The Amazon rain forest plant Uncaria tomentosa (cat’s claw) and its specific proanthocyanidin constituents are potent inhibitors and reducers of both brain plaques and tangles

Brain aging and Alzheimer’s disease both demonstrate the accumulation of beta-amyloid protein containing “plaques” and tau protein containing “tangles” that contribute to accelerated memory loss and cognitive decline. In the present investigation we identified a specific plant extract and its constituents as a potential alternative natural solution for preventing and reducing both brain “plaques and tangles”. PTI-00703 cat’s claw (Uncaria tomentosa from a specific Peruvian source), a specific and natural plant extract from the Amazon rain forest, was identified as a potent inhibitor and reducer of both beta-amyloid fibrils (the main component of “plaques”) and tau protein paired helical filaments/fibrils (the main component of “tangles”). PTI-00703 cat’s claw demonstrated both the ability to prevent formation/aggregation and disaggregate preformed Aβ fibrils (1–42 and 1–40) and tau protein tangles/filaments. The disaggregation/dissolution of Aβ fibrils occurred nearly instantly when PTI-00703 cat’s claw and Aβ fibrils were mixed together as shown by a variety of methods including Thioflavin T fluorometry, Congo red staining, Thioflavin S fluorescence and electron microscopy. Sophisticated structural elucidation studies identified the major fractions and specific constituents within PTI-00703 cat’s claw responsible for both the observed “plaque” and “tangle” inhibitory and reducing activity. Specific proanthocyanidins (i.e. epicatechin dimers and variants thereof) are newly identified polyphenolic components within Uncaria tomentosa that possess both “plaque and tangle” reducing and inhibitory activity. One major identified specific polyphenol within PTI-00703 cat’s claw was epicatechin-4β-8-epicatechin (i.e. an epicatechin dimer known as proanthocyanidin B2) that markedly reduced brain plaque load and improved short-term memory in younger and older APP “plaque-producing” (TASD-41) transgenic mice (bearing London and Swedish mutations). Proanthocyanidin B2 was also a potent inhibitor of brain inflammation as shown by reduction in astrocytosis and gliosis in TASD-41 transgenic mice. Blood-brain-barrier studies in Sprague-Dawley rats and CD-1 mice indicated that the major components of PTI-00703 cat’s claw crossed the blood-brain-barrier and entered the brain parenchyma within 2 minutes of being in the blood. The discovery of a natural plant extract from the Amazon rain forest plant (i.e. Uncaria tomentosa or cat’s claw) as both a potent “plaque and tangle” inhibitor and disaggregator is postulated to represent a potential breakthrough for the natural treatment of both normal brain aging and Alzheimer’s disease.


SUPPLEMENTARY METHODS AND RESULTS: a) PTI-777 fraction f identified as chlorogenic acid
Fraction F was the first material to be purified in quantity sufficient for structural elucidation work. Mass spectroscopy and NMR spectroscopy analysis was initially employed. Several different types of mass spectra ([(chemical ionization (CI), fast atom bombardment (FAB), and electron impact (EI)] were taken of the purified samples.
The 1 H-NMR (500 MHz) of fraction f in pyridine (d 5 ) showed 12 signals (Fig. S1a). A broad signal at about 8.4 ppm was attributed to OH groups on the compound. Two strongly coupled signals at 8.0 and 6.8 ppm and three aromatic signals at 7.5, 7.16 and 7.07 ppm were present in the spectrum indicative of hydrogen bound to sp 2 hybridized carbon atoms. There were also signals at 6.2, 4.75, 4.3, 2.9, 2.72 and 2.7 ppm. All of these signals with the exception of ones at 8.4 and 2.72ppm showed integration consistent with one proton. The integration of the 2.72 ppm signal was closer to 2 protons and the OH signal at 8.4ppm was not integrated. The Correlation Spectroscopy (COSY) spectrum (a two dimensional NMR experiment identifying adjacent protons) showed that the protons responsible for the signals at 8.0 and 6.8ppm were adjacent to each other. The large coupling constant of these two signals as well as their chemical shifts were indicative that these protons were attached to a carbon-carbon double bond system with trans geometry coupled to an aromatic ring. The COSY spectrum also revealed that the three aromatic protons were all on the same benzene ring and that the remaining six signals all showed connectivity indicating five contiguous carbon atoms.
