Phytochemistry 73 (2012) 95–105

Contents lists available at SciVerse ScienceDirect

Phytochemistry

journal homepage: www.elsevier .com/locate /phytochem

Analysis of commercial proanthocyanidins. Part 1: The chemical compositionof quebracho (Schinopsis lorentzii and Schinopsis balansae) heartwood extract

Pieter B. Venter, Mirek Sisa, Marthinus J. van der Merwe, Susan L. Bonnet,Jan H. van der Westhuizen ⇑Department of Chemistry, University of the Free State, Nelson Mandela Avenue, Bloemfontein 9301, South Africa

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 April 2011Received in revised form 23 June 2011Available online 5 November 2011

Keywords:Schinopsis lorentzii and Schinopsis balansaeAnacardiaceaeQuebrachoElectrospray mass spectrometryProanthocyanidinsNatural polymer

0031-9422/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.phytochem.2011.10.006

⇑ Corresponding author. Tel.: +27 51 4012782; fax:E-mail address: vdwestjh@ufs.ac.za (J.H. van der W

Quebracho (Schinopsis lorentzii and Schinopsis balansae) extract is an important source of natural poly-mers for leather tanning and adhesive manufacturing. We combined established phyto- and syntheticchemistry perspectives with electrospray mass spectrometry experiments to prove that quebracho pro-anthocyanidin polymers consist of an homologous series of flavan-3-ol based oligomers. The starter unitis always catechin which is angularly bonded to fisetinidol extender units. By comparison of the MS2 frag-mentation spectra of the oligomer with product ion scans of authentic catechin and robinetinidol sam-ples, we proved that quebracho extract contains no robinetinidol, as is often reported. Quebrachoproanthocyanidins have acid resistant interflavanyl bonds, due to the absence of 5-OH groups in fisetin-idol, and the aDP cannot be determined via conventional thiolysis and phloroglucinolysis. We used theMS data to estimate a conservative (minimum value) aDP of 3.1.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The wild quebracho forests in the Gran Chaco region of Argen-tina, Bolivia, and Paraguay have been harvested for more than100 years as an important source of vegetable tannins and timber.The timber is durable and extremely hard and the name quebrachois derived from the Spanish word quiebrahacha which means ‘‘axe-breaker’’. To obtain a warm water soluble quebracho extract, theheartwood is stripped of its bark, chipped, and extracted with boil-ing water. A cold water soluble extract (sulfited extract) is obtainedupon treatment of the warm water soluble extract with bisulfite ordirect extraction of wood chips with a boiling aqueous bisulfitesolution. Higher extraction rates are obtained with boiling aqueousbisulfite solution than with boiling water alone.

Quebracho extract is obtained from Schinopsis balansae (red‘‘chaqueno’’ quebracho, pure tannin content 20–21%) from theEastern Chaco region and Schinopsis lorentzii (red ‘‘santiagueno’’quebracho, pure tannin content 15–18%) from the Western Chacoregion. These two species were previously referred to as Quebrachocolorado chaqueño and Quebracho colorado santiagueño (Schinopsisquebracho-colorado) and belongs to the family Anacardiaceae. Athird tree species, Aspidosperma quebracho-blanco of the familyApocynaceae, is commonly referred to as white quebracho.

ll rights reserved.

+27 51 4448463.esthuizen).

Quebracho extract consists of about 95% proanthocyanidins(PAs) and 5% water soluble sugars on a dry basis. The term pro-anthocyanidin (PA) refers to the characteristic development of ared color upon heating PAs with dilute acid (Roux, 1992). PAs arealso referred to as condensed tannins to distinguish them fromhydrolysable tannins which do not produce a red color whenheated with aqueous acid. Hydrolysable tannin oligomers areesters of gallic acid and D-glucose. Important industrial sources ofPAs are mimosa bark extract (Acacia mearnsii) and quebrachoheartwood extract, and of hydrolysable tannins, tara pods, chest-nut bark, and oak gall extracts.

