in both plant and mammalian physiology: PACs function as chemical defense against biotic and abiotic stressors in plants;³ have diverse pharmacological properties and impact health in widely used dietary products and supplements.^4,5 They are an essential dietary factor in ruminants delivered via forage crops;⁶ and act as potent biomodifiers with promising potential to advance adhesive dental restorations.⁷ Relevant to the biomedical context of our interdisciplinary work, and in contrast to having been associated with broad, almost universal bioactivity profiles, the first target-specific structure-activity relationships (SARs) for PACs have been established only recently for dentin biomodification as the biological target.^8,9 These experiments have shown that PACs stemming from different plants have different efficacy, and that trimers and tetramers possessed the preferred degree of polymerization (DP = 3 or 4). Moreover, 3-O-galloylation did not exhibit sustained potency, A-type interflavan linkages (IFL) were favorable, and specific configurational parameters were preferred.^8,9

This not only calls into question the widespread perception that PAC bioactivity is non-specific, but connects PAC biochemistry with otherwise well-established concepts of structural specificity and permutational complexity of natural polymers in general. Analogous to proteins being poly-peptides, RNA/DNA being poly-nucleotides, and lectins being polysaccharides, flavan-3-ol oligomers constitute their distinct chemical space (“PACome”) that might potentially represent a diverse biochemical toolset. To accommodate both structural analyses and biological studies, previous studies have mainly focused on the major PACs from different plant sources.^10–14

While it is likely that PACs are affected by Residual Complexity (RC), as are natural products in general, both chemically and biologically, this issue remains unaddressed to date. Summarized briefly, RC refers to the “inherent deviation of nature-derived agents from single chemical entities, caused by their metabolomic origin” (go.uic.edu/residualcomplexity). In addition to the question to what degree PACs are residually complex, this raises the key question whether even well-defined PACs contain minor analogues that might not only present novel structures, but also affect the overall bioactivity and stability. To utilize PACs for human health benefits, it is, thus, imperative to explore this chemical diversity space of PACs, develop methodology for the detailed characterization of their complex chemical structures, and validate even highly purified PACs when used as the bioactive principles.

In this context, it should be kept in mind that the structure elucidation of PACs remains challenging, especially when seeking to define their absolute configuration: each monomeric unit contributes two stereogenic centers, and each inter-flavan linkage (IFL) adds one stereogenic element, plus atropisomeric complexity in the case of B-type IFLs. For an all-B-type PAC tetramer alone, this creates the impressive chemical space of 2^(2x4)+3=2,048 stereoisomers, not counting atropisomers emanating from three rotationally-restricted IFL linkages.

ECD is the spectroscopic method of choice for characterizing the absolute configuration of PACs. It was developed using synthesized 4-arylflavan-3-ols as models.¹⁵ Established rules are consistent with the experimental ECD data of several PAC standards¹⁶ and were further validated by the Mosher’s method¹⁷ and calculated ECD data.¹⁸ However, the ECD approach becomes less effective when determining the absolute configuration of higher oligomers due to the cumulative effect of the chromophores and the limited resolution in the 200–500 nm UV absorption window. Acid-catalyzed depolymerization^19,20 coupled with chiral-phase chromatography can solve the absolute configuration of B-type PACs, but only when reference standards are available.¹² As for A-type PACs, which are resistant to cleavage, differential ¹³C NMR chemical shift (Δδ) values between diastereomeric building blocks can be complementary to ECD data in determining the absolute configuration of higher oligomers with all-A-type IFLs.²¹

Towards our broader goal of developing new natural-sourced biomaterials that chemically and mechanically modulate the properties of dental tissue for clinical interventions, we have characterized nine A- and B-type PAC oligomers up to pentamers from Cinnamomum verum (Lauraceae) bark.^11,14 To develop clinically effective dentin biomodifiers, one trimeric and one tetrameric PAC have been selected as candidates for ongoing translational studies due to their overall in vitro performance, favorable yield, and accessibility in terms of preparation from spent cinnamon bark.^11,14 In the course of scaled-up purification, “knock-out” fractions of these bioactive lead compounds became available and provided the opportunity to explore previously untapped areas of the RC of PACs.

Employing diol-phase HPLC and LC-MS analysis to target PAC-like structures in residually complex fractions from PAC oligomer “knock-outs”, now led to the identification of three new compounds: two trimers (1, 2) and a tetramer (3) that feature unprecedented o- and p-benzoquinone motifs (“benzoquinonoid PACs”). The occurrence of these oxidatively modified PAC building blocks has two major chemical implications: (a) in terms of PAC chemistry, previously established ECD methodology for absolute configuration determination lacks the essential reference data, thereby calling for a new synthetic-analytical approach to their structure elucidation. (b) With regard to PAC biosynthesis, the newly discovered benzoquinonoid PACs provide important chemical evidence for the mechanisms underlying PAC biosynthesis.

This report presents the array of chemical methodology required to discover and characterize the three novel benzoquinonoid PACs (1–3) and enable their absolute configuration assignments: (i) targeted isolation from the two major but residually complex dentin bio-active compounds; (ii) full structural characterization via a combination of comprehensive NMR data analysis, experimental and computational ECD spectroscopy; and (iii) synthesis of a pair of diastereomeric model compounds as well as a truncated “dimeric” model, to evaluate the impact of the unique benzoquinone motifs on the ECD spectra of PACs. Moreover, the finding of benzoquinoid PACs provides the biosynthetic clue that phenolic precursors including benzoquinone monomers are part of the metabolomic flux and are being integrated during PAC oligomerization, thereby leading to a novel diversity element of this class of naturally occurring PACs.

Results and Discussion

Targeted Isolation of New PAC Types from C. verum.

The trimer, cinnamtannin B1 (PACBAR code: EC=8EC-8EC), and the tetramer, parameritannin A1 [PACBAR code: EC=8EC(6-EC)-8EC], are the principal compounds of the PAC-enriched extract of C. verum bark.^11,14 To isolate these PACs at gram-scale levels for comprehensive dental biology studies, the tri- and tetra-meric PAC fractions were first enriched by centrifugal partition chromatography and further purified by repeated Sephadex LH-20 chromatography. As this gradually produced “knock-out” fractions of the two increasingly pure PACs, the remaining fractions represented the RC of the bulk isolates, which potentially impacted the overall biological activity and stability of the two principal bioactives.

Chemical analysis of these fractions revealed similar TLC behavior with the major PACs and similar red to pinkish color reactions by the vanillin-sulfuric acid spraying reagent, triggering more detailed inspections by LC-MS for the presence of new PAC analogues forming the RC of both PAC trimers and tetramers. Initial clues for the presence of new trimers were distinct molecular ions observed at m/z 879, being 14 Da more than those of the major trimer cinnamtannin B1 (m/z 865 [M+H]⁺). The MS² spectrum (Figure 2a) of the ion at m/z 879 showed a fragment ion at m/z 589, which could be interpreted as being derived from the loss of an (epi)catechin (290 Da) unit via quinone methide cleavage (Scheme S1).²³ Similarly, the ion at m/z 1167 was observed from the minor fractions of the major tetramer, which was also 14 Da higher than the major tetramer, parameritannin A1[EC=8EC(6-EC)-8EC] (m/z 1153 [M+H]⁺), and the presence of a fragment ion with m/z 877 (−290 Da) in the MS² spectrum (Figure 2b, Scheme S2). Such observations suggested that these compounds are oligomeric PAC analogues with new substructures, which, no doubt, have escaped previous phytochemical studies due to being minor compounds.