The 13 C NMR (500 MHz) in pyridine (d5) showed 16 discrete signal regions (Fig. S1b). The two signals at 177 and 167 ppm indicated the presence of two carbonyl C atoms. There were 8 carbons found in the shift range for sp 2 carbons (signals at 150, 147, 146, 127, 122, 117, 116 and 115 ppm). There were four signals (76, 74, 72 and 71 ppm) indicative of sp 3 carbons bonded to oxygen atoms. Three of these signals were doubled (74, 72 and 71 ppm). The two signals at 39.7 and 39.2 ppm were representative of sp 3 carbon atoms not bonded to oxygen. The six doubled signals were assumed to be due to a mixture of isomers. The Distortionless Enhancement by Polarization Transfer (DEPT) experiment (distinguishes between carbon atoms bonded to 1, 2, 3 or no H atoms) indicated that the aromatic carbon atoms showed signals at 150, 147 and 127 ppm were not bonded to H. This was also the case for the atom responsible for the signal at 76 ppm. The Heteronuclear Correlation Spectroscopy (HETCOR) experiment helped confirm which 1 H signals were associated with individual 13 C signals.
The diode array ultraviolet (UV) spectrum of fraction f showed a peak at 33nm indicating a possible presence of an aromatic ring with extended conjugation. Analysis of both the 1 H and 13 C NMR data pointed to the presence of a tri-substituted aromatic ring with two phenolic groups and a conjugated ethylene group with trans geometry. The chemical shift of the carbonyl carbon at 177 ppm, suggested that this signal was due to a carboxylic acid group attached to this conjugated system. The COSY data suggested a chain of five continguous carbon atoms, three of which were oxygenated. Attachment of the remaining sp 3 quarternary carbon atom (76 ppm) to both ends of the five-carbon atom chain would form a cyclohexane ring, which would be consistent with the data. The remaining carbonyl carbon atom was assigned as a carboxylic acid group attached to the cyclohexane ring. The conjugated aromatic and the cyclohexane portions of the molecule were connected via an ester linkage, again consistent with the chemical shift data.
The structural features of fraction f were consistent with the compound, chlorogenic acid (C16H18O9, MW 354.31) (Fig. 11a). The structural assignment of fraction f as chlorogenic acid was confirmed by comparison of both the 1 H and 13 C NMR spectra of fraction f with the published spectra of chlorogenic acid [("Aldrich Library of 13 C and 1 H FT NMR Spectra" I (2) 1235C)]. The spectra were identical when solvent dependent chemical shift changes were taken into account. The published UV spectrum of chlorogenic acid was also compared with that of fraction f and found identical. The study indicated that the major compound within fraction f was identified as chlorogenic acid (Fig. 11a). b) PTI-777 fraction j identified as epicatechin Mass Spectroscopy: Numerous attempts to obtain a reliable molecule ion peak of the compound using both the fast atom bombardment (FAB+) and chemical ionization (CI) techniques were unsuccessful. A reliable molecule ion peak (M+1) with m/z 291.05 was however, obtained using electrospray techniques using both time-of-flight (Fig. S2a) and fourier transform mass spectroscopy (Fig. S2b). This mass to charge (m/z) ratio of 290 is consistent with a possible molecular formula of C14H12O7 or C15H14O6. An electron impact (EI) initiated mass spectrum showed large m/z fragments at 123 (C7H7O2), 140 (C7H7O2), and 153 (C9H8O3) (Fig. S2c).
