Proanthocyanidin Tetramers and Pentamers from Cinnamomum verum Bark and Their Dentin Biomodification Bioactivities

Abstract

To enable the further exploration of structure-activity relationships (SARs) of proanthocyanidins (PACs) with dentin biomodification abilities, Cinnamomum verum was selected for scaled-up purification of mixed A-/B-type, medium-size PAC oligomers. Sequential purification by centrifugal partition chromatography (CPC), Sephadex LH-20, and semi-prep HPLC chromatography yielded four underivatized tetrameric (5–8) and two pentameric (9–10) PACs. Their unambiguous structural characterization involved extensive spectral and chemical degradation approaches to show that epicatechin units are connected by plant-specific combinations of doubly-linked A- and singly-linked B-type interflavanyl linkages. The biomechanical properties (via dynamic mechanical analysis) and physico-chemical structure (via infrared spectroscopy) were assessed to evaluate the biomodification potency of PAC-treated collagen in a pre-clinical dentin model. This study revealed that (4→8) vs. (4→6) bonds in PAC interflavan linkages have limited influence on biomechanical outcomes of dentin. By exhibiting a 25-fold increase in the complex modulus of treated dentin compared to control, aesculitannin E (5) was found to be the most potent PAC known to date for enhancing the mechanical properties of dentin in this preclinical model.

Graphical Abstract

INTRODUCTION

Current dental restorative techniques rely on bonding resin composites to the dental tissue to recover the tooth form and function. Dentin is a complex mineralized tissue and the main substrate for adhesive restorations. Type I collagen comprises of 90% of the organic extracellular matrix (ECM) of dentin and, along with non-collagenous proteins, constitutes the 3-dimensional scaffold matrix of the tooth.^1,2 Plant species-specific proanthocyanidins (PACs) have been identified as potent dentin biomodification agents, mediating non-enzymatic collagen cross-linking and enhancing the biomechanical properties and biostability of dentin.¹ Among the most promising PAC-rich plant sources with identified dentin biomodification potential are Pinus massoniana, Cinnamomum verum, and Vitis vinifera. Varying by producing plant species, the degree of polymerization (DP) and interflavan linkage (IFL) of PACs have determinant effects on their interaction with dentin collagen.³

PACs are also known as condensed tannins and are oligomeric and polymeric flavan-3-ols. Analytical techniques, separation methods, as well as the structural elucidation strategies of PACs have been progressing substantially in the last two decades. Modern separation methods have enabled the selective and scalable isolation of enriched PAC fractions to single, underivatized compounds. This has enabled biological evaluation which was unfeasible when working with derivatized PACs (e.g., after phloroglucinolysis). Suitable advanced separation methods include high-speed counter current chromatography (HSCCC), centrifugal partition chromatography (CPC), and diol-phase HPLC.⁴

Mass spectrometry (MS) appears to be an efficient tool to differentiate PAC sources based on molecular weights of different types of PACs, such as UHPLC-MS, electrospray ionization (ESI) and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS, which the later has notable detectability of PACs with higher DP up to 30 with proper sample preparation.⁵ A recent review has summarized the advancement in MS based analytical methods on PAC mixtures.⁶ More recently, MALDI-TOF MS combined with multivariate analysis has been applied to discriminate the authenticity of different PAC sources and mixtures.⁷

Regarding structural characterization, the application of low temperature NMR allows the acquisition of NMR spectra of underivatized PACs with tendency to exhibit dynamic equilibria due to atropisomerism, which is caused by rotational hindrance around the interflavanyl bonds, often appearing as peak broadening in room temperature NMR spectra.^8–10 At the same time, ultra high field NMR instruments and more sensitive probes that can detect CH long-range correlations and the nuclear Overhauser effect (NOE) interactions more effectively are essential to determine the nature of the IFLs.¹¹ Moreover, ¹H iterative full spin analysis (HiFSA) has been recognized as a powerful tool in characterizing subtle but diagnostic effects in the ¹H NMR spectra of PACs, including higher order effects and H/D exchange reactions, making it feasible to accurately generate the full ¹H NMR assignments for even more complex PACs.^12,13

Chemical depolymerization remains a vital tool when establishing the chemical structures, particularly regarding the absolute configuration of the monomers in higher oligomeric PACs. This goal can be achieved by phloroglucinolysis,¹⁴ thiolysis,¹⁵ and furanolysis.¹⁶ The latter is a more recent and improved depolymerization method that uses menthofuran instead of phloroglucinol or 2-mercaptoethanol as the nucleophilic trapping reagent. Furanolysis is more efficient compared to previous methods and able to fully depolymerize condensed tannins.

Guided by dentin bioactivity assessment, our group has identified a series of A-type oligomeric PACs from pine bark,^{10,12,13,17,18} as well as B-type PACs from grape seeds,¹⁹ all with potent dentin biomodification activity. To enhance the communication across disciplines, particularly at the chemistry-biology interface, better reveal the complexity of PACs structure, and foster the establishment of dentin activity-structure relationships (SARs), we recently proposed the PAC Block ARray (PACBAR) system as more precise and better readable structural descriptors of PACs.²⁰ Applying PACBAR recently enabled the interdisciplinary definition of the first set of SARs that underlie the multi-facetted dental bioactivities of PACs isolated from pine bark.²¹

The inner bark of Cinnamomum verum (Lauraceae), a woody plant native to Sri Lanka, is widely valued as the source of cinnamon spice. In addition to its taste-determining essential oil, C. verum bark is also rich in mixed AB-type trimeric and tetrameric PACs, as evident from diol-HPLC and LC-MS profiling, and has shown promising dental bioactivities.³ Recently, two tetramers and one major trimer were identified from C. verum and their potential dentin biomodification abilities reported.^3,10 In the course of expanding the dentin bioassays screening library of chemically diverse PACs, C. verum was scaled-up targeting specifically the medium oligomeric PACs to enable extensive dentin biomodification studies and further explore the SAR space of the dental bioactivity of PACs.

Using monomeric to trimeric catechins (1–4) as reference points, we herein report the structural elucidation and dentin biomodification activities of four tetrameric (5–8) and two pentameric (9–10) PACs. The two-dimensional structures and absolute configuration of new PACs 7 and 9, as well as the previously isolated 5²² and 6²³, which had been identified via derivatization method only, were established by comprehensive spectroscopic analysis and furanolysis. The dentin biomodification activities of 5–10 were evaluated by dynamic mechanical analysis (DMA) and biochemical analysis via infrared spectroscopy.

RESULTS AND DISCUSSION

Scale-up Purification of PAC Oligomers.