Figure 2.

The MS/MS spectra of PAC analogues detected in the residual complex fractions of the designated bioactive PAC oligomers. Ions at m/z 879.17 (a) m/z 1167.24 (b) can be explained by the loss of an (epi)catechin unit (embedded structure).

Structure Elucidation of the Benzoquinonoid PAC Trimers 1 and 2.

Compounds 1 and 2 exhibited the same molecular formula of C₄₅H₃₄O₁₉ as determined by HRESIMS data analysis (Figures S32, S33) and their ¹³C NMR carbon counts (Table 1). Their highly similar 1D NMR spectra suggested a pair of stereoisomeric PAC trimers. The ¹H NMR spectrum of 1 (Figure S1) showed the characteristic PAC resonance patterns: AMX spin systems of trisubstituted B-rings (δ_H 7.07 ~ 5.68), an AB spin system of the A-ring (δ_H 6.26 and 6.05) of the upper unit, as well as the benzylic methylene resonances (δ_H 2.76 and 2.75) belonging to H₂-4 of the terminal unit. In the region of the oxymethine resonances, an AX spin system at δ_H 3.98 (dd, J = 3.48 Hz) and δ_H 4.06 (dd, J = 3.48 Hz) indicated the presence of a double IFL (2→O→7, 4→6 or 4→8). Three low frequency ¹³C NMR resonances (δ_C 28.46, 30.76, and 38.36; Figure S2) were ascribed to C-4 of three monomeric units and supported the trimeric nature of 1. Compared with the major trimer, cinnamtannin B1,¹⁴ a doubly-linked PAC from C. verum with one less oxygen and two more hydrogen atoms, the two carbonyl signals at δ_C 188.18 and 176.01 were the most distinct features of the ¹³C NMR data 1 (Figure S2). The HMBC spectrum showed correlations between two methines (δ_H 4.06 and 4.25) and the carbonyl group resonating at δ_C 188.18. Taking into account the ¹H-¹H COSY correlations, these methines were assigned as I-H-4 and II-H-4, respectively, and the carbonyl group as C-5 (Figure 3). However, the other carbonyl group (δ_C 176.01) showed no HMBC correlations, even in an experiment optimized for 2 Hz to capture small long-range J_CH correlations. In contrast, II-H-4 of 2 showed weak correlations to both carbonyl groups (Figure S12). This was attributed to the conformation of the heterocyclic F-ring of 2 providing favorable dihedral angles between H-4 and the carbonyl carbons, leading to detectable ³J_CH and ⁴J_CH couplings.

Table 1.

The ¹H and ¹³C{¹H} NMR Spectroscopic Data of 1 and 2^a,b Include the Results of ¹H Iterative Functionalized Spin Analysis (HifSA) for the Precise Determination of the ¹H NMR Spin Parameters.

		1		2
Ring	no.	δH (J, Hz)	δ C	δH (J, Hz)	δ C
Unit I
C	2		101.71 s		101.77 s
	3	3.977 dd (3.48, −0.30)	66.27 d	3.986 d (3.65)	66.11 d
	4	4.059 dd (3.48, 0.17)	28.46 d	4.077 dd (3.65, −0.48)	28.18 d
A	5		154.20 s		154.69 s
	6	6.259 dd (2.33, 0.17)	97.64 d	6.250 dd (2.33, −0.48)	97.40 d
	7		159.00 s		158.98 s
	8	6.053 d (2.33)	98.44 d	6.038 d (2.33)	98.68 d
	9		156.86 s		156.93 s
	10		101.97 s		102.35 s
B	1'		130.33 s		130.27 s
	2'	7.074 d (2.24)	115.91 d	7.050 d (2.21)	115.90 d
	3'		147.16 sc		145.69 sc
	4'		145.72 sc		147.17 sc
	5'	6.800 d (8.34)	115.72 d	6.779 d (8.34)	115.64 d
	6'	6.960 dd (8.34, 2.24)	120.14 d	6.936 dd (8.34, 2.21)	120.18 d
Unit II
F	2	5.321 dddd (1.70, 1.01, −0.69, −0.37)	79.15 d	4.569 d (9.61)	84.75 d
	3	4.081 dddd (1.94, 1.01, −0.40, −0.28)	72.22 d	4.176 dd (9.61, 8.97)	73.61 d
	4	4.253 dd (1.94, −0.20)	38.36 d	4.285 d (8.97)	38.68 d
D	5		188.18 s		186.85 s
	6		121.97 s		122.33 s
	7		150.34 s		150.11 s
	8		176.01 s		175.53 s
	9		152.49 s		152.35 s
	10		121.29 s		123.30 s
E	1'		131.02 s		130.29 s
	2'	7.014 dd (2.07, −0.37)	115.72 d	6.982 d (2.10)	116.35 d
	3'		146.05 s c		146.77 s c
	4'		145.95 s c		146.11 s c
	5'	6.758 dd (8.18, −0.28)	115.60 d	6.802 d (8.18)	116.09 s c
	6'	6.831 ddd (8.18, 2.07, −0.69)	119.80 d	6.896 dd (8.18, 2.10)	121.35 d
Unit III
I	2	3.933 dd (1.05, −0.38)	80.62 d	3.585 ddd (1.33, −0.70, −0.51)	79.73 d
	3	3.786 dddd (4.21, 2.71, 1.05, −0.19)	67.06 d	3.773 ddd (4.25, 2.26, 1.33)	67.31 d
	4a	2.759 ddd (−15.93, 4.21, 0.13)	30.76 t	2.768 ddd (−16.60, 4.25, 0.42)	29.88 t
	4b	2.750 dd (−15.93, 2.71)		2.725 dd (−16.60, 2.26)
G	5		156.58 s		155.86 s
	6	6.098 ddd (−0.40, 0.20, 0.13)	96.38 d	6.067 d (0.42)	96.76 d
	7		155.36 s		155.62 s
	8		107.22 s		108.39 s
	9		155.62 s		154.00 s
	10		100.00 s		100.30 s
H	1'		131.04 s		131.50 s
	2'	6.538 dd (2.09, −0.38)	114.75 d	6.631 dd (2.09, −0.51)	115.26 d
	3'		145.74 d c		145.51 s c
	4'		145.74 d c		145.85 s c
	5'	6.443 d (7.99)	116.32 d	6.557 d (8.09)	116.04 d c
	6'	5.680 ddd (7.99, 2.09, −0.19)	119.78 d	6.192 ddd (8.09, 2.09, −0.70)	119.68 d

^a1H and ¹³C NMR data were acquired in methanol-d₄ at 600 and 150 MHz at 298 K, respectively.

^bThe δ_H and J values were determined by HifSA.

^cInterchangeable assignment.

Figure 3.

Selected 2D NMR correlations of the p- and o-benzoquinone PACs 1 and 3. Key HMBC correlations to the carbonyl groups confirming the location and quinone motif of the A-ring of the middle unit are indicated in red.