Pentaacetate derivative of compound j: Mass spectra taken in both the FAB+ (Fig. S2d) and electron impact (EI) modes (Fig. S2e) gave molecular ion peaks of 523 and 500 respectively correlating with a sodiated and non-sodiated pentaacetate derivative of a compound with molecular weight 290. High resolution spectra taken in these modes gave observed m/z ratios of 523, 1214 (C25H24O11Na, error -0.4ppm/-0.2mmu) and 500.1317 (C25H24O11, error -0.2 ppm/-0.1 mmu). This information firmly established the molecular formula of the pentaacetate derivative of PTI-777 compound j as C25H24O11 and hence compound j to have the molecular formula corresponding to C15H14O6.
The sample remained stable in the acidified D2O when stored over the course of several months at room temperature, but it began to degrade in acetone within 24 hours.
Correlation spectroscopy (COSY) Correlation spectroscopy (COSY) spectra of the sample in both d6 acetone (Fig. S2j) and acidified D2O ( Fig. S2k-S2m) revealing coupling between the 3 signals from 6.8 to 7.1 ppm and the signals at 2.8 and 4.2 ppm.
Pentaacetate derivative of PTI-777 compound j The expected methyl groups and carbonyl carbon atoms were present in both the proton and carbon spectra of the acetylated derivative of j. The most striking changes in the 1 H NMR spectrum (CDCl3) upon acetylation (Fig. S2n) were the downfield shifts at about 0.6 ppm of the two singlets at 5.9 and 6.0 ppm and the downfield shift of 1.2 ppm of the apparent singlet at 4.85 ppm (compare to Fig. S2h). The 13 C spectrum (Fig. S2o) was less affected, but the aromatic carbon atoms were, in general, shifted downfield upon acetylation. All 15 carbon signals associated with compound j were readily seen in the spectrum. In addition to the 1 H-1 H coupling detected in the COSY spectrum of the unacetylated compound (Figs. S2j, S2k-S2m), the derivative (Fig. S2p) showed a correlation between the apparent singlet at 4.85 ppm and the sharp multiplet at 4.2 ppm of the original compound. The heteronuclear correlation spectroscopy (HETCOR) spectrum (Figs. S2q-S2s) of acetylated derivative was consistent with expectation and confirmed and the assignment and the identity of the protonated carbon atoms. Furthermore, these data are consistent with those reported by Huo et al 93 .
Ultra Violet (UV) Spectroscopy The UV spectrum of PTI-777 compound j showed a maximum at 278 nm consistent with an aromatic phenolic compound (Fig. S2t).

Structural Assignment
The 13 C NMR spectrum showed the presence of three sp 3 type carbon atoms, 9 sp 2 carbon atoms (no carbonyl carbons) and 3 carbon atoms that were either very upfield sp 2 type or very downfield sp 3 type carbon atoms. The 1 H NMR spectrum indicated the presence of 4 phenolic hydroxyl groups and one nonaromatic hydroxyl group. The major fragments observed in the EI mass spectrum corresponded to dihydroxylated benzene rings without and with carbon group substitution (m/z of 123, 139 and 152). The proton NMR data (6.8-7.02 ppm) showed evidence for a tri-substituted benzene ring that was hydroxylated. Comparison of 1 H NMR spectra data of model compounds with that of compound j, showed the observed splitting pattern and chemical shifts of the signals were consistent with a 1substituted, 3,4-dihydroxy benzene structure.
The COSY spectra showed 1 H-1 H coupling in a contiguous three-carbon fragment. The relative chemical shifts of two of these carbon atoms indicated that they were directly bonded to oxygen atoms. The large chemical shift change observed for the signal at 4.8 ppm upon acetylated indicated the location of the nonaromatic hydroxyl group and the chemical shift of the third carbon atom implied that it was benzylic. Together, these data were consistent with the following structure, Ar-CH2-CH(OH)-CH(R)O.