CPC has been successfully applied in preparation of the organic-soluble PAC-enriched fractions by DP from C. verum, but single compound purification still relied on semi-preparative HPLC which limited the exploration of bioactivities and the discovery of less abundant PACs.^10,24 As stated by the diol-phase HPLC analysis (Figure 2A), the trimers (DP3) and tetramers (DP4) were dominant components of C. verum extract, which contained A/B-mixed type IFLs based on molecular ion peaks (e.g. m/z 865.1981, 1153.2634) detected by LCMS (Figure 2B). After extraction by 70% aqueous acetone, followed by partition with methyl acetate (MeOAc) to remove polar components such as PAC polymers and sugars, the PAC-enriched fraction was subjected to Diaion HP-20 resin chromatography column (CC), taking advantage of it higher loading capacity, to further separate PAC oligomers from monomers and polymers. Using an optimized two-phase solvent system (HEMaWat, 0.4/4/1/4),²⁵ the PAC-enriched fraction was separated into three DP clusters in a single run by CPC. Fraction A consisted of phenolic compounds, monomers and dimers; fraction B was predominantly trimers; fraction C from the extrusion portion of the CPC separation contained oligomeric PACs with DP 4~5 (Figure 2C–2D). The major trimer cinnamtannin B1 (4) was further purified from Sephadex LH-20 CC separation of fraction B, which resulted in a good overall yield for this compound, in the amount of 1.11% of dried plant material. The rest of the subfraction as well as fraction C were further purified by reverse-phase C18 CC and semi-prep HPLC afforded several structurally diverse underivatized tetramers and pentamers (5–10) in substantial amounts and with purities from 85% to 98% as determined by quantitative ¹H NMR (Figures S59–S64), which not only allowed comprehensive spectroscopic analysis and acid-catalyzed cleavage reactions, but also enabled extensive dentin biomodification studies. Therefore, the preparation method reported here can serve as an efficient and feasible workflow for the scale-up purification of AB-mixed type oligomeric PACs for comprehensive bioactivity evaluation.

Figure 2.

Analysis and gram-scale purification of PAC oligomers from C. verum. (A) Diol HPLC profile (280 nm) and (B) UHPLC-MS (TIC: m/z 100~2000) profile of C. verum extract. (C) CPC separation by DP clusters. (D) TLC of CPC fractions.

Structure Elucidation of PACs - Stereochemical Reference Points.

Until all possible permutations of enantiomeric flavan-3-ols and their IFLs have been characterized rigorously for a given oligomerization level by ¹H and ¹³C NMR, depolymerization reactions remain essential for the study of absolute stereochemistry of PACs. Furanolysis advances this element of PAC structural analysis as it exhibits higher efficiency at low concentration. This is an advantage over the classical reagents phloroglucinol and 2-mercaptoethanol, which require excess amounts, and simplifies the preparation of the cleavage products for further analysis.¹⁶ Therefore, this study employed furanolysis for PAC depolymerization. The terminal unit epicatechin (1) as well as the menthofuran-substituted derivatives 2 and 3 were obtained from the dimer, procyanidin B2 (11), and the trimer, cinnamtannin B1 (4), under acidic conditions using just one equivalent of menthofuran (Scheme 1). Compound 2 was determined to be epicatechin-(4→2′′)-menthofuran by comparing it 1D NMR spectrum (Table 1) with reported data.¹⁶ The relative configuration of C-4 was determined as being trans-configured relative to C-2 based on the observed γ-gauche effect, which led to the chemical shift of C-2 appearing at relatively low frequency (δ_C 75.29).¹⁷ The high-amplitude positive Cotton Effect (CE) at 220~240 nm in the electronic circular dichroism (ECD) spectrum (Figure 4A) indicated that the C-4 menthofuran group was in β orientation. Therefore, the absolute configuration of 2 was established as (2R,3R,4S)-epicatechin-(4β→2′′)-menthofuran. According to the newly proposed PAC nomenclature,¹⁷ the microPACBAR code of 2 is EC-2″MF.

Scheme 1.

Furanolysis of procyanidin B2 (11) and cinnamtannin B1 (4) under acid-catalysis condition led to the identification of reference compounds epicatechin (1), epicatechin-(4→2′′)-menthofuran (2) and epicatechin-(2β→O→7, 4β→8)-epicatechin-(4β→2′′)-menthofuran (3).

Table 1.

¹H and ¹³C NMR data of 2 and 3 in CD₃OD^a

		2b			3c
Ring	No.	δC, type	δH (J, Hz)		δC, type	δH (J, Hz)
Unit I
C	2	75.29, CH	4.9171, s		100.07, C
	3	70.47, CH	3.9809, d (2.15)		68.10, CH	4.1233, d (3.36)
	4	38.07, CH	4.2259, d (2.15)		29.33, CH	4.4540, d (3.36)
A	5	157.06, C			157.08, C
	6	95.14, CH	5.9412, d (2.31)		98.35, CH	6.0330, d (2.32)
	7	157.50, C			158.13, C
	8	94.39, CH	5.9808, d (2.31)		96.69, CH	6.0816, d (2.32)
	9	156.21, C			154.21, C
	10	98.68, C			104.14, C
B	1′	130.92, C			132.26, C
	2′	113.91, CH	6.8982, d (2.00)		115.60, CH	7.1313, d (2.12)
	3′	144.66, C			145.67, C
	4′	144.43, C			146.77, C
	5′	114.66, CH	6.7429, d (8.23)		115.92, CH	6.8121, d (8.31)
	6′	117.90*, CH	6.6926, dd (8.23, 2.0)		119.77, CH	7.0366, dd (8.31, 2.12)
Unit II
F	2				78.61, CH	5.0976, brs
	3				71.33, CH	4.0002, d (1.97)
	4				39.83, CH	4.2525, d (1.97)
D	5				157.39, C
	6				96.69, CH	6.0817, s
	7				152.98, C
	8				106.87, C
	9				152.29, C
	10				102.38, C
E	1′				131.21, C
	2′				115.93, CH	7.1476, d (2.12)
	3′				146.01, C
	4′				146.31, C
	5′				115.65, CH	6.8153, d (8.18)
	6′				120.34, CH	6.9680, d (8.18, 2.12)
MF	2″	147.62, C			148.78, C
	3″	114.83, C			116.03, C
	3a″	117.90*, C			119.23, C
	4″A	19.72, CH2	2.2825, dddd (−16.08, 10.11, 4.81, 2.01)		20.97, CH2	2.3109, dddd (−15.64, 10.61, 5.07, 2.65)
	4″B		2.3345, ddddd (−16.08, 5.33, 3.42, 2.05, 1.27)			2.3642, dddd (−15.64, 5.40, 3.16, 1.73)
	5″A	31.26, CH2	1.8234, ddddd (−13.05, 4.81, 3.76, 3.42, 1.68)		32.56, CH2	1.3584, dddd (−13.07, 10.63, 10.61, 5.40)
	5″B		1.3446, dddd (−13.05, 10.58, 10.11, 5.33)			1.8348, dddd (−13.07, 5.07, 3.16, 2.82)
	6″	29.77, CH2	1.8858, ddddd (10.58, 9.56, 6.72, 5.37, 3.76)		31.00, CH2	1.8976, ddddd (10.63, 9.62, 6.69, 5.38, 2.82)
	7″A	31.01, CH2	2.1116, dddd (−16.01, 9.56, 2.21, 2.05)		32.21, CH2	2.1201, ddd (−16.06, 9.62, 2.65)
	7″B		2.5897, dddd (−16.01, 5.37, 1.68, 1.27)			2.6023, ddd (−16.06, 5.38, 1.73)
	7a″	147.77, C			149.26, C
	8″	6.68, CH3	1.6848, s		8.12, CH3	1.7826, s
	9″	20.55, CH3	1.0696, d (6.72)		21.83, CH3	1.0778, d (6.69)

^aThe δ_H and J values were generated via HiFSA.

^bRecorded on JEOL 400/100 MHz, resp.

^cRecorded on Bruker 600/150 MHz, resp.

^*Overlapped signals.

Figure 4.

ECD spectra of 2 and 3 (A) and PACs 4–10 (B), with an expansion of the 260 ~300 nm region (C).