Progress in defining quebracho PA composition has been slow,mainly due to the complexity of the extracts and the difficulty ofisolating pure PAs with silica gel based chromatography materials.Uncertainties include different hydroxylation patterns of the con-stituent flavan-3-ol aromatic rings, different configurations at theC-2, C-3 and C-4 stereogenic centers, the possibility of a secondether interflavanyl bond (A-type PAs), the average chain length(degree of polymerization), and the presence of angular oligomers.

Progress is further hampered by the absence of 5-OH groups inthe constituent monomers, which imparts stability to the interflav-anyl bond against acid hydrolysis (Roux and Paulus, 1962; Rouxet al., 1975). This renders the classical method to analyse PAs viaacid hydrolysis of the interflavanyl bond and subsequent trappingof intermediates with toluene-a-thiol or phloroglucinol (thiolysisand phloroglucinolysis) (Thompson et al., 1972; Foo and Porter,1978; Kennedy and Taylor, 2003; Rigaud et al., 1991) and analysis

Fig. 1. Flavan-3-ol and flavan-3,4-diol monomers from the heartwood of S. lorentzii(putative building blocks of quebracho PAs).

Fig. 2. Quebracho dimers from S. balansae.

Fig. 3. Trimer isolated from S. balansae [ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol].

Fig. 4. Tetramer synthesized by Viviers and co-workers.

96 P.B. Venter et al. / Phytochemistry 73 (2012) 95–105

of such trapped intermediates with HPLC (Shen et al., 1986;Koupai-Abyazani et al., 1993; Rigaud et al., 1991; Kennedy andTaylor, 2003), unreliable. Vivas et al. (2004), for example, failedto isolate any known flavan-3-ol toluene-a-thiol adducts uponthioacidolysis of quebracho tannins.

Most of the properties and industrial applications of vegetabletannins are attributed to the ability of the constituent PAs orhydrolysable tannins to form complexes with proteins via hydro-gen bonds (Haslam, 1974, 1988, 1997). This includes astringencyin tea and red wine (interactions between tannins and proteinbased taste receptors in the mouth) (Bate-Smith, 1954; Hofmannet al., 2006), anti-feeding properties (the indigestibility of tan-nin–protein complexes) (Hagerman et al., 1992), and growth inhi-bition of many micro-organisms (irreversible deactivation ofenzymes) (Akin, 1982). Complexation of vegetable tannins withhide proteins transform biodegradable raw hide into leather whichresists bacterial degradation, has a nice touch and is abrasion, heat,and water resistant (Haslam, 2005). Quebracho is extensively usedto produce vegetable tanned leather. It is also used to manufactureadhesives via cross linking of the nucleophilic aromatic A-rings ofthe constituent PAs with formaldehyde (Pizzi, 1978). It is a sourceof oenological tannins, used to enhance the ‘‘mouth feel’’ proper-ties of young or poor quality red wines. The absence of the 5-OHgroup and corresponding stability of the PA oligomer to interflav-anyl bond fission (Roux and Paulus, 1962; Roux et al., 1975) isprobably an important factor in the industrial application of que-bracho and mimosa PAs as it imparts longevity to leather andadhesives manufactured from it. A better understanding of themolecular composition of vegetable tannins will assist industrialapplications. The relative affinity for collagen, rate of penetrationinto hides and skins during commercial tannage, mobility withinleather, and desorption from finished leather under moist condi-tions are determined by oligomer composition (Covington, 2009).The availability of nucleophilic centers for cross linking with form-aldehyde on the periphery of oligomers determines curing timeand pot life of thermosetting PA based adhesives.

Electrospray ionization (ESI) and matrix-assisted laser desorp-tion ionization (MALDI) are soft ionization techniques that canfractionate a mixture of oligomers, such as quebracho PA extract,into fractions of different degrees of polymerization (DP) and esti-mate the average degree of polymerization (aDP). Soybean seedcoat extract (Takahata et al., 2001) and hop PAs (Taylor et al.,2003) with a DP of 30 and 22, respectively, have been characterisedby MALDI-TOF MS, and litchi PAs with a DP of 22 (Le Roux et al.,1998) with ESI. Mouls and co-workers (2011) compared aDP valuesobtained from thiolysis of PAs with the aDP values obtained fromESI-MS. They confirmed that poorer ionization of high DP PAs ledto the underestimation of the aDP with MS, but concluded thatESI is appropriate to analyse low molecular weight PA samples(aDP below 20).