The ¹³C NMR chemical shifts of C-6, C-8, and C-10 in both catechin and epicatechin units typically fall into the 96~110 ppm range. In the spectrum of 1, seven ¹³C resonances observed within this range showed well-resolved HMBC correlations (Figure S4) and were assigned to the six C-6/C-8/C-10 carbons of units I and III as well as the acetal carbon (C-2) of unit I, while the C-6/C-8/C-10 resonances of unit II were absent. Meanwhile, two quaternary olefinic carbons resonated at higher frequencies of δ_C 121.97 and 121.29. HMBC correlations between I-H-4/I-H-3 and the carbon resonating at δ_C 121.97, as well as II-H-3/II-H-4 and the carbon resonating at δ_C 121.29 supported their assignment as C-6 and C-10 of unit II. Their unusually high δ_C values, compared with regular PACs such as cinnamtannin B1,¹⁴ reflected the influence of a p-benzoquinone substructure of the A-ring in the middle unit of 1. This was in accordance with the reported NMR data of p-benzoquinone derivatives synthesized from (epi)catechin.²⁴ Thus, the carbonyl group at δ_C 176.01 was assigned to II-C-8. HMBC correlations from II-H-4 and III-H₂-4 to carbons resonating at δ_C 107.22 (III-C-8) and 155.62 (III-C-9), as well as a ROESY cross peak indicating spatial proximity between II-H-2 and III-H-2′ defined the presence of a 4→8 single IFL between units II and III. Thereby, the indistinguishable 2D structures 1 and 2 were established as PAC trimers with a p-benzoquinone A-ring in the middle unit as shown in Figure 3 and recognized as being stereoisomeric, likely in the middle units.

This stereochemical difference was reflected in the ¹H NMR coupling constants of the C-ring hydrogens in unit II. In PACs, the relative configurations of C-2/C-3/C-4 can be assigned via the ³J_2,3 and ³J_3,4 values.²⁵ Briefly, in the 2,3-cis configuration, H-2/H-3 resonate as broad singlets and exhibit a small J_2,3 value of 1.0~2.4 Hz; whereas for the 2,3-trans configuration, the J_2,3 value is in the range of 8.0~10.0 Hz. The C-4 configuration may be deduced from the chemical shift of C-2, which resonates at higher frequencies (84~86 ppm) in 2,4-cis configured C-rings vs. at lower frequencies (77~80 ppm) in 2,4-trans orientated moieties due to a pronounced C-4 γ-gauche effect.¹² Therefore, two steps were required: (i) precise determination of the J_2,3 and J_3,4 coupling constants by ¹H iterative functionalized Spin Analysis (HifSA, see ref. ²⁶ and references therein; Table 1); and (ii) unambiguous assignment of the II-C-2 chemical shift at δ_C 79.15 in 1 vs. δ_C 84.75 in 2. Thereby, the relative configuration in the middle unit of 1 was established as 2,3-cis-3,4-trans and that of 2 as 2,3-trans-3,4-trans based on the respective J_2,3 = 1.70 and 9.61, J_3,4 = 1.94 and 8.97 values, while the relative configurations of units I and III were identical in both compounds.

The structures of 1 and 2, were thus, established as the two p-benzoquinonoids, epicatechin-(2β→O→7, 4β→6)-epicatechin-5,8-dione-(4β→8)-epicatechin and epicatechin-(2β→O→7, 4β→6)-ent-catechin-5,8-dione-(4β→8)-epicatechin, respectively (Figure 1). The new flavan-3-ol-5,8-dione substructures were assigned the “pQ” prefix for inclusion in the PACBAR coding system.²² Accordingly, the micro-PACBAR code of 1 is EC=6pQEC-8EC, while 2 is EC=6pQeC-8EC.

Figure 1.

The chemical structures as well as the graphical macro- and micro-PACBAR codes of the new benzoquinone PACs, 1–3. The proposed extension of the PACBAR terminology²² denotes p- (5,8-) and o- (5,6) quinone motifs as pQ and oQ, respectively.

Structure Elucidation of the Benzoquinonoid PAC Tetramer 3.

In its (+)-HRESIMS spectrum (Figure S35), 3 exhibited a molecular ion at m/z 1167.2425 [M+H]⁺. Considering the MS/MS fragment ion at m/z 877.1601 [M+H]⁺ (Figure 2b) and the ¹³C NMR spectrum (Figure S16) showing two carbonyl groups at δ_C 189.64 and 178.55 as well as four putative C-ring C-4 carbon resonances between 40~28 ppm, 3 was recognized as a tetrameric PAC analogue, also containing a benzoquinone substructure. The B-type linkage between units III and IV was established by HMBC correlations from the methine signals at δ_H 4.56 (III-H-4) and 4.21 (III-H-3) to the quaternary carbon at δ_C 105.93, which was determined as C-8 of the terminal unit based on its chemical shift, which is often shielded compared to C-6, and further confirmed by ROESY correlations between III-H-2 and IV-H-2′/6′.

After assigning the AB spin system (δ_H 5.97 and 6.01) to the A-ring of unit I, the remaining NMR resonances of unit I were resolved by their HMBC correlations (Figure 3). The acetal carbon at δ_C 99.85 (I-C-2) suggested an A-type linkage between units I and II. The ROESY correlation between I-H-4 and II-H-2′ indicated a (2→O→7, 4→8) IFL from unit I to II. The carbonyl group at δ_C 189.64 showed a single HMBC correlation with II-H-4, hence, it was assigned to II-C-5. Subsequently, the oxygenated aromatic carbon at δ_C 149.56 could be assigned as II-C-9 based on its correlations with I-H-4/II-H-4. The ¹³C NMR chemical shifts of the conjugated unsaturated systems of the II-A-ring in 1, i.e., Δ^6,7 (δ_C 121.97, 150.34) and Δ^9,10 (δ_C 152.49, 121.29) indicated that the double bonds formed a p-benzoquinone A-ring in unit II. The ¹³C NMR resonances of the Δ^7,8 (δ_C 150.49, 108.39), and Δ^9,10 (δ_C 149.56, 120.64) double bonds in 3, defined its II-A-ring as an o-benzoquinone moiety.

The absence of aromatic hydrogens (H-8/6) in the III-A-ring suggested the occurrence of H/D exchange, a reaction that often occurs with phloroglucinol-type A-rings in deuterated protic solvents such as methanol-d₄ and D₂O.²⁶ This was further confirmed by HRMS data analysis of the sample recovered from the NMR tube, which exhibited a molecular ion of m/z 1168.2495 [M+²H]⁺ (Figure S36), indicative of hydrogen-deuterium exchange. Such H/D exchange reactivity made the NMR assignment of the already proton deficient compound more challenging. In the ¹³C NMR spectrum of 3, one oxygenated aromatic carbon was absent from the typical region of C-5/7/9 (149~159 ppm), while unusual HMBC correlations were observed from I-H-4 and III-H-4 to the carbon at δ_C 150.49, suggesting II-C-7 and III-C-9 were overlapping in the ¹³C NMR spectrum. After differentiating C-5 and C-9 in unit III, the linkage site was determined to be C-6, as corroborated by HMBC correlations between III-C-5 and both II-H-3/4 and III-H-4.

The 2,3-cis relative configurations of the monomeric units were determined based on the broad singlet resonances of H-2/3 and the small J_2,3 values as determined by HifSA (Table 2). The 3,4-trans configurations were established via the aforementioned γ-gauche effect. Therefore, the structure and relative configuration of 3 was established as the o-benzoquinonoid, epicatechin-(2β→O→7, 4β→8)-epicatechin-5,6-dione-(4β→6)-epicatechin-(4β→8)-epicatechin. Coding the o-benzoquinone substructure as “oQ”, 3 may be represented as EC=8oQEC-6EC-8EC in micro-PACBAR notation.

Table 2.