The above fragments accounted for all but 4 hydrogen atoms (two of which are found in phenolic type OH groups) and for all but 6 carbon atoms (3 sp 2 type and 3 intermediates between sp 3 and sp 2 type). The C6H4 formula indicated a high degree of unsaturation consistent with a second aromatic ring. The remaining accounted for singlets in the 1 H NMR at 5.9 and 6.0 ppm revealed that this ring was tetra substituted and electron rich. These data indicated that this benzene ring was bonded to three oxygen atoms (two hydroxyl groups and one ether) and once carbon atom. These three structural units (two phenolic rings and the three carbon fragment) when connected together form a flavanol structure identical to the diastereomers, catechin and epicatechin. Comparison of the 1 H and 13 C NMR spectra, as well as the infrared (IR) spectra of compound j with the published spectra (Aldrich collection) for catechin and epicatechin (Figs. S2u, S2v and S2w) establishing the identity of PTI-777 compound j as epicatechin. The splitting pattern of doublet of doublets center on 2.8 ppm in the 1 H NMR spectrum matches that of the epicatechin reference spectrum. The pattern of the aromatic signals between 6.8 and 7.02ppm is also most similar to that of epicatechin. The IR spectrum of compound j matches closely with that of epicatechin, while the IR spectrum of catechin differs significantly in the fingerprint region.

c) PTI-777 fraction h2 identified as epicatechin-4β-8-epicatechin (proanthocyanidin B2)
Acetylation of h2 A sample of h2 (7 mg) was dissolved in a mixture of acetic anhydride (0.5 ml) and pyridine (0.5 ml). The mixture stood at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the h2 peracetate (6 mg) as a colorless gum.
A sample of a fraction rich in peaks h1 and h2, from a second silica gel column of PTI-777 (50 mg) was dissolved in a mixture of acetic anhydride (0.5 ml) and pyridine (0.5 ml). The mixture stood at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the h2 peracetate (28 mg) as a colorless gum.
Identification of peak h2 from PTI-777 as epicatechin-4β-8-epicatechin (proanthocyanidin B2) The main component of peak h, called h2, from the extract PTI-777, was isolated by a series of chromatographic techniques, monitored by HPLC. We initially separated the original extract PTI-777 by column chromatography over silica gel, when 20% methanol in chloroform gave a fraction rich in the two components of peak h on a preparative scale, to give us mostly pure h1(16 mg) and pure h2 (23 mg). Ave ion electrospray mass spectrum of peak h2 gave a clean 100% ion at 577 Da. This is appropriate for the molecular ion (M-H + )of a molecular formula of C30H26O12, such as a dimer of two epicatechin, or isomeric units. A 1 H NMR spectrum (Fig. S3a) of peak h2 showed unusual broadening of the signals, whilst the 13 C NMR (Fig. S3b) showed sharp and broad signals, consistent with some kind of flavonol dimer. We were surprised to see no signals in the 5.8 -6.3 ppm region of the 1 H NMR spectrum, or in the 90-99 ppm region of the 13 C NMR spectrum, where the characteristic H-6/H-8 and C-6/C-8 signals would appear. Running the NMR spectra in deuteroacetone ( Fig. S3c; S3d) instead of deuteromethanol, showed the expected signals to be present, indicating that in deuterated protic solvents, an exchange of these H-6 and H-8 protons for deuterons took place.

Acetylation of h2
Since the compound h2 was unstable under the conditions necessary to prove its structure by NMR spectroscopy, we had to make a stable derivative. Acetylation of a sample of pure h2 gave a peracetate (Fig. S3e), which was purified by column chromatography over silica gel. A larger sample of this peracetate, identical by NMR and thin layer chromatography, was also obtained by silica gel separation of the two main products from acetylation of a fraction rich in h1 and h2.