Pasch et al. (2001) investigated commercial sulfited quebrachotannin extract using MALDI-TOF mass spectrometry and observedoligomers to a maximum of decamers (2798 Da) (c.f. octamers formimosa PAs). This is in line with the aDP of 6.74 (c.f. 4.9 for mimo-sa PAs) found by Thompson and Pizzi (1995) and Fechtal and Riedl(1993) with NMR methods. The individual PA oligomers consistingof clusters of ions 16 Da apart, was attributed to combinations andpermutations of fisetinidol (274 Da) and robinetinidol (290 Da)constituent units. They concluded that quebracho PAs consistmostly of profisetinidins. The same authors claim that quebrachoPAs were, in contrast with angular mimosa PAs, linear and that thislinear structure explains the relative ease with which quebrachoPAs undergo acid catalysed hydrolysis compared to smaller, lessviscous oligomers.

Table 1ESI (negative mode and positive mode) ions for hot water soluble quebracho extract.

Oligomer m/z Value(negative mode)

m/z Value(positive mode)

Catechin Fisetinidol

Dimer 561 563 1 1Trimer 833 835 1 2Tetramer 1105 1107 1 3Pentamer 1377a 1379d 1 4Hexamer (1649)b 1651e 1 5Heptamer (1921)c (1923)f 1 6

(Ions in brackets were not detected directly but indirectly as water adducts).a The 13C isotope peak at m/z 1378 was automatically annotated in Fig. 5a. The slightly less intensive 12C peak at m/z 1377 is also

visible. Water adducts (+18 Da) of these two peaks are visible at m/z 1395 and 1396.b The m/z 1649 value was indirectly detected as a water adduct of the 13C isotope peak at m/z 1668. Close inspection of a magnified

spectrum reveals the presence of a 12C water adduct at m/z 1667.c The expected heptamer was not detected in negative mode at m/z 1921 in Fig. 5a.d The m/z 1379 peak is also detected as the 13C isotope peak at m/z 1380, and as their water adducts (+18 Da) at m/z 1397 and 1398,

respectively.e The m/z 1651 peak is also detected in Fig. 5c as single and double water adducts at m/z 1669 and 1687, respectively. Magnification of

the spectrum also reveals the corresponding 13C isotope peaks at 1652, 1670, and 1688, respectively.f The heptamer is mainly detected as a double water adduct (+36 Da) at m/z 1959 (and the corresponding 13C isotope peak at 1960).

NOPQ

m/z1700200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1800 1900

%

0

100NP_Pol_100216_23 11 (0.126) Cm (11:14-1:4) TOF MS ES-

374561.1

552.1

287.0238.9 449.1423.0

343.0

495.1

833.1

562.1

601.1

688.1602.1

603.1688.6

689.1 831.1

834.1

1105.2

835.1

849.1

850.1 985.1869.1 1009.2

1106.2

1107.2

1395.21123.21378.2 1668.31396.3

Fig. 5a. Negative mode ESI spectrum of hot water soluble quebracho extracts (m/z 200 to 2000 m/z range).

P.B. Venter et al. / Phytochemistry 73 (2012) 95–105 97

2. Phytochemistry

Roux and Evelyn (1960) found only catechin 1 and ent-fise-tinidol-4b-ol [(�)-leucofisetinidin] 2 (Fig. 1) as monomeric con-stituents in the heartwood of S. lorentzii. This suggests that 1and 2 are the precursors of quebracho PAs. The flavan-3,4-diol2 is present in high concentrations at the sapwood/heartwoodinterface and declines rapidly from the heartwood edge and isabsent from the center heartwood of mature (120–140 yearold) trees. An increase in average molecular weight from 910in the outer heartwood to 1784 Da in the central heartwoodPAs (determined with ebulliometry) suggests that PA oligomer

formation continues away from the sapwood after heartwoodformation.