Compound 4 was isolated as the major trimer from C. verum, the structure was determined as cinnamtannin B1 by comparison of the ¹H and ¹³C NMR data with those from a sample previously authenticated by our group.¹⁰ The furanolysis of 4 produced the menthofuran substituted A-type dimer 3. Its molecular formula was C₄₀H₃₆O₁₃ based on its HRESIMS spectrum (Figure S46) and ¹³C NMR data (Table 1). The 2,4-trans configuration of 3 was determined by the observed γ-gauche effect, which shifted the resonance of C-2 (δ_C 78.61) to lower frequency compared to 2,4-cis configuration.¹⁷ The ECD spectrum (Figure 4A) showed a positive CE at 220~240 nm, indicating the β orientation of C-4. Accordingly, 3 was established as epicatechin-(2β→O→7, 4β→8)-epicatechin-(4β→2′′)-menthofuran, equal to EC=8EC-2″MF in microPACBAR code.

Characterization of Tetrameric and Pentameric PACs.

Compound 5 was isolated as a light-brownish amorphous powder. The molecular ion at m/z 1153.2616 [M+H]⁺ observed in its HRESIMS spectrum (Figure S47), and four carbon resonances (CH/CH₂) in the low frequency of the ¹³C NMR spectrum (Figure S10), which belonged to the C-4 carbons of (epi)catechin monomers, indicated a tetrameric PAC structure containing one A-type and two B-type IFLs. Four AMX spin systems of 1,3,4-trisubstituted aromatic rings (B-ring) were observed between δ_H 7.35~6.60 ppm in the ¹H NMR spectrum (Figure S9). In general, the AB spin system at δ_H 5.980 and 6.028 corresponded to the H-6/8 (A-ring) of the top unit, and singlet aryl resonances at δ_H 5.875 and 5.903 belonged to either H-6 or H-8 of lower units. The absence of one aryl hydrogen (H-6/8) resonance in tetramer 5 indicated the H/D exchange reactivity, which occurs frequently in deuterated protic solvents such as CD₃OD or D₂O and can occur relatively rapidly under acid conditions via electrophilic-substitution or keto-enol tautomerization.¹² In the present case, residual amounts of formic acid (2%, Figure S59) retained in the sample after semi-prep HPLC facilitated this process.

In the ¹H NMR spectrum of 5 (Figure S9), the AX spin system at δ_H 3.190 (d, J = 3.5 Hz) and 4.133 (d, J = 3.5 Hz) were determined as H-3 and H-4 of a doubly linked epicatechin unit, which was further assigned to unit I by the HMBC correlations between H-3/4 and H-6/8 with C-10 (unit I). The methylene resonances (δ_H 2.808, 2.929) appearing at lower frequency in the ¹H NMR spectrum belonged to the H₂-4 of terminal epicatechin unit. Detailed analysis of the 2D NMR data allowed the assignment of all hydrogen and carbon resonances, thereby establishing the ABB-type IFLs of 5. The (2→O→7, 4→8) A-type linkage between unit I and II was determined based on ROESY correlation between I-H-4 with II-H-2, as well as by HMBC cross peaks from I-H-4 to II-C-7/8/9. The (4→8) B-type IFLs from units II to III and III to IV were supported by ROESY correlations between II-H-2 with III-H-2′/6′, as well as III-H-3 with IV-H-2′, respectively. The linkage sites were further confirmed by the HMBC correlations between II-H-4/III-H-4 to II-C-9 and III-H-4 with IV-C-7/8/9, IV-H-4 with IV-C-9 (Figure 3).

Figure 3.

Key HMBC (H→C) and COSY (H–H) correlations in 5.

The 2,3-cis configurations of units II to IV were established on the basis of observed small coupling constants or the occurrence of broad singlets for H-2/3 (Table 3). According to the γ-gauche effect, the B-type IFLs were determined as trans-configured to C-2 between unit II/III and III/IV, for which the carbon chemical shifts (II-C-2: δ_C 78.91, III-C-2: δ_C 77.66) were observed at relatively low frequencies compared to 2,4-cis configured PACs (C-2: δ_C ~ 84) as reported recently.¹⁷ Collectively, 5 was established as an ABB-type tetramer in terms of its IFLs, and as a PAC that consisted of four epicatechin units. This structure has been reported previously as aesculitannin E, albeit without providing complete NMR data and solely on the basis of thiolysis.²²

Table 3.

¹H NMR data of 5–7 and 9 in CD₃OD.

Unit	Ring	No.	5a	6b	7a	9a
I	C	2		5.004, s
		3	3.190, d (3.5)	3.932, d (2.2)	3.915, d (3.1)	3.447, d (3.6)
		4	4.133, d (3.5)	4.612*, brs	4.630, d (3.1)	4.340*, brs
	A	6	6.028, d (2.3)	6.007, d (2.3)	5.979, d (2.3)	5.995, d (2.4)
		8	5.980, d (2.3)	5.936, d (2.3)	6.051, d (2.3)	6.061, d (2.4)
	B	2′	7.004, d (2.2)	6.897, d (2.1)	7.070, d (2.1)	7.113, d (2.1)
		5′	6.807, d (8.3)	6.825–6.843*	6.902, d (8.4)	6.731, d (8.4)
		6′	6.851–6.870*	6.666, m	6.966, dd (8.4, 2.1)	6.775*
II	F	2	5.689, s			5.656, s
		3	4.215, m	3.361, d	3.330, d (3.2)	4.148, d (2.4)
		4	4.591, brs	4.125, d (3.6)	4.273, d (3.2)	4.459, brs
	D	6	5.875, s		6.133, s
		8		5.981 s
	E	2′	7.324, d (2.1)	7.046, d (2.1)	7.333, d (2.3)	7.367, d (2.1)
		5′	6.851–6.870*	6.825–6.843*	6.835, d (8.3)	6.852–6.872*
		6′	7.210, dd (8.3, 2.1)	6.870, dd (8.2, 2.1)	7.214, dd (8.3, 2.3)	7.274, dd (8.4, 2.1)
III	I	2	4.555, s	5.771, s	5.738, s	4.670, s
		3	3.640, d (2.1)	3.992, brs	4.138, brs	3.811, brs
		4	4.679, brs	4.612*, brs	4.590, s	4.552, brs
	G	6	Abs	5.830, s	5.782, s	6.121, s
	H	2′	6.662, d (2.0)	7.414, d (2.1)	7.315, d (2.2)	6.824, d (2.0)
		5′	6.688, d (8.2)	6.825–6.843*	6.771, d (8.3)	6.696, d (8.2)
		6′	6.607, dd (8.2, 2.0)	7.228, brd (8.0, 2.1)	7.200, dd (8.3, 2.2)	6.456, dd (8.2, 2.0)
IV	L	2	4.913, s	4.407, s	4.402, s	4.340*, brs
		3	4.292, brs	3.869, m	3.861, brs	3.843 brs
		4	2.808, dd (17.0, 2.9) 2.929, dd (17.0, 4.6)	2.852, m	2.836, m	2.165, dd (17.3, 4.8) 2.309, d (16.6)
	J	6	5.903, s	6.111, s	6.112, s	6.023 s
	K	2′	7.089, d (2.0)	6.825–6.843*	6.809, d (2.0)	7.008, d (2.0)
		5′	6.745, d (8.2)	6.782, m	6.763, d (8.3)	6.775*
		6′	6.879, dd (8.2, 2.0)	6.825–6.843*	6.729, dd (8.3, 2.0)	6.852–6.872*
V	M	2				4.890, s
		3				3.919, brs
		4				4.320, brs
	O	6				5.918, d (2.3)
		8				6.026, d (2.3)
	N	2′				7.113, d (2.0)
		5′				6.731, d (8.4)
		6′				6.442, dd (8.4, 2.0)

^aRecorded in 600 MHz, at 298 K.