One and 2D NMR experiments ( Fig. S3f; S3g) on the h2 peracetate showed it to be a decaacetate. Two sets of signals were seen in both the 1 H and 13 C NMR spectra, in a ratio of three to one. These were due to rotational isomers (atropisomers), shown by opposite phase cross peaks in the NOESY spectrum to interconvert in the time frame of the NMR experiment. We solved the structure using the signals of the major atropisomer. The presence of two flavan-3-ol units could be seen from the four 13 C signals for the C-2 and C-3 positions in the 60 -80 region, as well as a signal at 26.65 for the free C-4 position of the lower unit and a signal at 33.99 for the linked C-4 of the upper unit. A CIGAR 1 H -13 C correlation experiment (Figs. S3h; S3i) showed that the two units were connected from the 4(u) position to the 8(l) position, by the correlations from H-4(u) to C-8(l) and C-8a(l). The stereochemistries at C-2 and C-3 of both upper and lower units was shown to be the same as in epicatechin by the similar chemical shifts of the 1 H and 13 C signals for the lower unit, as well as the similar low coupling constants between H-2 and H-3 in both units. The stereochemistry of the linkage was shown to be 4β-8 from the NOE interactions (Figs. S3j; S3k; S3l), in particular the lack of an interaction between H-2(u) and H-4(u), and the presence of an interaction between H-2(u) and H-6'(l), and between H-3(u) and H-6'(l). The structure of the natural product h2 was therefore assigned to be epicatechin-4β-8-epicatechin (Fig. S3m).
Epicatechin-4β-8-epicatechin is also known as procyanidin B2 or proanthocyanidin B2. Our NMR data on PTI-777 h2 matched partial NMR data published on procyanidin B2 [104][105] and our data of the PTI-777 h2 peracetate exactly matches the published data on peracetylated procyanidin B2 106 . The optical rotation of +29.0 o compared to a literature 5 value of +25 o showed the absolute stereochemistry to be the same as found previously.  (Fig. S4a) from a silica gel column were separated by HPLC. The peaks between 14.5 and 16.2 and 16.2 and 19.0 minutes were collected, then freeze dried to give two products, about 80% pure h1, retention time 15.1 minutes (16mg) (Fig. S4b) as a white solid; and pure h2 (23 mg) as a white solid.
Acetylation of h1 protocol A sample of h1 (5 mg) was dissolved in a mixture of acetic anhydride (0.5 ml) and pyridine (0.5 ml). The mixture stood at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the h1 peracetate (2 mg) as a colorless gum. A sample of a fraction rich in peaks h1 and h2, from a second silica gel column of PTI-777 (50 mg) was dissolved in a mixture of acetic anhydride (0.5 ml) and pyridine (0.5 ml). The mixture stood at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the h2 peracetate (38 mg) followed by the h1 peracetate (15 mg) as a colorless gum.

Results:
The minor component of peak h, called h1, of the extract PTI-777 was isolated by a series of chromatographic techniques, monitoring by HPLC. We initially separated the original extract PTI-777 by column chromatography over silica gel, when 20% methanol in chloroform gave a fraction rich in the two components of peak h (134 mg). An HPLC method was developed to separate the two main components of peak h on a preparative scale, to give us a mostly pure h1 (16 mg) and pure h2 (23 mg).
A -ve ion electrospray mass spectrum of h1 gave a 100% ion at 577 daltons. This is approximate for the molecular ion (M-H + ) of a molecular formula of C30H26O12 (molecular weight 578), such as a dimer of two epicatechin, or isomeric units. We had previously isolated and identified epicatechin from the PTI-77 extract (described above).
A 1 H NMR spectrum (Fig. S4c) and 13 C NMR spectrum (Fig. S4d) of peak h1 showed two sets of signals, consistent with some kind of flavonol dimer, with major and minor atropisomers present.
Acetylation of a sample of h1 gave a peracetate (Fig. S4e), which was purified by column chromatography over silica gel. A larger sample of this peracetate, identical by NMR and thin layer chromatography, was also obtained by silica gel separation of the two main products from acetylation of a fraction rich in h1 and h2.