Viviers and co-workers (1983) isolated the two diastereoiso-mers ent-fisetinidol-(4b ? 8)-catechin 3 and ent-fisetinidol-(4a? 8)-catechin 4 (m/z 562) from S. balansae. Smaller quantities ofent-fisetinidol-(4b ? 6)-catechin and ent-fisetinidol-(4a ? 6)-cat-echin diastereoisomers 5 and 6 were also isolated (Fig. 2). The ratioof 3:4:5:6 approximated 2.5:1:0.6:0.2.

The same team also isolated the angular trimer ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol 7 (4,6;4,8-bis-ent-fisetinidol-catechin) (m/z 834) (Fig. 3) and three diastereo-isomers from S. balansae. However, no tetramers were reported.

NOPQ

m/z

200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900

%

0

100NP_Pol_100216_11 31 (0.339) Cn (Cen,4, 70.00, Ar); Sm (SG, 1x5.00); Sb (1,40.00 ); Cm (23:31-2:4) TOF MS ES+

2.00e4563.1

411.1393.1

301.1287.1273.1

271.1231.1

315.1

365.1

437.1 545.1

485.1

683.2

564.2

565.2

681.2

585.1

835.2

684.2

725.2

726.2

817.2801.2

955.2836.2

857.2

858.2

873.2

874.2

1107.3956.2

997.2998.3

1033.3

1108.3

1129.3

1227.31130.3

1131.3

1148.3

1686.41397.41228.31229.3

1270.3

1414.4 1669.4

1499.4

1692.41959.5

1693.41694.4 1965.5

Fig. 5b. Positive mode ESI spectrum of hot water soluble quebracho extracts (m/z 200 to 2000 m/z range).

NOPQ

m/z1300 1350 1400 1450 1500 1550 1600 1650 1700 1750 1800 1850 1900 1950

%

0

100NP_Pol_100216_11 31 (0.339) Cn (Cen,4, 70.00, Ar); Sm (SG, 1x5.00); Sb (1,40.00 ); Cm (23:31-2:4) TOF MS ES+

1.86e31686.4

1397.4

1379.3

1300.41361.3

1307.3

1329.4

1414.4

1419.3

1669.4

1420.3

1499.4

1421.3

1433.31435.31445.4

1481.41451.3

1500.4 1668.4

1651.41517.4

1518.4

1559.41519.4

1577.41634.4

1687.5

1691.4

1692.4

1959.5

1958.5

1693.4

1694.41943.5

1941.51790.51709.4

1771.41711.4

1924.51923.51862.51839.5

1864.5

1964.5

1965.5

1979.5

1982.5

Fig. 5c. Positive mode ESI spectrum (expansion of 5b) of hot water soluble quebracho extracts (m/z 1300 to 2000 m/z range).

98 P.B. Venter et al. / Phytochemistry 73 (2012) 95–105

3. Synthesis

Viviers and co-workers (1983) investigated the biomimetic syn-thesis of quebracho PAs via acid catalysed condensation of catechin1 and ent-fisetinidol-4b-ol 2. The products closely resemble thoseisolated by the same authors.

Condensation of 1 eq. of catechin 1 with ent-fisetinidol-4b-ol 2(1 eq.) gives mainly ent-fisetinidol-(4b ? 8)-catechin 3 and smallquantities of the epimeric ent-fisetinidol-(4a ? 8)-catechin 4(Fig. 2). The presence of a second equivalent of 2 led to formationof the trimer, ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol 7 (Fig. 3). A further equivalent of 2 leads to the

Fig. 6. Structures of rDA fragments of m/z 563, 835 and 1107 dimers, trimers and tetramers.

Table 2Diagnostic rDA fragments associated with their corresponding oligomer precursors.