The ¹H and ¹³C{¹H} NMR Spectroscopic Data of 3^a,b Including the Results of HifSA for the Precise Determination of the ¹H NMR Spin Parameters.

Ring	no.	δH (J, Hz)	δ C			δH (J, Hz)	δ C
Unit I				Unit III
C	2		99.85 s	I	2	5.726 ddd (1.10, 0.89, −0.38)	78.93 d
	3	3.377 d (3.70)	65.43 d		3	4.213 dd (2.14, 0.89)	72.58 d
	4	4.205 dd (3.70, 0.13)	29.04 d		4	4.557 ddd (2.14, 1.10, 0.58)	38.51 d
A	5		156.68 s	G	5		151.90 s
	6	5.974 d (2.39)	98.35 d		6		109.87 s
	7		157.85 s		7		152.62 s
	8	6.014 d (2.39)	96.52 d		8		96.44–96.55
	9		154.12 s		9		150.49 s*
	10		105.34 s		10		107.98 s
B	1'		130.85 s	H	1'		131.53 s
	2'	6.935 dd (2.30, 0.28)	115.52 d		2'	7.331 dd (2.18, 0.28)	116.56 d
	3'		145.27 s c		3'		145.99 s c
	4'		146.25 s c		4'		146.34 s c
	5'	6.710 dd (8.30, 0.28)	116.84 d		5'	6.860 dd (8.22, 0.28)	116.14 d
	6'	6.660 dd (8.30, 2.30)	119.94 d		6'	7.213 ddd (8.22, 2.18, −0.38)	121.15 d
Unit II				Unit IV
F	2	5.102 ddddd (1.62, 0.96, −0.37, −0.25, 0.20)	78.92 d	L	2	4.703 ddd (1.13, −0.22, −0.12)	79.95 d
	3	3.530 ddd (1.74, 0.96, −0.13)	72.42 d		3	3.918 dddd (4.90, 1.13, 0.86, −0.18)	67.67 d
	4	4.240 dd (1.74, 1.62)	37.07 d		4a	2.940 ddd (−17.32, 4.90, 0.67)	29.88 t
					4b	2.862 ddd (−17.32, 0.86, 0.67)
D	5		189.64 s	J	5		157.34 s
	6		178.55 s		6	6.168 dd (0.67, 0.58)	96.44 d
	7		150.49 s*		7		155.99 s
	8		108.39 s		8		105.93 s
	9		149.56 s		9		156.36 s
	10		120.64 s		10		101.11 s
E	1'		130.90 s	K	1'		133.19 s
	2'	6.794 ddd (2.13, −0.25, 0.22)	115.84 d		2'	6.791 dd (2.02, −0.22)	115.74 d
	3'		145.79 s		3'		145.26 s c
	4'		146.08 s		4'		145.46 s c
	5'	6.633 dddd (8.20, 0.22, 0.20, −0.13)	115.59 d		5'	6.782 dd (8.07, −0.18)	115.71 d
	6'	6.358 ddd (8.20, 2.13, −0.37)	120.13 d		6'	6.722 ddd (8.07, 2.02, −0.12)	119.69 d

^a1H and ¹³C NMR data were acquired in methanol-d₄ at 600 and 150 MHz at 298 K, respectively.

^bThe δ_H and J values were generated via HiFSA.

^cInterchangeable assignment.

^*Overlapped signals.

Synthesis and Spectroscopic Properties of 4-(2,5-Benzoquinone)flavan-3-ols as Stereochemical References for PACs with Benzoquinone Motifs.

The general chiroptical rules for the determination of the absolute configuration of PACs were developed from analysis of the Cotton Effects (CEs) in the ECD spectra of 4-arylflavan-3-ols. These compounds were synthesized from flavan-3,4-diols of known absolute configuration by stereoselective substitution at C-4, utilizing the strong nucleophiles, phloroglucinol or resorcinol.^15,27 The ECD spectra of 4-arylflavan-3-ols exhibit multiple CEs, and the C-4 chromophore is the major contributor to the sign of the high-intensity low-wavelength CE, as well as the spatial orientation and conformation of the O-heterocycle.^15,25

To assess the impact of a p-benzoquinone motif on the CEs in the ECD spectra of benzoquinonoid PACs, and to attempt the definition of the absolute configurations of 1-3, the C-4 benzoquinone substituted flavan-3-ols were synthesized as models for stereochemical referencing. Inspired by previous studies of the synthesis of 4-arylflavan-3-ols,^15,27 the 4-(benzoquinone)flavan-3-ols were generated via oxidation of 4-arylflavan-3-ols, where the C-4 substituted phenolic group possesses a 2,5-dihydroxy substitution pattern. However, the 3,4-dihydroxy substituted B-ring of both catechin and epicatechin is susceptible to oxidation into o-semiquinone and o-benzoquinone products.²⁸ Therefore, it was necessary to protect the phenolic groups of the incipient flavan-3-ols prior to the oxidation step.

The chiroptical properties of free phenolic 4-arylflavan-3-ols are consistent with their corresponding O-methyl ethers, as validated by experimental and calculated ECD data.¹⁸ The tetra-O-methylation of (+)-taxifolin (4, 2R,3R), available from our pool of monomeric flavonoid reference compounds, using dimethyl sulfate (DMS)²⁹ was performed at ambient temperature to prevent C-2 epimerization,¹³ affording (2R,3R)-5,7,3′,4′-tetra-O-methyltaxifolin (5) in 70% yield (Scheme 1). The NaBH₄ reduction products of 5 comprised a mixture of unstable flavan-3,4-diol epimers, thus, the nucleophilic benzene-1,2,4-triol was added prior to the reduction step. However, in contrast to the mild conditions of coupling the free phenolic entities, the reaction of the 3,4-diol equivalent of 5 and benzene-1,2,4-triol only proceeded when the temperature and acid concentration were increased, presumably due to the lower nucleophilicity of benzene-1,2,4-triol compared to phloroglucinol, and the reduced electrophilicity and higher steric hindrance of C-4 in the 3,4-diol of 5 compared to the free phenolic form.²⁷ Eventually, stereoselective condensation in the presence of 0.3 M HCl yielded 6 and 7 at 60 °C in a ca. 2:1 ratio in 62% overall yield. Being susceptible to slow oxidation into the p-benzoquinone analogues, compounds 6 and 7 were subjected to mild oxidation using Ag₂CO₃ at ambient temperature,³⁰ ultimately affording the 4-(4-hydroxy-2,5-benzoquinone)flavan-3-ols, 8 and 9.

Scheme 1.

Synthesis of the 4-(4-Hydroxy-2,5-benzoquinone)flavan-3-ols, 8 and 9.