One and 2D NMR experiments (Figs. S4f; S4g) on the h1 peracetate showed it to be a decaacetate. The structure was solved using the signals of the dominant atropisomer. The presence of two flavan-3-ol units could be seen from the four 13 C NMR signals in the 60 -80 ppm region, as well as a signal at 26.56 for the free C-4 position of the lower unit and a signal at 36.72 ppm for the coupled C-4 of the upper unit (Fig. S4g). The positions of the 13 C signals and the small couplings of the 1 H signals of the lower unit were typical of an epicatechin unit, while the positions of the 13 C signals and the much larger couplings of the 1 H signals of the upper unit were typical of a coupled catechin 107 .
A CIGAR 1 H -13 C correlation experiment (Figs. S4h; S4i) showed that the two units were connected from the 4(u) position to the 8(l) position, by the correlations from H-4(u) to C-8(l) and C-8a(l).

e) PTI-777 fraction k2 identified as epicatechin-4β→8-epicatechin-4β→8-epicatechin or proanthocyanidin C1
The major component of peak k, called k2, of the PTI-777 extract was also isolated by a series of chromatographic techniques, monitored by HPLC. We initially separated the original PTI-777 extract by column chromatography over silica gel, when 40% methanol in chloroform gave a fraction rich in the major component of peak k. Preparative HPLC on a fraction rich in k2 (Fig. S5a) gave a pure sample of peak k2 (Fig. S5b). A -ve ion electrospray mass spectroscopy of this showed it to have a molecular ion M + of 866 (Fig. S5c). This is appropriate for a molecular formula of C45H38O18 (molecular weight = 866), such as a trimer of three epicatechin or catechin units. The initial 1 H NMR (Fig. S5d) showed there to be similar broad peaks to that seen in h2, so it was decided to acetylate the compound to definitely identify the structure of k2. A further fraction from the silica gel column that was rich in peak k2 was acetylated as before to enable us to obtain more material for structure elucidation. The peracetate of k2 was purified by column chromatography over silica gel. One (Fig. S5e) and 2D NMR experiments (Fig. S5f) were carried out on the k2 peracetate. Two sets of signals were seen in both the 1 H and 13 C NMR spectra in a ratio of three to one. These were due to rotational isomers (aptropisomers). The structure was determined from signals of the major isomer.
The position of the 13 C NMR signals and the small couplings of the 1 H NMR signals of the lower unit was typical of epicatechin, and the positions of the 13 C signals and the small couplings of the 1 H signals of the other two units were typical of coupled epicatechin 107 . The presence of three flavon-3-ol units could be seen from the six 13 C signals in the 60-80 ppm region, as well as a signal at 26.39 from the free C-4 position of the lower unit and signals at 34.36 and 35.04 for the coupled C-4's of the other units.
A CIGAR 1 H -13 C correlation experiment (Figs. S5g; S5h) showed that the two units were connected from the 4 (upper) position to the 8 (lower) positions, by the correlations form H-4(u) to C-8(m) and C-8a(m), and from H-4(m) to C-8(l) and C-8a(l).
K2 peracetate was therefore determined to be the structure shown in Fig. S5i. Our NMR data on the K2 peracetate shown in Fig. S5i matched partial NMR data published on proanthocyanidin C1 104 but we could not find any 13 C NMR data published on structure k2 (Fig. S5j). The optical rotation of +60.9 o (MeOH) for structure k2 compared to the literature value of +92 o (H2O) showed that it to have the same absolute stereochemistry as that published. K2 was therefore identified as epicatechin-4β→8-epicatechin-4β→8-epicatechin or proanthocyanidin C1 (Fig. S5j). 13  Acetylation of k2 protocol A sample of k2 (5mg) was dissolved in a mixture of acetic anhydride (0.5ml) and pyridine (0.5ml). The mixture was kept at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the k2 peracetate (2 mg) as a colorless gum. A fraction rich in k2 (34 mg) (Fig. S5b) was dissolved in a mixture of acetic anhydride (0.5ml) and pyridine (0.5ml). The mixture was kept at room temperature for 18 hours, then the solvents were removed in vacuo. Purification by column chromatography over silica gel, eluting with 20% ethyl acetate in dichloromethane gave the k2 peracetate (15mg) as a colorless gum.