^bRecorded in 800 MHz, at 278 K.

^*Overlapped signals; Abs, absence of resonance due to H/D exchange.

Noteworthy, the same planar structures of 5 were retrieved from the Dictionary of Natural Products (https://dnp.chemnetbase.com), with different stereochemistry or incomplete NMR assignment.^22,26,27 Although differentiate the monomeric unit, e.g. epicatechin from catechin can be achieved from the ³J coupling constants of H-2 and H-3, and the relative configuration of C-4 can be established via γ-gauche effect by comparing the C-2 chemical shifts, it can be risky to determine whether the monomeric units are (epi)catechin or its enantiomers based on NMR comparison with reference data.²⁶ Unfortunately, these kinds of erroneous assignments were quite common in PAC literatures, which press the need for rigorous structural elucidation including the absolute stereochemistry determination for both new and known PACs which lack prior comprehensive characterizations. The intricate stereochemistry of PACs contributes to the chemical diversity of this class of natural products which are widely recognized but still poorly understood.

The absolute configuration of 5 was therefore determined by a combination of ECD spectroscopy and furanolysis. As shown in Figure 4, 5 exhibited a positive CE at 220~240 nm and a negative CE at 260~280 nm, which established the β orientation of the IFL and the α orientation of B-ring, respectively. The furanolysis products of 5 were identified as epicatechin (1), EC-2″MF (2), and EC=8EC-2″MF (3) by LCMS (Figure S55) and chiral-phase HPLC comparison with the respective reference compounds (Figure 5). Thus, PAC 5 was finally determined as epicatechin-(2β→O→7,4β→8)-epicatechin-(4β→8)-epicatechin-(4β→8)-epi-catechin or EC=8EC-8EC-8EC in microPACBAR code.

Figure 5.

Identification of furanolysis products of 5–7 and 9 by chiral phase HPLC comparison with reference compounds 1–3.

Compound 6 was also a PAC tetramer, bearing one A-type and two B-type IFLs based on its molecular ion at 1153.2616 [M+H]⁺ in the HRESIMS (Figure S48). The NMR spectra of 6 were acquired at 278 K to resolve the dynamic rotational isomerism that caused peak broadening in the ambient temperature NMR experiment (Figures S15–S20).¹⁰ The AMX spin system resonating at δ_H 5.004 (s, H-2), 4.612 (brs, H-4), and 3.932 (d, J = 2 Hz, H-3) observed in the ¹H NMR spectrum (Figure S15) was assigned to unit I, based on the HMBC correlations observed from H-3/4 and H-6/8 to C-10, which already established the B-type IFL between unit I and II. The furanolysis cleavage products of 6 were identified as epicatechin (1), EC-2″MF (2), EC=8EC-2″MF (3), and cinnamtannin B1 (4), based on LC-MS analysis in comparison with authentic samples (Figure S56). This suggested BAB IFLs between the epicatechin monomeric units in 6. The exact linkage location was determined to be 4→6 from unit I to II, as the singlet resonance at δ_H 5.981 (II-H-8) showed NOE correlation to II-H-2′, as well as due to the HMBC correlations between II-H-8 and II-C-9/10. The linkage between unit III and the terminal unit was established as 4→8 according to the NOE correlation between III-H-2 and IV-H-2′, as well as via HMBC cross peaks between III-H-4/IV-H-4 and IV-C-9. The B-type IFLs were both determined as involving trans configuration relative to C-2, because the chemical shifts of C-2 (I-C-2: δ_C 76.69; III-C-2: δ_C 79.50) appearing at relatively lower frequency according to the γ-gauche effect. By analyzing the ECD spectrum (Figure 4) and chiral phase HPLC results (Figure 5), the absolute configuration of 6 was established unambiguously as epicatechin-(4β→6)-epicatechin-(2β→O→7,4β→8)-epicatechin-(4β→8)-epicatechin, microPACBAR coded as EC-6EC=8EC-8EC (Figure 1). This structure had previously been established as pavatannin C-1 through NMR analysis of its peracetate derivative.²³

Figure 1.

The chemical structures and PAC Block ARray (PACBAR) designations of 1–10 (MF, menthofuran residue; EC, epicatechin).

The HRESIMS (Figure S49–S50) of compounds 7 and 8, besides their ¹³C NMR carbon count (Table 2 and Table S2) established the molecular formula of C₆₀H₄₆O₂₄, indicating they are a pair of isomeric tetramers both bearing two A-type and one B-type IFL. The ¹H NMR spectrum of 7 showed signals of major and minor rotamer at a ratio of 2.7:1 (Figure S61). In the aryl hydrogen region of the ¹H NMR spectrum of 7 (Figure S21), the AB spin system resonating at δ_H 5.979 and 6.051 (d, J = 2.3 Hz) were ascribed to the H-6/8 of the top unit, whereas the three singlets at δ_H 5.782, 6.112 and 6.133 were ascribed to H-6 or H-8 of three lower units. Two sets of AX spin systems of the oxymethine resonances at δ_H 3.915 and 4.630 (d, J = 3.1 Hz), as well as δ_H 3.330 and 4.273 (d, J = 3.2 Hz), were assignable to H-3/4 of the doubly linked units.

Table 2.

¹³C NMR data of 5–7 and 9 in CD₃OD [δ_C, type]