Compounds 8 and 9 both possessed the molecular formula of C₂₅H₂₄O₉ on the basis of their HRESIMS spectra (Figures S39, S40). The ¹H NMR spectra (Table 3) displayed the expected AB spin system of the taxifolin A-ring (δ_H 6.12~6.20, J = 2.23 and 2.40 Hz), an AMX spin system of the B-ring (δ_H 6.82~7.05), as well as four methoxy groups. The heterocyclic C-ring protons appeared as two doublets (δ_H 4.95 and 4.80, for H-2 and δ_H 4.22 and 4.85 for H-4 in 8 and 9, respectively), with mutual coupling to H-3 giving rise to doublet of doublets (δ_H 4.31 and 4.28, respectively). The two doublets ascribed to H-2 and H-4 were differentiated via HSQC spectra. The chemical shifts of C-2 (8: δ_C 81.50; 9: δ_C 77.52) as well as the coupling constants (8: J_2,3 6.60, J_3,4 5.15 Hz; 9: J_2,3 9.60, J_3,4 6.69 Hz) further revealed the 2,3-trans-3,4-trans and 2,3-trans-3,4-cis configurations of 8 and 9, respectively. Two olefinic singlets (δ_H 5.84~6.21) and two carbonyl groups (δ_C 184.27~188.00) observed in the ¹H and ¹³C NMR spectra of 8 and 9 indicated the oxidation of the C-4 phenolic moiety into the 2,5-benzoquinone derivative. The deshielded carbonyl carbons (8: δ_C 187.80; 9: δ_C 187.99) showed HMBC correlations with H-4/H-6″, while H-3″ showed HMBC cross peaks with C-1″/C-4″ and carbonyl carbon resonances (8: δ_C 184.77; 9: δ_C 187.27; Figure S30), thereby supporting the p-disposition of the carbonyl groups. Collectively, this confirmed the presence of the intended 4-hydroxy-2,5-benzoquinone moiety as the C-4 substituent in the synthetic model compounds.

Table 3.

The ¹H (600 MHz) and ¹³C{¹H} (150 MHz) NMR Data of 8 and 9 in Acetone-d₆.

No.	8		9
	δH (J, Hz)	δ C	δH (J, Hz)	δ C
2	4.961 d (6.60)	81.50 d	4.801 d (9.60)	77.52 d
3	4.322 dd (6.60, 5.15)	73.03 d	4.286 dd (9.60, 6.69)	69.79 d
4	4.224 d (5.15)	40.32 d	4.848 d (6.78)	35.47 d
5		159.35 s		158.92 s
6	6.148 d (2.40)	93.02 d	6.156 d (2.23)	92.84 d
7		161.59 s		161.73 s
8	6.202 d (2.40)	94.54 d	6.116 d (2.23)	93.91 d
9		157.01 s		157.29 s b
10		104.31 s		103.61 s
1′		132.59 s		132.06 s
2′	6.945 d (2.23)	111.56 d	7.052 d (2.32)	112.77 d
3′ a		150.11 s		150.13 s
4′ a		150.35 s		150.37 s
5′	6.820 d (8.32)	112.72 d	6.905 d (8.23)	112.35 d
6′	6.879 dd (8.32, 2.23)	120.36 d	6.965 dd (8.23, 2.32)	121.56 d
1″		153.68 s		154.52 s
2″		187.80 s		187.99 s
3″	5.88 s	108.57 d	6.031 s	109.19 d
4″		158.15 s		157.29 s b
5″		184.77 s		184.27 s
6″	5.836 brs	128.37 d	6.211 s	131.28 d
5-OMe	3.624 s	55.89 q	3.680 s	56.06 q
7-OMe	3.788 s	55.63 q	3.770 s	55.64 q
3′-OMe a	3.741 s	56.11 q	3.798 s	56.12 q
4′-OMe a	3.730 s	56.07 q	3.781 s	56.10 q

^aInterchangeable assignments.

^bOverlapped signals.

Absolute Configuration of 8 and 9 by ECD Spectroscopy.

The absolute configurations of compounds 8 and 9 were defined by means of both their experimental and computed ECD data analysis. Overlay of their experimental and computed ECD spectra showed good matches for both compounds (Figure 4) and proved the consistency of the stereochemical assignments.

Figure 4.

Experimental (black) and calculated ECD spectra (red) in MeOH. A shows the Boltzmann averaged spectra of conformers e and f of compound 8, whereas B shows those of conformers d and e of compound 9. The σ-value (artificial line broadening) of the calculated spectra was set to 0.22 eV and 0.29 eV, respectively.

It should be emphasized that the two π→π* electronic transitions in the near ultraviolet region of the UV spectra of p-benzoquinones are analogous to those observed in benzene and aromatic hydrocarbons in general.³¹ Thus, the ECD spectra (Figure 4) of both 8 and 9 showed distinct Cotton effects (CEs) near 300, 240, and 210 nm, reminiscent of those of 4-arylflavan-3-ols.^15,18 However, a conspicuous difference involved the shifts of the CEs of the aryl and para-benzoquinone chromophores at C-4 in 8 and 9 to ca 210 nm, compared to those of the classical 4-arylflavan-3-ols at 220–240 nm.¹⁵ Such a shift in the CEs of the bifunctional C-4 chromophores in 8 and 9 not only revealed the ¹L_a CEs of the aromatic B-ring chromophore, but also permitted the full assignment of the absolute configurations of 8 and 9. Moreover, the ECD spectra of both 8 and 9 displayed low-amplitude positive and negative CEs near 350 and 450 nm, respectively. These CEs arose from the n→π* electronic transitions of the lone-pair electrons of the carbonyl groups of the p-benzoquinone units (vide infra).

The respective negative and positive ¹L_b bands at ca. 300 nm of the aromatic B-ring chromophores in the spectra of 8 and 9 indicated (2R) and (2S) absolute configurations, respectively. Both compounds displayed negative ¹L_a bands at ca. 240 nm of the aromatic B-ring chromophores. Such an observation corroborated the relative configuration assignments, i.e., the ¹L_a band is similar in sign to the ¹L_b band for 2,3-trans-3,4-trans configured analogues, but opposite in sign to the ¹L_b band for 2,3-trans-3,4-cis analogues.²⁵ The prominent positive CE at ca. 260 nm in the spectrum of 8 resulted from a π → π* type electronic transition from the B-ring to the p-benzoquinone moiety (vide infra).

The respective J_2,3 and J_3,4 values of 6.6 and 5.5 Hz for 8 indicated a 2,3-trans-3,4-trans relative configuration and a C-ring in a twisted boat conformation.³² The high-amplitude negative CE at ca. 210 nm in its ECD spectrum was evidence for an α-oriented p-benzoquinone moiety at C-4 and hence a (4S) absolute configuration. When taken in conjunction with the assigned relative configuration based on the negative ¹L_b and ¹L_a bands of the aromatic B-ring chromophore, the absolute configuration of compound 8 could be unambiguously assigned as (2R, 3S, 4S).

In the 210 nm region of the spectrum of 9, the sign of the CE deviated significantly from those observed in our previous studies of 2,3-trans-3,4-cis-4-arylflavan-3-ols.^15,18,27,32 Based upon the aromatic quadrant rule (see ref.³² for interpretation of the rule as applied to 4-arylflavan- 3-ols), a high-amplitude positive CE near 210 nm, as was observed for compound 9, would be indicative of a C-4 substituent located in the upper left quadrant, thus, requiring the p-benzoquinone function to be β-oriented.³² However, the β-orientation of the C-2 aryl group as established by the signs of the ¹L_b and ¹L_a bands of the aromatic B-ring chromophore, in conjunction with the NMR data (vide supra), necessitated an α-orientation of the C-4 substituent, for which a negative CE (substituent in the lower left quadrant) would be anticipated. The resolution of this apparent contradiction involved consideration of the C-ring conformation as follows: the presence of the C-2 carbonyl group of the p-benzoquinone unit in 9 allows formation of a hydrogen bond with the C-3 hydroxy group, leading to a slightly distorted C-ring conformation (conformer 9d, Figure S44) and, thereby, exhibiting M-helicity (see ref. ²⁰ for definitions of M- vs. P-helicity of benzopyran moieties). As a net result, the benzoquinone moiety is located in the lower right rather than the lower left quadrant and, hence, a positive high-amplitude CE near 210 nm is observed. Collectively, these observations strongly supported the absolute configuration of 9 as (2S, 3R, 4S). Analysis of the molecular orbitals involved in the electronic excitations that led to the observed CEs in the ECD spectra of 8 and 9 are discussed in detail in the SI.