Oligomer ESI+ mass Rel. comp.a RDA mass Rel. comp.a

Dimer 563 125 411 36Trimer 835 79 683 110Tetramer 1107 33 955 49Pentamer 1379 8 1227 13Hexamer 1651 3 1499 4Heptamer 1923 61 1771 1.5

a Relative composition is based on peak height.

P.B. Venter et al. / Phytochemistry 73 (2012) 95–105 99

formation of the tetramer, ent-fisetinidol-(4b ? 6)-ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol 8 (Fig. 4).

The most reactive nucleophilic position on catechin 1 is C-8since at this position the highest occupied molecular orbital

-MS2 (561.20): 0.312 to 3.359 min from Sample 1 (TuneSam taeH(ffiw.14103101301102TMfo)DIelp

50 100 150 200 250 300m/z

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Rel

. Int

. (%

)

289.3

161.3

271.3125.1 245.4

109.0205.0 257.0137.3 253.1151.1

31295.0189.3 9.6822.311

Fig. 7a. Product ion scan of the m/z 561 d

(HOMO) exhibits it’s maximum amplitude (Elliot et al., 1982). Thefirst condensation product (dimer) will thus predominantly beent-fisetinidol-(4b ? 8)-catechin 3. The C-6 position of the phloro-glucinol A-ring of catechin (two enolic OH groups, one enolic ethergroup) is more nucleophic than the resorcinol A-ring (one enolic OHand one enolic ether group) of the competing ent-fisetinidol unit.The second condensation product (trimer) will thus predominantlybe ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol 7.

In constructing the tetramers from the trimers (Young et al.,1985), it must be emphasized that both reactive positions of thephloroglucinol A-ring of the catechin moiety are occupied in thetrimer. Thus, the resorcinol A-ring of the upper ent-fisetinidol moi-ety is the most reactive remaining nucleophilic position. The trimerwill thus react via the sterically less hindered C-6 position with athird ent-fisetinidol-4b-ol molecule to yield the tetramer

.spc1.6634.xaM)rezilubeNde

350 400 450 500 550 600, amu

561.3

409.5

391.3

451.33.4 328.5

imer ion (APCI in the negative mode).

.spc8.221.xaM)rezilubeNdetaeH(ffiw.34824101301102TMfo)DIelpmaSenuT(1elpmaSmorfnim102.12ot750.2:)03.338(2SM-

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800m/z, amu

10%

15%

20%

25%

30%

35%

40%

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Rel

. Int

. (%

)

289.5

833.5561.2

529.3

409.4

161.0 600.4391.8

270.8

680.8

451.4 510.85.7736.3021.901

310.9 329.8 359.9999 2222222294.88267.5 41113.9 4 333333.3 4475.5198..8192.21 593.111154.9 635.9 663.3 719.1 824.2222

Fig. 7b. Product ion scan of the m/z 833 trimer (APCI in the negative mode).

100 P.B. Venter et al. / Phytochemistry 73 (2012) 95–105

ent-fisetinidol-(4b ? 6)-ent-fisetinidol-(4b ? 8)-catechin-(6 ? 4b)-ent-fisetinidol 8 (Fig. 4).

Owing to the increased thermodynamic stability of 3,4-transcompared to 3,4-cis isomers (Forest et al., 2004), the isolated andsynthesised oligomers possess predominantly, but not exclusively,3,4-trans configured constituent units. Owing to the fact that massspectrometry cannot distinguish between diastereoisomers orregioisomers, implies that we will not refer in our furtherdiscussion to configuration or position of the interflavanyl linkand replace, e.g., the terms ent-fisetinidol-(4b ? 6)-catechin 5and ent-fisetinidol-(4a ? 6)-catechin 6 with fisetinidol-catechin.

Phytochemistry thus suggests that:

1. Quebracho heartwood contains only catechin and ent-fisetin-idol-4-ol and no robinetinidol-4-ol.

2. Dimers and trimers consist of a catechin starter unit and one ortwo fisetinidol extender units. The trimer is angular with onefisetinidol in the ‘‘upper’’ C-8 position and the other in the‘‘terminal’’ C-6-position. No linear fisetinidol–fisetinidol dimers,fisetinidol–fisetinidol–fisetinidol trimers, or robinetinidol con-taining dimers and trimers were reported in the literature.