Unit	Ring	No.	5a	6b	7a	9a
I	C	2	99.96, C	76.69, CH	99.94, C	100.22, C
		3	67.16, CH	72.78, CH	68.78, CH	67.07, CH
		4	28.79, CH	37.59, CH	29.87, CH	29.05, CH
	A	5	154.09, C	157.22, C	157.78, C	156.78, C
		6	96.56, CH	95.84, CH	97.89, CH	98.46, CH
		7	157.81, C	157.77, C	158.13, C	157.95, C
		8	98.27, CH	96.02, CH	96.48, CH	96.52, CH
		9	156.67, C	157.66, C	154.32, C	154.32, C
		10	104.92, C	101.82, C	103.69, C	105.01, C
	B	1′	132.39, C	132.44, C	132.36, C	132.23, C
		2′	115.72, CH	115.16, CH	115.59, CH	115.65, CH
		3′	145.66, C	145.31c, C	146.85c, C	145.39c, C
		4′	146.56, C	145.20c, C	146.84c, C	146.56c, C
		5′	115.64, CH	115.11, CH	115.85, CH	116.86, CH
		6′	119.92, CH	119.02, CH	119.50, CH	119.88, CH
II	F	2	78.91, CH	99.65, C	100.50, C	78.70, CH
		3	72.51, CH	66.83, CH	66.44, CH	72.51, CH
		4	38.56, CH	28.98, CH	29.24, CH	39.04, CH
	D	5	157.35, C	152.03, C	156.06, C	154.84, C
		6	95.91, CH	109.63, C	98.65, CH	106.95, C
		7	151.03, C	156.45, C	153.70, C	148.71, C
		8	106.45, C	97.18, CH	108.89, C	106.66, C
		9	151.58, C	154.23, C	148.60, C	150.32, C
		10	106.49, C	104.90, C	106.65, C	106.77, C
	E	1′	132.02, C	132.27, C	131.95, C	131.55, C
		2′	116.57, CH	115.63, CH	115.56, CH	116.89, CH
		3′	146.24, C	145.31c, C	146.60c, C	145.97c, C
		4′	145.59, C	146.38c, C	146.73c, C	146.46c, C
		5′	115.98, CH	115.76, CH	116.08, CH	116.06, CH
		6′	121.20, CH	119.83, CH	120.23, CH	121.70, CH
III	I	2	77.66, CH	79.50, CH	78.70, CH	77.70, CH
		3	73.26, CH	72.93, CH	72.67, CH	72.55, CH
		4	37.30, CH	38.07, CH	38.21, CH	37.78, CH
	G	5	156.06, C	155.86, C	155.82*, C	157.39, C
		6	Abs	95.92, CH	95.83, CH	96.98, CH
		7	151.06, C	150.88, C	151.39, C	157.27, C
		8	109.12, C	106.06, C	106.57, C	109.18, C
		9	157.33, C	152.34, C	151.40, C	155.64, C
		10	100.72, C	106.31, C	106.80, C	98.93, C
	H	1′	132.97, C	131.12, C	132.73, C	132.90, C
		2′	115.42, CH	117.20, CH	116.51, CH	115.87, CH
		3′	145.86, C	145.28c, C	146.65c, C	145.33c, C
		4′	145.41, C	146.16c, C	146.23c, C	145.76c, C
		5′	116.19, CH	115.68, CH	115.62, CH	115.66, CH
		6′	119.46, CH	122.20, CH	120.99, CH	119.43, CH
IV	L	2	79.61, CH	80.13, CH	80.25, CH	79.53, CH
		3	66.78, CH	67.14, CH	67.61, CH	72.55, CH
		4	29.70, CH2	29.60, CH2	29.79, CH2	29.55, CH2
	J	5	156.70, C	155.76, C	155.62, C	155.32, C
		6	97.51, CH	96.14, CH	96.45*, CH	96.09, CH
		7	156.50, C	155.37, C	151.32, C	155.13, C
		8	107.00, C	108.76, C	108.74, C	107.80, C
		9	154.18, C	151.95, C	156.10, C	156.30, C
		10	100.73, C	99.67, C	100.05, C	101.83, C
	K	1′	132.01, C	133.00, C	133.11, C	132.74*, C
		2′	115.18, CH	115.31, CH	115.34*, CH	115.56, CH
		3′	146.54, C	145.31c, C	145.94c, C	145.64c, C
		4′	145.89, C	145.70c, C	145.45c, C	145.78c, C
		5′	116.15, CH	115.65, CH	116.13*, CH	115.89, CH
		6′	119.07, CH	119.41, CH	119.43, CH	119.79, CH
V	M	2				77.10, CH
		3				72.88, CH
		4				37.52, CH
	O	5				159.62, C
		6				96.80, CH
		7				159.73, C
		8				96.36, CH
		9				158.04, C
		10				98.39, C
	N	1′				132.74*, C
		2′				115.96, CH
		3′				145.29, C
		4′				145.56, C
		5′				116.02, CH
		6′				120.54, CH

^aRecorded in 150 MHz at 298 K.

^bChemical shifts were read from HSQC and HMBC spectra. Recorded in 200 MHz at 278 K.

^cInterchangeable assignment.

^*Overlapped signals. Abs - absent signals due to H/D exchange.

LC-MS analysis of the furanolysis products of 7 showed a molecular ion at m/z 1011.2715, corresponding to a menthofuran substituted A-type trimer, and a free epicatechin unit at m/z 313.0671 (Figure S57), collectively confirming 7 to have AAB-type IFLs. All its NMR resonances were fully assigned by analyzing the HSQC, HMBC, and ¹H-¹H COSY spectra. The exact positions of the A-type linkages of unit I/II and II/III were determined as (2→O→7, 4→8), according to the ROESY correlations found between I-H-4 with II-H-2′/6′, I-H-2′ with II-H-6, and II-H-4 with III-H-2′/6′. The (4→8) B-type linkage between units III and IV was established based on the ROESY cross peaks between III-H-2 and IV-H-2′/6′, as well as the HMBC correlations from both III-H-4 and IV-H-4 to IV-C-9. The 2,3-cis configurations were established for unit III and IV based on the observation of broad singlet resonances for H-2/3. The C-4 was in trans position to C-2 (δ_C 78.70) in unit III according to the γ-gauche rule. As 7 showed an ECD pattern congruent to those of the ECD spectra of 4–6 (Figure 4), its monomers had to have 2ɑ,4β absolute configuration. The terminal unit was confirmed as epicatechin by chiral phase HPLC analysis of its furanolysis products (Figure 5). In conclusion, 7 was established as the previously unreported PAC tetramer, epicatechin-(2β→O→7,4β→8)-epicatechin-(2β→O→7,4β→8)-epicatechin-(4β→8)-epicatechin, equivalent to EC=8EC=8EC-8EC in microPACBAR code.

Comparing the ¹H NMR spectrum of 8 with 7, there were two sets of AB spin systems and one singlet resonance in the aryl hydrogen region (Figure S27), indicating the existence of a “branched (4→6 IFL)” unit in 8. After detailed assignment of the hydrogen and carbon resonances of each unit in 8, the linkages were determined as (2→O→7,4→8) between unit I and II, (4→8) from unit II to the terminal unit III, and (2→O→5,4→6) between unit II and the “branched” unit IV. These assignments reflected the observed HMBC correlations between I-H-4/IV-H-4 with C-6/7 of unit II, and ROESY cross peaks between II-H-4 with III-H-2′/6′. Thereby, tetramer 8 was established as previously described parameritannin A-2 via comparison of the ¹H and ¹³C NMR resonances (Table S3) as well as ECD spectra (Figure 4) with reported data.²⁸

The HRESIMS spectra (Figure S51–S52) of 9 and 10 showed both have the identical molecular formula of C₇₅H₆₀O₃₀. The presence of five aliphatic methine or methylene resonances (δ_C 39~29 ppm) observed at the low frequency of its ¹³C NMR spectrum ascribed to the C-4 carbons of (epi)catechin monomers, suggested that compounds 9 and 10 were isomeric pentamers having one A-type and three B-type IFLs. In the ¹H NMR spectrum of 9 (Figure S33), the AB spin systems resonating at δ_H 5.995 and 6.061 (d, J = 2.4 Hz), δ_H 5.918 and 6.026 (d, J = 2.3 Hz) were ascribed to the H-6/8 of the top unit or a branch unit, whereas the singlets at δ_H 6.023 and 6.121 were assigned to H-6 or H-8 of the lower units. The AX spin system resonating at δ_H 3.447 and 4.340 (d, J = 3.6 Hz) were assigned to H-3 and H-4 of a doubly linked unit, which was further confirmed by the observed HMBC correlation between H-4 and the ketal carbon at δ_C 100.22. The methylene group (δ_H 2.165 and 2.309) appearing at lower frequency in the ¹H NMR spectrum of 9 was assigned to H₂-4 of the terminal unit. The NMR resonances of each unit were fully assigned by analysis of the HSQC, HMBC, and ¹H-¹H COSY spectra.