Absolute Configuration of the new PACs 1–3 with o- and p-Benzoquinone Motifs.

The comprehensive interpretation of the ECD spectra of the model compounds, 8 and 9, enabled the definition of the absolute configurations of the new benzoquinonoid PACs, 1-3. Their ECD spectra (Figure 5) displayed low-amplitude negative CEs between 325 and 400 nm, which are due to the n→π* electronic transitions of the lone-pair electrons of the carbonyl oxygen atoms. All three spectra also exhibited positive CEs at ca. 300 nm based upon the ¹L_b electronic transitions of the aromatic B-ring chromophores. Although it is risky to assign C-2 absolute configurations in oligomeric proanthocyanidins based upon the sign of these CEs, the positive CEs observed for 1–3 most likely indicated a majority of 2α-oriented B-rings in the flavanyl constituent units,³³ hence (2S) absolute configurations was concluded for the A-type units in 1–3, as well as (2R) absolute configurations deduced for the remaining flavanyl units, except for a (2S) configuration of the middle unit of compound 2, which is based upon the 2,3-trans-3,4-trans relative configuration assigned via the ³J ¹H NMR coupling constants of the hydrogen atoms of the heterocyclic ring.

Figure 5.

The experimental ECD spectra of 1–3.

The crucial 200–240 nm region of the ECD spectrum of benzoquinonoid tetramer, 3, displayed the characteristic positive couplet of oligomeric PACs possessing β-linked flavanyl constituent units,^16,34 reminiscent of (4R) absolute configuration of the A-type top unit and (4S) absolute configuration of the subsequent two flavanyl moieties. In the same region of the spectra of trimers 1 and 2, only the short-wavelength component of the positive couplet was observed, thus permitting the definition of the (4R) and (4S) absolute configurations of the top and middle units, respectively, of both compounds. The latter assignments were confirmed when the ^IH NMR coupling constants of the hydrogens of the heterocyclic rings of unit II (³J_2,3 = 1.70 Hz; ³J_3,4 = 1.94 Hz for 1: ³J_2,3 = 9.61 Hz; ³J_3,4 =8.97 Hz for 2) and, thus, the relative configurations were considered in conjunction with the assigned (2R) absolute configurations of the A-type units as well as the respective (2R) and (2S) configurations of 1 and 2. Such a predominance of epicatechin constituent units and the corresponding 4β IFLs is supported by our previous studies on PAC oligomers from the same source plant, C. verum.^11,14

Because of the novelty of PACs 1-3 and the uncertainty of the effects of the o- and/or p-benzoquinone moieties on the CEs in the ECD spectra of 1-3 compared to the typical CEs of PACs devoid of these structural units, the ECD spectrum of the truncated trimer, 2, i.e., dimer 2' was calculated. An overlay of the spectra of trimer 2 vs. 2' is shown in Figure 6 and demonstrates an excellent match except for the slight shifts in the wavelengths of the CEs. Such shifts presumably result from the removal of the GHI-units in 2 and, hence, loss of the significant C-4 (F-ring) and C-2 (I-ring) stereocenters and their associated CEs on the overall ECD spectrum. Molecular orbital analysis involved in the electronic excitations that led to the observed CEs in the ECD spectrum of 2' are discussed in the SI.

Figure 6.

Experimental (black) and calculated ECD spectra (red) in MeOH of the Boltzmann averaged spectra of conformers c, f, i, l, and p of truncated 2'. The σ-value (artificial line broadening) in the calculation was set to 0.23 eV. UV wavelength correction of 1 nm was applied to generate this graph.

Collectively, the additional ECD data permitted the definition of the (2S, 3R, 4R: C-ring, 2S,3R: F-ring) absolute configuration of truncated 2'. Thus, the structures of 1-3 were unequivocally assigned as epicatechin-(2β→O→7, 4β→6)-epicatechin-5,8-dione-(4β→8)-epicatechin, epicatechin-(2β→O→7, 4β→6)- ent-catechin-5,8-dione-(4β→8)-epicatechin, and epicatechin-(2β→O→7,4β→8)-epicatechin-5,6-dione-(4β→6)-epicatechin-(4β→8)-epicatechin, respectively (Figure 1).

Biosynthetic Implications.

The pathways depicted in Scheme 2 reflect the contemporary state-of-the-art knowledge in PAC biosynthesis and establish the new quinonoid PACs into the context of the PACome of the source plant. While the enzymatic parameters controlling the coupling reaction between the nucleophilic flavan-3-ol starter and the electrophilic flavan-3,4-diol extender unit remains elusive, the current finding not only add new quinone methide reaction chemistry and fully elucidated benzoquinonoid structures to the current consensus, but are also evidence for post-oligomerization modification as an important mechanism in PAC biosynthesis.

Scheme 2.

Plausible biosynthetic pathway to the formation of the quinonoid-type PAC trimer, 1, involves PAC oligomerization (OLI), mid-stream A-type double linkage (ALF) formation, and two post-oligomerizational modification steps (POM_1/2) generating the p-quinone D-ring. The starter unit is drawn in blue, the extension units in red and brown.

Recognizing the post-oligomerization nature of benzoquinonoids requires consideration of the reaction schemes and intermediates starting with the monomers: the most abundant monomeric component of C. verum, epicatechin (10, EC), is generated from leucocyanidin via a combination of an anthocyanidin synthase (ANS) to afford cyanidin followed by an anthocyanidin reductase (ANR).³⁵ Cyanidin may be trapped by nucleophiles such as water, L-cysteine, and ascorbate,³⁶ forming the 2,3-cis-leucocyanidin to establish a precursor pool comprising EC (10) as starter and 2,3-cis-leucocyanidin (11) as extender units (Scheme S3).

Coupling of EC (10) and (11) via a C-4 resonance-stabilized carbocation (12) then affords procyanidin B2 (13, EC-8EC),³⁷ which may be susceptible to oxidation at C-8 (unit I) [the Post-Oligomerization Modification (POM₁) process] to give the 8-hydroxyprocyanidin B2 (14, 8-OHEC-8EC). While it is tempting to invoke oxidation of the substituted hydroquinone A-ring unit of 14, C-6 of the resulting 5,8-benzoquinone moiety would no longer serve as an essential nucleophilic center to sustain the oligomerization process. Thus, 14 would rather be susceptible to a further oligomerization (OLI) step with 12 to afford the trimeric intermediate 15. A subsequent oxidation step would create the B-ring quinone-methide (16) (unit I), which would afford the C-O-C linkage in the A-type procyanidin (17)³⁸ via intramolecular trapping by HO-7 (unit II). The fully substituted hydroquinone-type A-ring of its middle unit would be readily oxidized (POM₂) to form the naturally occurring trimer (1). Its C-2 epimer (unit II) is plausibly derived from a C-ring seco-intermediate (18) and subsequent ring closure to afford the thermodynamically more stable ent-catechin moiety of the naturally occurring trimer (2) (Scheme S3).³⁹ The epicatechin-5,6-dione unit II of the naturally occurring tetramer (3) would arise via a similar sequence from a C-8 substituted 6-hydroxyepicatchin (6-OHEC) unit II in tetrameric intermediate (23) (Scheme S4).