Synthetic organic chemistry suggests that:

1. The fisetinidol–catechin–fisetinidol trimer will be the soleintermediate in the construction of all higher oligomers andall higher oligomers will have this moiety attached to one ormore additional fisetinidol extender units.

2. The formation of tetramers and higher oligomers are inhibitedby the lower reactivity of the 5-deoxy fisetinidol A-ring. Wethus expect that dimers and trimers will be the major compo-

nents in quebracho (and mimosa) PAs and higher oligomers willbe relatively less common. This is not the case with 5-oxy PAswhere reactive catechin-4-ol or gallocatechin-4-ol are theextender units and large oligomers (DP of 20 and more) arecommon (Takahata et al., 2001; Taylor et al., 2003; Le Rouxet al., 1998).

We thus postulate that quebracho PA oligomers consist of ahomologous series of flavan-3-ol based oligomers. The starter unitis always catechin which is angularly bonded to fisetinidol exten-der units. Herein, we report our results on the composition of ahot water soluble (unsulfited) quebracho extract from S. lorentziiwith electrospray ionization (ESI) and atmospheric pressurechemical ionization (APCI) mass spectrometry and use these re-sults to test our hypothesis. When comparing our ESI and APCI datawith published MALDI data, it should be taken into account thatMALDI-ionization is observed via sodium [M+23]+ or potassium[M+39]+ adducts. The two MALDI m/z values are 16 Da apart andcan be misinterpreted in PA mass spectra as evidence for the pres-ence of oligomers with additional OH-groups (Reed et al., 2005).

4. Experimental

Spray dried, hot water soluble quebracho extract fromS. lorentzii was supplied by Mimosa Extract Company (Pty) Ltd.,24 van Eck Place, Pietermaritzburg, 3201, South Africa.

HPLC grade (P99.9% purity) methanol and water were pur-chased from Merck. The mass spectrometer was a Sciex API 2000MS/MS system, equipped with an ESI or APCI source and operatedin the negative ion mode. The operating conditions in the ESI

Scheme 1. Fragmentation of m/z 561 quebracho dimer (M�H)�.

P.B. Venter et al. / Phytochemistry 73 (2012) 95–105 101

source were as follows: ionspray voltage, �4500 V; declusteringpotential, �40 V; probe temperature, 450 �C. Nitrogen was usedas the nebulizer gas (20 units) curtain gas (20 units), and the

-MS2 (289.00): 1.150 to 5.635 min from Sample 1 (TuneSample ID) H(ffiw.83804101301102TMfo

40 60 80 100 120 140 160m/z

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Rel

. Int

. (%

)

109.0

123.2

137.5

151.497.3 121.6

95.1161

81.3149.2

57.2139.3 160.135.3

110.892.8

145.2163.2107.869.3 142.980.259.141.1 133.496.2 117.044.9 85.2 154.4

Fig. 8a. Product ion scan of the m

collision gas (5 units). The operating conditions in the APCI sourcewere as follows: nebulizer current, �2.0 lA; probe temperature,450 �C; declustering potential �20 V. Nitrogen was also used as

Max. 2640.7 cps.ated Nebulizer)e

180 200 220 240 260 280 300, amu

203.2

.4

205.0 289.4175.06 245.4187.2 221.1

177.1174.1 201.7 227.5 230.2 247.4217.4185.7270.9 9.7824.1623.171

/z 289 fragment in Fig. 7a.

-MS2 (289.30): 2.695 to 6.316 min from Sample 1 (TuneSampleID) of MT20110310162022.wiff (Turbo Spray) Max. 2.8e4 cps.