The presence of an A-type IFL (2→O→7, 4→8) was established between units I and II due to I-H-4 showing a ROESY correlation with II-H-2′/6′. The branching unit was determined to be unit V, being connected to unit II via a (4→6) IFL, based on the observation of HMBC correlations between I-H-4/V-H-4 with II-C-7 and II-H-4/V-H-4 with II-C-5, as well as ROESY cross peaks from V-H-3 to I-H-2′/6′. Similarly, the IFLs between units II/III and III/IV were both determined as (4→8) due to observed HMBC correlations between II-H-4/III-H-4 with III-C-9, II-H-4/II-H-3/III-H-6 with III-C-8, III-H-4/IV-H-4 with IV-C-9, and III-H-4/III-H-3/IV-H-6 with IV-C-8. Moreover, in the ROESY spectrum of 9, II-H-5 was correlated with III-H-2′/6′, and III-H-5 was correlated with IV-H-2′/6′.

The 2,3-cis configuration of 9 was established by the observation of small coupling constants as indicated by the occurrence of broad singlets for H-2 and H-3 (Table 3). According to the γ-gauche effect rule, the C-4 of units II/III/V had to be in trans position to C-2, as the chemical shifts of II/III/V-C-2 were at relatively low frequency at δ_C 78.70, 77.70, and 77.10, respectively. The high-amplitude positive CE at 220~240 nm and low-amplitude negative CE at 260~290 nm shown in the ECD spectrum of 9 (Figure 4) led to the assignment of the absolute configurations of C-2 and C-4 as ɑ and β, respectively. This was further confirmed by LC-MS (Figure S58) and chiral phase HPLC (Figure 5) analysis of the furanolysis products of 9, which were identified as epicatechin (1), EC-2″MF (2), and EC=8EC-2″MF (3) by comparison with authentic compounds. In summary, compound 9 was determined as epicatechin-(2β→O→7,4β→8)-[epicatechin-(4β→6)]-epicatechin-(4β→8)-epicatechin-(4β→8)-epicatechin. This pentameric PAC represents a new natural product that can be described as EC=8EC(6-EC)-8EC-8EC in microPACBAR code.

The 1D NMR spectra of 9 and 10 were highly similar. Careful comparison found that some of the oxymethine signals ascribable to H-2 were shifted in the ¹H NMR spectrum of 10 (Figure S39) compared to those in 9. Among them, the oxymethine proton (δ_H 4.030) shifted to lower frequency was assigned to H-2 of terminal epicatechin unit based on COSY and HMBC correlations with H-3 and H₂-4. Besides, H-2′/6′ of the terminal unit were correlated to II-H-2 in the ROESY spectrum. Thus, it evidenced the substructure of the main chain as “EC=8EC-8EC”. The linkages of branched units were established as “EC-8EC-6-” based on the observation of ROESY correlations between IV-H-4 with I-H-2′, and V-H-4 with IV-H-2′/6′, as well as HMBC cross peaks between IV-H-4 with I-C-5/6/7, and V-H-4 with IV-C-7/8/9. These data composed together determined 10 as epicatechin-(2β→O→7,4β→8)-[epicatechin-(4β→8)-epicatechin-(4β→6)-]-epicatechin-(4β→8)-epicatechin, PACBAR named as EC=8EC(6-EC8-EC)-8EC. This structure was previously characterized as machiphilitannin A, and the NMR and ECD data of 10 were consistent with data reported for this compound.²⁹

Biomechanical and Biochemical Analysis of the Biomodification Properties of PACs 5–10 on Dentin Extracellular Matrix.

Dynamic mechanical analysis (DMA) is used to assess the bulk viscoelastic behavior of the dentin extracellular matrix (ECM), expressed on their elastic and viscous properties.³⁰ By calculating the complex modulus (E*) and tan δ, we are able to determine the material’s overall resistance to deformation and damping capacity, respectively. Overall, the E* of all biomodified dentin exhibited a 10- to 25-fold increase. There was no significant difference between tetramers (5–8) and pentamers (9–10), except for 5 (p = 0.003), which showed a 25-fold increase in E* compared to untreated dentin (control) (Figure 6). PAC 5 showed the highest E* reported (228.6 MPa), while the control showed the lowest value (9.8 MPa) as expected. Despite high fold increase, the damping capacity of dentin-treated with 5 was similar to 6–10 (p > 0.173). These results are consistent with our previous findings that oligomeric PACs, especially those with DP between 3–5, show the best dentin biomodification activity.^10,17,21,25

Figure 6.

Dynamic mechanical analysis (DMA) of PAC-modified dentin matrices. Data are shown as mean and standard deviation (n = 3 for each compound) for (a) Complex modulus (E*, in MPa), (b) fold variation of E* and (c) damping capacity (tan δ) after dentin biomodification with PACs 5–10. ANOVA/Tukey post-hoc test (α = 0.05) revealed significant differences in all variables (E* - p < 0.003, Fold variation of E* - p = 0.026, damping capacity - p = 0.039). Different letters represent significant statistical differences among groups (p < 0.05).

All tetrameric PACs had mixed AB-type ILFs, with 5 showing the highest biomechanical activity and bearing ABB-type IFLs. The main difference between 7 and 8, both with AAB-type IFLs, was that 8 bears a branched unit that is linked via a (2→O→5, 4→6) A-type IFL. However, this variation in spatial structure did not influence the dentin biomodification potential. Also, the pentamers 9 and 10, bearing a (4→6) IFL and branched unit showed similar viscoelastic behavior of dentin, regardless of the number of units attached in the branched linkage.

Fourier-transform infrared (FTIR) spectroscopy is a powerful analytical method for the biochemical/structural characterization of most types of collagens, as well as collagen-based materials.^31,32 Figure 7 shows the FTIR spectra of unmodified and PAC-treated collagen of dentin ECM. The presence of new peaks and the shifts or broadening of absorbance bands between 1160 ~ 1000 cm⁻¹ in the spectra of PAC-treated dentin indicated structural modification of dentin collagen. Moreover, the biomodified dentin matrix exhibited alterations to the intensities of collagen amide II and III bands. These modifications were quantified by the ratios of amide II and amide III to CH₂ band respectively (Figure 8), which are considered a measure of collagen integrity.²¹ Amide II/CH₂ and amide III/CH₂ ratios decreased in both tetramers and pentamers treated dentin matrix, compared to the control (p < 0.001). No differences existed between the treatments regarding amide III/CH₂ ratio, while one change was observed in amide II/CH₂ where treatment with tetramer 6 was only similar to treatment with pentamer 10.

Figure 7.

Biochemical characterization of type I collagen in the dentin matrix using FTIR spectroscopy. The dashed vertical lines indicate peaks assignment for the collagen amide I at 1630 cm⁻¹, amide II at 1550 cm⁻¹, CH₂ band at 1450 cm⁻¹, and amide III at 1240 cm⁻¹, and exclusive peaks in PAC-treated samples (~1150, 1120, 1070 and 1010 cm⁻¹). CTRL: control.

Figure 8.

Semi-quantitative analysis of FTIR spectra of the dentin matrix. The graphs depict means and standard deviation bars (n = 3) of (a) amide II and (b) amide III ratios of untreated (CTRL) and biomodified collagen matrices with PACs 5–10. Ratios were calculated from the intensities and areas under the assigned peaks of amide II vs. CH₂ (1,550/1,450 cm^–1) and amide III vs. CH₂ scissoring (1,240/1,450 cm^–1). ANOVA/Tukey’s post-hoc test (α = 0.05) showed significant differences among biomodified and control groups (p < 0.001). Different letters represent significant statistical differences among groups (p < 0.05). CTRL : Control.