The discovery of the new PACs 1-3 containing benzoquinone structural motifs is both an attestation to and an expansion of the current paradigm of PAC biosynthesis, supported by three new pieces of chemical evidence (Scheme 2):

1. After the second oligomerization step producing the BB-type trimer 15, oxidation produces the quinone methide intermediate 16. This reactive intermediate can be stabilized by A-type linkage formation (ALF, Scheme 2) via intramolecular trapping by HO-7 to yield the AB-type trimer 17. This has two implications: it not only supports the current understanding of A-type IFL formation, but also rationalizes the following.

2. The presence of a strongly nucleophilic center at the A-ring of the starter flavan-3-ol unit or a subsequent oligomeric intermediate is a prerequisite for sustained PAC biosynthetic flux. Since oxidation of the substituted hydroquinone A-ring moiety in, e.g., dimeric intermediate 14 would turn C-6 into an electrophilic center, A-ring oxidized PACs such as the new quinonoids 1-3 must be products of POM steps. Thus, oxidation of the BB-type trimer 15 to the AB-type trimer 17 precedes a second oxidative step, that is required for the formation of the o- and p-quinone structural motifs in the A-ring of the second flavan-3-ol unit in 1-3 (D-ring; shown for trimer 1 in Scheme 2).

3. The reactivity of the intermediates discussed in (i) and (ii) is evidence for the presence of two distinct oxidative POM steps (Scheme 2) in the formation of 1-3: first, POM₁ introduces the 8-OH group that enables A-type linkage formation (ALF); subsequently, POM₂ generates the D-ring quinonoids. The family of Polyphenol Oxidases (PPOs) may be responsible for those oxidation steps, such as catechol oxidase and tyrosinase, which can oxidize o-diphenols to their corresponding o-quinones, and hydroxylate monophenols to o-diphenols, respectively.^40,41

Collectively, (i)-(iii) indicate that the biosynthesis of quinonoids 1-3 are driven by highly selective, most likely enzymatic, oxidation steps.

Notably, post-oligomerization modification of PACs is a previously unrecognized biosynthetic route. It clearly increases chemical diversity and the structural complexity of PACs and may significantly modify bioactivities as has been recognized for galloylation and glycosylation.^42,43 Thus, the SAR impact of PACs containing these benzoquinonoid moieties, including as minor bioactive species in dentin biomodification, is worth further investigation.

Conclusions

The present study comprised the discovery and full characterization of three novel oligomeric quinonoid PACs, 1–3, that constitute the Residual Complexity of the principal dentin bioactive tannins of C. verum and possess unique o- and p-benzoquinone substructures. The impact of the benzoquinone moieties on the Cotton Effects (CEs) in their ECD spectra was defined through synthesis of the 4-(2,5-benzoquinone)flavan-3-ol derivatives as model compounds and detailed analysis of their experimental and computed ECD data, as well as simulation of the ECD data of dimer 2' as the truncated model of trimer 2. Subsequently, this enabled the unequivocal establishment of the absolute configurations of the new PACs based on robust NMR and ECD data. This integrated approach can now be applied in future structural studies aimed at novel oligomeric PACs, which are in the vast majority of cases either resistant to crystallization for X-ray diffraction analysis or unfit for computational studies due to structural complexity.

The new PAC chemistry established in this study advanced our current understanding of PAC biosynthesis. In particular, it highlights the occurrence of post-oligomerizational modifications (POMs; Scheme 2) as oxidative steps that are distinct from oligomerization and IFL formation. POM is a well-established concept in the biosynthesis of polyketides (PK) and nonribosomal peptides (NRP), where it is known as post-translational modification of oligomers formed by multimodular synthetases (PKS and NRPS) that elongate activated monomers of amino and hydroxy acids, respectively.⁴⁴ Oxidative POMs also commonly occur in all major classes of terpenoids (mono-/sesqui-/di-/tri-terpenoids).^45,46

While POMs are a recurring theme in Nature, the present discovery of quinonoid-type PACs is the first evidence for an oxidative POM in PAC biosynthesis. Moreover, the quinone-containing PACs expand the chemical space of condensed tannins and go beyond galloylation and glycosylation, which are common biosynthetic modifications that do not alter the core molecular structures. As the quinone-forming POM oxidation step (POM₂) likely requires high selectivity, it would be challenging to establish synthetic routes to these novel quinone structures.

Being designated bioactive compounds and/or impurities in bioactive unmodified PAC oligomers, the new benzoquinone PACs are part of ongoing SAR studies.

EXPERIMENTAL SECTION

General experimental procedures.

Column chromatography (CC) was performed on DIAION^™ HP20 (Supelco, Bellefonte, PA, USA), Sephadex LH-20 gel (Pharmacia, Uppsala, Sweden), TOYOPEARL^® HW-40F (TOSOH Bioscience LLC, PA, USA) and Biotage^® SNAP KP-C18 flash cartridges (30 g) (Biotage, Uppsala, Sweden). TLC was using SIL G/UV254 (Macherey-Nagel, Inc., Bethlehem, PA, USA), visualized under UV light (254 nm) and spraying with vanillin-sulfuric acid reagent followed by heating. All reactions were performed in oven-dried glassware fitted with rubber septa under a positive pressure of nitrogen, unless otherwise noted. All reaction mixtures were stirred throughout the course of each procedure using Teflon-coated magnetic stir bars. Air- and moisture-sensitive liquids were transferred via syringe. The NMR data was processed with MestReNova (ver. 14.2.0–26256, Mestrelab Research, S.L.). Structural assignments were made with additional information from HSQC, HMBC, ¹H-¹H COSY and ROESY experiments.

Material.

The dried powder (8.17 Kg) of the stem bark of Cinnamomum verum was purchased from Oregon’s Wild Harvest (Sandy, OR, USA) in 2019 (No. CIN-07011P-OMH01). All solvents used were obtained from Fisher Scientific (Fair Lawn, NJ, USA) or Sigma-Aldrich (St. Louis, MO, USA). NMR solvents were purchased from Cambridge Isotope Laboratories (Tewksbury, MA, USA). Benzene-1,2,4-triol was purchased from TCI (Portland, OR, USA). The material of taxifolin (4) was sourced from the compound repository of one of the authors (DF) at the University of Mississippi.

Instrumentation.

The preparative fractionation was performed on a centrifugal partition chromatography (CPC) extractor, SCPE-250 (Armen Instrument, Saint Ave, France). Semi-preparative HPLC was performed on a Shimadzu HPLC equipped with a binary solvent delivery pump LC-20AB, UV-Vis detector SPD-20A and a vacuum degasser, using a YMC-Pack ODS-AQ (250 × 10 mm, S-5, 12 nm) column. 1D and 2D NMR spectra were acquired at 298K on 600 MHz Bruker AVANCE spectrometer equipped with a 5 mm DCH CryoProbe, or 600 MHz JEOL spectrometer with a Royal Probe. High-resolution electrospray ionization (ESI) mass spectra were determined by a Bruker Impact II mass spectrometer, quadrupole time of flight (qTOF). ECD spectra were recorded on a JASCO-815 spectrometer with a 0.1 cm quartz cuvette, the sample concentration was 0.1 or 0.2 mg/mL in MeOH. The heated reaction was performed on Discover (CEM) microwave synthesizer.

Extraction and Isolation.