40 60 80 100 120 140 160 180 200 220 240 260 280 300

m/z, amu

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Rel

. Int

. (%

)

108.8

124.8

289.4

123.2

137.6 151.4203.3

97.3 245.5205.1

121.494.9 161.5

179.5149.381.4 187.283.3 175.0165.1

221.1139.1 159.157.6 135.3167.2 227.693.0 185.5 201.7 217.6105.4 247.6110.9 133.3 230.2144.6 288.1271.269.3 157.141.5 80.167.355.2 181.343.238.9 257.1

Fig. 8b. Product ion scan of an authentic catechin sample.

102 P.B. Venter et al. / Phytochemistry 73 (2012) 95–105

the nebulizer gas (60 units), curtain gas (40 units), and the collisiongas (5 units). The collision energy used for both the APCI and ESIsource was �30 eV. The chromatograph consisted of an Agilent1200 series auto-sampler, pump and column department. Theinjection solvent consisted of water and methanol (1:1, v/v) witha flow speed of 50 lL/min.

Additional mass spectrometric information was obtained bydirect infusion of a solution of quebracho extract into a WatersAPI Q-TOF Ultima mass spectrometer, using a carrier solution ofacetonitrile:water:formic acid (80:20:0.1, v/v/v) delivered by aWaters Acquity Ultra Performance Liquid Chromatography (UPLC)system at a flowrate of 0.3 lL/min. The operating conditions for theESI source in the negative ion mode were as follows: capillaryvoltage, 3.5 kV; cone voltage, 35 kV; source temperature, 100 �C;desolvation temperature, 350 �C; desolvation gas, 350 L/h; conegas, 50 L/h.

5. Results and discussion

5.1. Q1 scan of hot water soluble quebracho extract

The negative mode ESI mass spectrum of hot water soluble que-bracho extract (Fig. 5a) has salient m/z values at 561.1 and

833.1 Da, and less intense ions at m/z 1105, 1378, and 1668. Theseions correspond with fisetinidol–catechin dimers, fisetinidol–catechin–fisetinidol trimers, and higher oligomers correspondingwith one catechin starter unit and three to five fisetinidol extenderunits. The pentamer, with a small peak at m/z 1377 and 13C isotopepeak at 1378, is additionally observed as a more intense wateradduct (+18 Da) at m/z 1395. The pentamer is further confirmedby doubly charged ions at m/z 688.1, 688.6, and 689.1 correspond-ing with singly charged m/z values of 1376, 1377 and 1378 (13Cisotope peak), respectively. The 1376 value indicates neutralhydrogen radical transfer between ionic species. The oligomersidentified are in accordance with our predictions based on isolatedmonomers, dimers, and trimers, and in vitro reactions of catechinwith ent-fisetinidol-4b-ol. We cannot say whether extension tothe higher oligomers takes place via the ‘‘upper’’ or ‘‘lower’’ fisetin-idol unit. The structural conclusions from Fig. 5a are summarisedin Table 1. The spectrum is relatively simple. No monomers (m/z273 or 289) and little fragmentation are observed.

The positive mode ESI mass spectrum of hot water soluble que-bracho extract (Fig. 5b) is similar to the negative mode spectrum(Fig. 5a), but with more evidence of fragmentation. Prominent frag-ments at m/z 411, 683 and 955 correspond with a retro Diels–Alder(rDA) fragmentation of the m/z 563 (dimer), 835 (trimer), and 1107(tetramer) ions, respectively (Fig. 6). Interestingly, only the

-MS2 (289.20): 2.606 to 3.270 min from Sample 1 (TuneSampleID) of MT20110310161127.wiff (Turbo Spray) Max. 3.6e5 cps.

40 60 80 100 120 140 160 180 200 220 240 260 280 300m/z, amu

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Rel

. Int

. (%

)

139.3

110.9

289.4125.0 149.3

108.8

123.2 165.2121.4 137.5 163.594.8 148.2 179.5 205.192.8 271.0252.9 288.197.281.1 105.5 243.5214.9188.865.5 228.977.141.2 44.5 255.859.1

Fig. 8c. Product ion scan of an authentic robinetinidol sample.