Isolation of peak k1: Fractions 38 to 42 contained compound k1 (22mg) as a pale brown gum. The retention time of this k1 peak was 15.0 minutes as monitored by HPLC. For acetylation of k1 to determine the structure, a sample of k1 (15 mg) was dissolved in a mixture of acetic anhydride (0.5ml) and pyridine (0.5ml). The mixture stood at room temperature for 18 hours, then the solvents were removed in vacuo to give the k1 peracetate (16mg) as a colorless gum.
Isolation of k1 and the k1 peracetate: The minor component of peak k, called k1, of the PTI-777 extract was isolated by column chromatography over sephadex LH20, monitored by HPLC. Elution with 95% ethanol followed by increasing amounts of acetone and water, followed by methanol, gave pure peak k1 in fractions 38 to 42 . The structure of the k1 peracetate is shown in Fig. S6a, whereas the structure of k1 is shown in Fig. S6b. To arrive at these structures, the following analysis and results were obtained.
A -ve ion spectroscopy mass spectrum of k1 gave a 100% ion at 561 daltons (Fig. S6c). This is appropriate for the molecular ion (M + -H) of a molecular formula of C30H26O11 (molecular weight = 562), such as a mixed dimer of one epicatechin, or isomeric unit and one epiafzelechin, or isomeric unit. The 13 C NMR of k1 showed signals consistent with some kind of flavanol dimer (Fig. S6d). The 1 H NMR spectrum of k1 showed there to be similar broad peaks to that seen in compound h2 (Fig. S6e), so it was decided to acetylate the compound to determine the final structure. Acetylation of pure k1 gave a peracetate (structure shown in Fig. S6a). The 1 H and 13 C spectrum of the k1 peracetate are shown in Fig.  S6f and Fig. S6g, respectively. Two sets of signals were seen in both the 1 H and 13 C NMR spectra, in a ratio of three to one. These were due to rotational isomers (atropisomers). The structure was determined using the signals of the major atropisomer.
The presence of the two flavan-3-ol units could be seen from the four 13 C signals for the C-2 and C-3 position in the 60-80 ppm region, as well as a signal at 26.61 ppm for the free C-4 position of the lower unit and a signal at 34.14 for the linked C-4 of the upper unit. A CIGAR 1 H-13 C correlation experiment (Figs. S6h to S6j) showed that the two units were connected from the 4(µ) position to the 8(l) position between H-4(µ) and C-8(l). The sterochemistries at C-2 and C-3 of both upper and lower units was shown to be the same as in epicatechin by the similar chemical shifts of the 1 H and 13 C signals for the lower unit, as well as the similar low coupling constants between the H-2 and H-3 in both units. The lower flavan-3ol unit was shown to be epicatechin by CIGAR correlations from H-2(l) to C-2' and C-6' signals of the 3'-4'-dioxygenated aromatic ring. The upper flavan-3-ol unit was identified by CIGAR correlations from H-2(µ) to equivalent C-2'/C-6' signals of a 4'-oxygenated ring. This constitutes an epiafelechin unit. The structure of the natural product k1 was therefore assigned to be epiafzelechin-4β→8-epicatechin. This compound is a known compound [109][110] . Our NMR data on the structure for k1 matched partial NMR data published on epiafzelechin4β→8-epicatechin. The optical rotation of -

g) PTI-777 fraction l identified as epicatechin-4β→8-epicatechin-4β→8-epicatechin4β→8-epicatechin4β→8-epicatechin (epicatechin tetramer)
Peak l from the extract PTI-777 was isolated and shown from electrospray mass spectroscopy, degradation studies and by partial NMR studies to be the epicatechin tetramer epicatechin-4β→8epicatechin-4β→8-epicatechin4β→8-epicatechin (Fig. S7). The compound proved to be unstable in solution, and to acetylation conditions, which prevented full characterization in the same manner as that used for the other proanthocyanidins previously isolated from this extract. The epicatechin tetramer has been reported in the literature, and our partial data match that partial data reported.