The amide II band represents the secondary structure of collagen (N-H bending vibrations coupled with C-N stretching vibration).³¹ The decrease of amide II/CH₂ (A₁₅₅₀/A₁₄₅₀) ratios observed in the PAC-treated dentin matrix suggested the formation of PAC-collagen cross-linking (N-H hydrogen bonding), and alteration of its secondary structure. The ratios of amide III/CH₂ (A₁₂₄₀/A₁₄₅₀) of all the PAC-treatments were lower than control, which values were close to 1, revealed that PACs induced conformational changes to the dentin matrix, while the dentin collagen did not denature during treatments and exhibited intact triple helical structure.³¹ Moreover, the occurrence of additional IR absorbances (Figure 7) points to characteristic functional groups of PACs (PACs IR spectra, Figure S65), the absorptions around 1100~1070 and 1010 cm⁻¹ were assigned to the C-O stretching of the pyran ring (C-ring),³³ and bands between 1160 and 1150 cm⁻¹ were attributed to the symmetric stretching of C-O-C.²¹

Summary and Outlook.

The native complexity of oligomeric PACs pose unique analytical challenges in structural determination. It is then arbitrary to identify a known structure solely based on comparing with the published NMR data, for the variances of chemical shifts between isomers esp. stereoisomers were subtle. Therefore, it is imperative to employ integrated methods to consolidate the accurate chemical structure of PAC using the techniques we gathered so far in this “toolbox”, including comprehensive spectroscopic analysis (MS and NMR) to solve the 2D structure and relative configuration, ECD spectroscopy and depolymerization to conclude the absolute configuration, which were fully implemented in this work.

The scale-up preparation method allowed the extensive dentin bioassays and SAR studies on a series of tetrameric and pentameric PACs isolated from C. verum. Among them, the tetramer 5 possessing the “ABB” type IFLs was the most potent PAC with dentin biomodification ability identified so far. Furthermore, PACs with “4→6” versus “4→8” type interflavan linkages showed no significant difference regarding their dentin biomodification ability, which suggested the differential linkage sites had limited influence on the PAC-dentin collagen interactions. These results lead our future interests toward different types of interflavan linkages (A-type vs B-type), and disparate stereochemistry of oligomeric PACs, in order to further explore the SAR of PACs with dentin biomodification abilities.

EXPERIMENTAL SECTION

General Experimental Procedures.

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. 1D and 2D NMR spectra were acquired at 298K on Bruker 600 MHz AVANCE III spectrometer equipped with a 5 mm DCH CryoProbe, and Bruker 800 MHz AVANCE II spectrometer equipped with a TXI RT probe at 278K. HiFSA calculations were carried out using Cosmic Truth (CT) platform @ctb.nmrsolutions.fi. High-resolution electrospray ionization (ESI) mass spectra were determined by a Bruker Impact II mass spectrometer, quadrupole time of flight (qTOF). Column chromatography (CC) was performed on DIAION™ HP-20 (Supelco, Bellefonte, PA, USA), Sephadex LH-20 gel (Pharmacia, Uppsala, Sweden), and TOYOPEARL HW-40F (TOSOH Bioscience LLC, PA, USA). Preparative partition chromatography was performed on a centrifugal partition chromatography (CPC) extractor, SCPE-250 (Gilson Inc., Middleton, WI, USA). 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. 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 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).

Plant Materials.

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).

Diol Phase HPLC Profiling and UHPLC-MS analysis.

Sample for Diol phase HPLC analysis was prepared as 20 mg/mL, by directly dissolving the 70% acetone crude extract in MeOH. ACN/HOAc (98:2) was solvent A, and MeOH/H₂O/HOAc (95:3:2) was solvent B. HPLC analysis was performed according to a method previously reported³⁴ with optimization of the analysis time (0–5 min, 7% solvent B; 5–30 min, 7 to 30% solvent B; 30–32 min, 30 to 100% solvent B; 32–40 min, 100% solvent B), using Develosil™ HPLC 100 Diol (5 μm, 4.6 × 250 mm) column. The PDA chromatograms were extracted at 280 nm. Sample for UHPLC-MS analysis was prepared as 1 mg/mL. The separation was performed on the CORTECS^® C18 (2.7 μm, 3.0 × 100 mm) column in gradient elution of ACN/H₂O/FA from (5:95:0.01) to (30:70:0.01) in 8 min.

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 evaporating 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 HP-20 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 before¹⁰ with a modified solvent system (Hexane/EtOAc/MeOAc/H₂O, 0.4/4/1/4, v/v), the stationary retention volume ratio (Sf) was 0.66. The CPC fractions were pooled into 3 subfractions A-C according to their TLC profiles. Subfraction B (3.56 g) was separated by Sephadex LH-20 CC (MeOH), affording three subfractions B1-B3. Subfraction B1 (3.00 g) contains the major trimer cinnamtannin B1 (4, 2.80 g), which was further purified by repeating Sephadex LH20 CC (MeOH). The remaining fraction (240 mg) from B1 was further separated by TOYOPEARL HW-40F CC (90% MeOH in water) to yield subfraction B1–1, which was finally purified by semi-prep HPLC to give tetramer 7 (5.7 mg). Subfraction C (3.6 g) was subjected to reverse phase C₁₈ CC with gradient elution of MeOH in 0.1% formic acid water from 5% to 30%, combining the similar fractions into subfraction C1 to C4 by TLC. Subfraction C2 (10~15% MeOH, 500 mg) was further fractionated by semi-prep HPLC (ACN/H₂O/FA, 15:85:0.1, 3 mL/min) and yielded seven subfractions C2–1 to C2–7. PAC 8 (15 mg) was purified from subfraction C2–4 (68.5 mg) by semi-prep HPLC (ACN/H₂O/FA, 15:85:0.1, 3.5 mL/min). Subfraction C3 (20% MeOH, 500 mg) was separated by semi-prep HPLC (ACN/H₂O/FA, 14:86:0.1, 3.5 mL/min) to obtain PACs 5 (28.7 mg) and 6 (15.8 mg). Pentamers 9 (12.2 mg) and 10 (3.5 mg) were isolated from subfraction C4 (20% MeOH, 300 mg) by semi-prep HPLC in an isocratic elution of MeCN/H₂O/FA (20:80:0.1, 3 mL/min). The purities (%, w/w) of 5–10 were calculated as 86.62%, 85.86%, 98.29%, 96.26%, 94.88%, and 85.65%, using 100% qHNMR method (Figures S59–S64).³⁵

(2R,3R,4S)-Epicatechin-(4ꞵ⟶2′′)-menthofuran; EC-2″MF (2):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 238 (12.02), 274.5 (−2.65) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₂₅H₂₇O₇ 439.1751; Found 439.1757. ¹H and ¹³C NMR data see Table 1.

Epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-epicatechin-(4ꞵ⟶2′′)-menthofuran; EC=8EC-2″MF (3):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 231.5 (35.61), 272.5 (−5.20) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₄₀H₃₇O₁₃ 725.2229; Found 725.2246. ¹H and ¹³C NMR data see Table 1.

Epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-epicatechin-(4ꞵ⟶8)-epicatechin-(4ꞵ⟶8)-epicatechin; aesculitannin E; EC=8EC-8EC-8EC (5):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 229 (81.98), 274 (−5.84) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₆₀H₄₉O₂₄ 1153.2608; Found 1153.2616. ¹H and ¹³C NMR data see Table 2 and 3.