The dried powder of C. verum (8.17 Kg) was extracted with 70% acetone in water (30 L) by continuous percolation at room temperature. After evaporation in vacuo, the residue was then freeze-dried to afford the crude extract (648.4 g). 200 g of the crude extract was suspended in water and extracted with MeOAc (3 L × 3). The MeOAc-soluble extract (72.6 g) was fractionated by CC over DIAION HP20 with gradient elution by MeOH/H₂O (v/v) from 0:10 to 5:5 and washed by 10:0 to give seven fractions 1–7. Fraction 5 (MeOH/H₂O 4:6, 7.54 g) was subjected to separation on a CPC extractor using the method described before¹ with a modified solvent system (Hexane/EtOAc/MeOAc/H₂O, 0.4/4/1/4, v/v), the stationary retention volume ratio (S_f) was 0.66. The CPC fractions were pooled into 4 subfractions A-D according to their TLC profiles. Subfraction B (2.40 g) was separated by Sephadex LH20 CC (MeOH), affording three subfractions B1-B3. Subfraction B1 (1.66 g) contains the major trimer cinnamtannin B1, which was further purified by repeat Sephadex LH20 CC (MeOH). Subfractions B2 (122 mg) and B3 (149.5 mg) were fractionated over TOYOPEARL^® HW-40F CC (MeOH/H₂O, 9:1 v/v), respectively, to afford mixtures B2a and B3a. The mixture B2a was purified by semi-preparative HPLC (MeCN/H₂O/formic acid, 18:82:0.1, 3 mL/min) to yield compounds 1 (4.3 mg) and 2 (3.5 mg). Similarly, compound 3 (4.0 mg) was obtained by semi-preparative HPLC (MeCN/H₂O/formic acid, 12:88:0.1, 3 mL/min).

Epicatechin-(2β→O→7, 4β→6)-epicatechin-5,8-dione-(4β→8)-epicatechin; (Micro-PACBAR: EC=6pQEC-8EC) (1). Brownish amorphous powder; ECD (MeOH) λ_max (Δε) 220.5 (−11.3), 244 (1.4), 300.5 (3.0), 359.5 (−1.6) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₄₅H₃₅O₁₉ 879.1767; Found 879.1763. ¹H and ¹³C{¹H} NMR data see Table 1.

Epicatechin-(2β→O→7, 4β→6)-ent-catechin-5,8-dione-(4β→8)-epicatechin; (Micro-PACBAR: EC=6pQeC-8EC) (2): Brownish amorphous powder; ECD (MeOH) λ_max (Δε) 220.5 (−24.5), 244 (3.3), 301 (6.0) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₄₅H₃₅O₁₉ 879.1767; Found 879.1783. ¹H and ¹³C{¹H} NMR data see Table 1.

Epicatechin-(2β→O→7, 4β→8)-epicatechin-5,6-dione-(4β→6)-epicatechin-(4β→8)-epicatechin; (Micro-PACBAR: EC=8oQEC-6EC-8EC) (3): Brownish amorphous powder; ECD (MeOH) λ_max (Δε) 211.5 (−46.9), 236.5 (57.3), 272.5 (−0.1), 296 (20.2) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₆₀H₄₇O₂₅ 1167.2401; Found 1167.2425. ¹H and ¹³C{¹H} NMR data see Table 2.

Preparation of tetra-methylated taxifolin.

The tetramethlyation of taxifolin was modified from the method reported by Taniguchi and Monde.²⁹ Briefly, (+)-taxifolin (300 mg, 986 μM) in acetone (5 mL) added K₂CO₃ (1.02 g, 7.40 mM) and dimethyl sulfate (DMS) (935 μL, 9.86 mM), stirred under room temperature overnight, protected by N₂. The final solution was quenched by equivalent volume of distilled water, then added dichloromethane (30 mL) for partition (×3) and recovered the organic layer under vacuo. The 5,7,3′,4′-tetra-O-methylated taxifolin (5, 210 mg) was purified via RP-C18 and eluted in gradient elution from 10% to 100% MeOH in water.

(2R, 3R)-5,7,3′,4′-tetra-O-Methyl taxifolin (5). Pale yellow amorphous powder; HRMS (ESI) m/z [M + Na]⁺, Calcd for C₁₉H₂₀O₇Na 383.1101; Found 383.1107. ¹H NMR (DMSO-d₆, 600 MHz) δ 7.13 (s, H-2′), 7.02 and 6.95 (d, H-5′ and H-6′), 6.21 (s, H-6), 6.18 (s, H-8), 5.28 (brs, 3-OH), 5.03 (d, J = 11.4 Hz, H-2), 4.44 d (brd, J = 11.4 Hz, H-3), 3.79~3.76 (4×s, 4×3H, 4×OMe). ¹³C{¹H} NMR (DMSO-d₆, 150 MHz) δ 190.28 (C-4), 165.71 (C-5), 163.85 (C-9), 161.65 (C-7), 149.09 (C-3′/4′), 148.52 (C-3′/4′), 129.76 (C-1′), 120.75(C-6′), 111.59 (C-5′), 111.34 (C-2′), 103.55 (C-10), 93.59 (C-6), 93.00 (C-8), 82.65 (C-2), 72.42 (C-3), 55.95~55.57 (4×OMe).

Condensation of tetra-methylated taxifolin (5) and benzene-1,2,4-triol.

The solution of 5 (100 mg, 55 mM) in ethanol and acetone (5 mL, 4:1), added NaBH₄ (20 mg, 52.8 mM) stirred at room temperature. The reaction was monitored by HPLC, after 3 hours, benzene-1,2,4-triol (400 mg, 317.4 mM) was added into the reaction mixture, stirred to dissolve, then added 0.3 M HCl (5 mL), reaction was under 60 ⁰C microwave heating for 40 min. The reaction mixture was neutralized by 1 M NaHCO₃ (30 mL). The solution was diluted by ethyl acetate (100 mL) and water (100 mL) for partition (×3). After evaporation in vacuo, the residue was then subjected to semi-HPLC (MeCN/H₂O/formic acid, 36:64:0.1, 4.8 mL/min) for purification, yielding the condensation products 6 (43.5 mg) and 7 (18.5 mg).

Preparation of 4-quinone flavan-3-ols.

Ag₂CO₃ (240 mg, 217.6 mM) was added into the solution of 6 (13 mg, 6.9 mM) in THF (4 mL), stirred for 2 hours, and filtered the reaction mixture by celite. The reaction mixture of 7 (13 mg, 6.9 mM) was reacted for 4 hours under the same conditions. The two reactions yielded 8 (11.3 mg, 26.8%) and 9 (10.5 mg, 10.5%), respectively.

Compound 8. Brownish amorphous powder; ECD (MeOH) λ_max (Δε) 236 (−3.3), 257 (1.6), 283 (−2.9) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₂₅H₂₅O₉ 469.1493; Found 469.1502. ¹H and ¹³C{¹H} NMR data see Table 3.

Compound 9. Brownish amorphous powder; ECD (MeOH) λ_max (Δε) 237 (−1.6), 275.5 (3.6) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₂₅H₂₅O₉ 469.1493; Found 469.1489. ¹H and ¹³C{¹H} NMR data see Table 3.

Data Availability Statement.

The data underlying this study are made openly available in the Harvard Dataverse at https://doi.org/10.7910/DVN/GOM39V; this includes the raw NMR FIDs and ECD data for compounds 1–3, 8, and 9.