Results: Peak l of the extract PTI-777 was isolated by a series of chromatographic techniques monitored by HPLC. We separated the original extract of PTI-777 by column chromatography over silica gel, when 50% methanol in chloroform gave a fraction rich in peak l. Preparative HPLC of this fraction, using method 2 gave a pure sample of peak l. A -ve ion electrospray m.s. of a sample of peak l, showed it to have a pseudo molecular ion [M-H] of m/z 1153. The spectrum also showed a strong ion with m/z 576 due to [M-2H] 2-. This is appropriate for a molecular formula of C80H50O24, such as a tetramer of four epicatechin or isomeric units.
1 H and 13 C NMR spectra of peak l showed the typical peaks for an epicatechin oligomer, the peaks in the 60 to 90 ppm region are diagnostic of the flavanol units present, peaks at 67 and 80 ppm were seen for the C-2 and C-3 of a terminal epicatechin, then a group of peaks between 77 and 78 ppm for the -2 of linked epicatechins, and a group of peaks between 72 and 74 for the C-3 of linked epicatechins. The spectra were limited due to the small amount of sample, as well as its instability in solution over time, but the 13 C NMR spectrum did match (allowing for slight differences due to solvent) that of proanthocyanidin B2 ( Fig. S7) reported previously 110 . Attempts to acetylate a newly isolate sample of peak l, gave only the peracetate of the epicatechin dimer (PTI-777 fraction h2), which indicated that peak l is not stable to the acetylation conditions.
The small amount of sample available of a four flavanol unit oligomer, with the additional complication of rotational isomers and its instability in solution, means that the structure of peak l couldn't be elucidated by NMR alone. Indeed, the structure of most oligomers of this size and many of the trimer flavanols have been previously proven by degradation studies. The units are cleaved in ethanolic HCl, then the resultant flavanol cations, formed from all of the linked units, are trapped out by reaction with phloroglucinol, to give their 4-phloroglucinol adducts 111 . The terminal unit is left as the free flavanol. Therefore, this reaction can be used to work out which units are present, which is the terminal unit and the ratio of linked units to terminal units.
HPLC analysis of the reaction products of peak l, with phloroglucinol in ethanolic HCL showed three peaks, two equivalents of epicatechin-4-phloroglucinol, one equivalent of epicatechin and one equivalent of the epicatechin dimer, proanthocyanidin B2 (i.e. Fraction PTI-777 peak h2). The epicatechin dimer, proanthocyanidin B2 is the result of incomplete degradation, but can be used in conjunction with other products to prove the structure. This showed that peak l must be comprised of 4 units of epicatechin. The results are very similar to those obtained from phloroglucinol degradation of the proanthocyanidin previously described 111 . Further evidence that peak l is made of epicatechin units comes from the formation of peak l from a disproportionate reaction of the proanthocyanidin dimer, Proanthocyanidin B2, in acidic methanol, which leads to the formation of epicatechin (PTI-777 peak j), the epicatechin trimer C1 (PTI-777 peak k2), as well as peak l, and post-peak l material. Also, it was noted that one of the products from a disproportion reaction of the proanthocyanidin trimer C1 (PTI-777 peak k2), formed when left in methanol, was the proposed tetramer peak l. The formation of peak l from dimer proanthocyanidin B2 and trimer proanthocyanidin C1, combined with the degradation studies indicate that the tetramer must be made up of 4β→8 linkages, giving us the structure of peak l as epicatechin-4β→8-epicatechin-4β→8-epicatechin4β→8-epicatechin4β→8-epicatechin.