Epicatechin-(4ꞵ⟶6)-epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-epicatechin-(4ꞵ⟶8)-epicatechin; pavatannin C-1; EC-6EC=8EC-8EC (6):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 232.5 (60.64), 274 (−3.13) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₆₀H₄₉O₂₄ 1153.2608; Found 1153.2598. ¹H and ¹³C NMR data see Table 2 and 3.

Epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-epicatechin-(4ꞵ⟶8)-epicatechin; EC=8EC=8EC-8EC (7):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 230.5 (226.88), 273.5 (−19.96) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₆₀H₄₇O₂₄ 1151.2452; Found 1151.2484. ¹H and ¹³C NMR data see Table 2 and 3.

Epicatechin-(2ꞵ⟶O⟶7,4ꞵ⟶8)-[epicatechin-(2ꞵ⟶O⟶5,4ꞵ⟶6)]-epicatechin-(4ꞵ⟶8)-epicatechin; Parameritannin A-2; EC=8EC(6=EC)-8EC (8):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 225.5 (20.27), 240.5 (33.97), 273 (−2.24) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₆₀H₄₇O₂₄ 1151.2452; Found 1151.2449. ¹H and ¹³C NMR data see Table S2.

Epicatechin-(2β⟶O⟶7,4β⟶8)-[epicatechin-(4β⟶6)]-epicatechin-(4β⟶8)-epicatechin-(4β⟶8)-epicatechin; EC=8EC(6-EC)-8EC-8EC (9):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 235.5 (146.14), 276.5 (−6.83) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₇₅H₆₁O₃₀ 1441.3242; Found 1441.3288. ¹H and ¹³C NMR data see Table 2 and 3.

Epicatechin-(2β⟶O⟶7,4β⟶8)-[epicatechin-(4β⟶8)-epicatechin-(4β⟶6)]-epicatechin-(4β⟶8)-epicatechin; Machiphilitannin A; EC=8EC(6-EC8-EC)-8EC (10):

Brownish amorphous powder; ECD (MeOH) λ_max (Δɛ) 238.5 (33.46), 279 (−3.71) nm; HRMS (ESI) m/z [M + H]⁺ Calcd for C₇₅H₆₁O₃₀ 1441.3242; Found 1441.3248. ¹H and ¹³C NMR data see Table S2.

Furanolysis.

To further confirm the configurations of isolated PACs, furanolysis was performed.¹⁶ Procyanidin B2 (11, 10 mM, 15 mg) and cinnamtannin B1 (4, 30 mg) were each reacted with 20 mM menthofuran in 0.1 N HCl in MeOH at 40 °C for one hour. Adding 5 times the volume of 40 mM sodium acetate solution to the mixture quenched the reaction. The mixture was then extracted twice with ethyl acetate, the organic layer dried, and subjected to semi-prep HPLC for purification. This resulted in the isolation of 1 (6.4 mg), 2 (5.0 mg), and 3 (15.3 mg). Similarly, PACs 5 (1.8 mg), 6 (2.0 mg), and 7 (1.2 mg), were prepared as 10 mM solutions in MeOH, reacted with 40 mM menthofuran; 10 mM PAC 9 (0.9 mg) was reacted with 60 mM menthofuran. All reactions were performed in 0.1 N HCl in MeOH solution at 40 °C for one hour. Each reaction mixture was neutralized by 40 mM sodium acetate and extracted with ethyl acetate. The reaction products were identified by LCMS and chiral phase HPLC analysis. Chiral HPLC analysis was performed using CHIRALPAK^Ⓡ IA (5 μm, 4.6 mm × 250 mm), eluting with an isocratic solvent system composed of hexane/isopropanol/TFA (65:35:0.1) at 1.0 mL/min.

Dynamic Mechanical Analysis (DMA) of Dentin.

A DMA protocol was used to determine the viscoelastic properties of the dentin ECM.³⁰ Briefly, dentin specimens (0.5 mm thickness x 1.7 mm width x 7 mm length, n = 3) were demineralized (10% phosphoric acid, 5h) and treated with PAC 5–10 (1% w/v in 20 mM HEPES) for 1 h. After rinsing, specimens were subjected to strain sweep using a 3-point bending submersion clamp (Q800 DMA, TA Instruments, New Castle, DE, USA) following the preset parameters: frequency of 1 Hz, amplitude sweep from 1 to 100 μm and at a preload force of 0.01 N. DMA measurements were carried out before and after treatment. The viscoelastic properties of dentin were determined by the complex modulus (E*) and tan δ. The E* is an overall representation of the elastic (storage modulus, E′) and viscous (loss modulus, E″) properties of a material, expressed by the formula E* = (E′² + E″²)^0.5, whereas the tan δ (E″/E′) represents the damping capacity and is calculated as the ratio of the viscous and elastic components.³⁶ The fold variation of E* was calculated using before and after treatment mean values. An initial screening of all demineralized dentin specimens included only the ones in the range of 6 – 11 MPa of E* baseline values. The equality of variances intra-groups was determined by Levene’ test and found to meet the assumption of homogeneous distribution for E* fold variation (p = 0.139) and tan δ (p = 0.249). Data were statistically analyzed using one-way ANOVA and Tukey post-hoc test (α = 0.05, SPSS v.25, SPSS) to assess the effects of the DP and specific compound/structure on the mechanical properties of dentin ECM [E* fold variation (FV) and damping capacity (tan δ)].

Biochemical Analysis of the Dentin Extracellular Matrix.

The characterization of the functional groups of the dentin extracellular matrix (ECM) was carried-out by Fourier-transform infrared spectroscopy (FTIR, Nicolet 6700, Thermo Fisher Scientific), equipped with an attenuated total reflectance (ATR) apparatus (single reflection diamond, Thermo Fisher Scientific). Spectra were collected at the range of 4,500 to 600 cm⁻¹ and with the accumulation of 128 scans per sample (n = 3). Background transmission spectra were collected before each analysis without the sample on the ATR crystal. The dentin specimens were pressed on the ATR plate surface and spectra were obtained. The spectra were treated by a 14-point interpolated baseline subtraction, and normalized based on the mean values of all data sets. Assessment focused on the predominant bands of type I collagen: amide I (C=O stretching mode at 1,630 cm⁻¹), amide II (N-H bending and C-N stretching modes at 1,550 cm⁻¹), amide III (C-N stretching modes and N-H bending at 1,240 cm⁻¹), and wagging vibrations of CH₂ groups at 1,450 cm^-1. The IR spectra of compounds 5–10 were collected with the same equipment parameters by dissolving them in MeOH (0.1~1 mg/mL) and applying one drop onto the ATR plate surface; after solvent evaporation, the spectra were collected, with the MeOH residue serving as blank control. The band areas and intensities of the assigned bands were fitted (peak fitting function in Origin), calculated, and the structural variations in the collagen cross-linking and collagen triple helices were investigated using indices calculated as ratios between each amide band (II/III) and the CH₂ scissoring band. FTIR indices (amide II/CH₂ and amide III/CH₂) of the native and modified dentin matrices were statistically analyzed to assess the effects of the DP and specific compound/structure using one-way ANOVA and Tukey’s post-hoc test (α = 0.05, SPSS v.25, SPSS). Spectra were analyzed with OMNIC Spectra (Thermo Fisher Scientific) and Origin Pro 8 (OriginLab, Northampton, MA) software.