P.B. Venter et al. / Phytochemistry 73 (2012) 95–105 103

catechin moiety undergoes RDA fragmentation and no similar frag-mentation of the fisetinidol moieties is evident. The hexamer thatappear as a minor peak at m/z 1651 is observed as more intensesingle and double water adducts at m/z 1669 and 1687, respec-tively, and the heptamer that should appear at m/z 1923 as a dou-ble water adduct (+36 Da) at m/z 1959. These water adducts are notobserved in the smaller oligomers, indicating that molecular sizeplays a role in their stability. It is well known that PAs are veryhydrophilic and water adducts are not unexpected. Similarly, PAsare antioxidants, indicating the presence of labile hydrogen radi-cals (Wright et al., 2001) and thus possible neutral hydrogen trans-fer between ionic species. Fragments corresponding to extraoxygen (+16 or +32 Da), are however not observed. PAs thatcontain more than six flavan-3-ol building blocks (hexamers andupwards) have 90 and more carbon atoms and explains the moreintense 13C isotope peaks (1 Da bigger) than 12C peaks.

Table 2 collates the oligomers up to the heptamer level andtheir corresponding rDA fragments. The rDA fragments have m/zvalues that correspond exactly with the calculated value withoutany evidence of water adducts. Rather important signals at m/z725 and 997 correspond to the loss of a resorcinol or catechol moi-ety (110.0 Da) from either the A- or B-ring of fisetinidol extenderunits.

5.2. Product ion scans of the dimers (m/z 561) and trimers (m/z 833) inhot water soluble quebracho extract

A product ion scan (APCI in the negative mode) of both the m/z561.2 (dimer) (Fig. 7a) and m/z 833.3 (trimer) (Fig. 7b) yields them/z 289.4 product ion as base peak as would be expected fromfission of a fisetinidol–catechin interflavanyl bond. The comple-mentary m/z 273 ion, associated with fisetinidol, is not observed

(although the loss of a neutral 273 Da is observed). The m/z 409and 391 ions are the result of rDA fragmentation (Scheme 1).

5.3. Product ion scans of the m/z 289 fragment (MS2), purerobinetinidol, and catechin

Comparison of the product ion scan of the m/z 289 fragment(Fig. 8a) with that of pure catechin 1 (Fig. 8b) and robinetinidol 9(Fig. 8c) confirms the aforementioned conclusion that the m/z289 fragment is catechin and not robinetinidol as was previouslyreported by Pasch et al. (2001) and Vivas et al. (2004).

5.4. Relative composition of quebracho PAs

Reliable quantification with mass spectrometry requires inter-nal standards that are not available for complex PA mixtures. PAsin quebracho, however, form a homologous series of oligomers thatdiffers only in the number of ent-fisetinidol extender units permolecule. We thus assume that the amount of each oligomer pres-ent is related to the intensity of the corresponding peak and that

Table 3Composition of quebracho extract calculated from ESI (Fig. 5a).

Oligomer M [M�H]�1 (int.) 13C[M�H]�1a (Int.) [M�H+H2O]�1 (int.) [M�2H]�2 (int.) 13C[M�2H]�2a (Int.) Total int. Weighted compos.

Dimer 562 561b (124)c 562 (44) 580 (0) 280 (0) 280.5 (0) 168 33Trimer 834 833 (118) 834 (66) 852 (0) 416 (0) 416.5 (0) 184 37Tetramer 1106 1105 (32) 1106 (22) 1124 (0) 552 (31) 552.5 (22) 107 21Pentamer 1378 1377 (0) 1378 (5 + 5) 1396 (4) 688 (15) 688.5 (10) 41 8Hexamer 1650 1649 (0) 1650 (0) 1668 (6) 824 (0) 824.5 (0) 6 1Number average molecular mass Mn = 854.96Degree of polymerization DP = 3.14Weight average molecular mass Mw = 938.76

a 13C isotope peaks.b Observed peak.c Measured intensity.

104 P.B. Venter et al. / Phytochemistry 73 (2012) 95–105

measurement of peak intensities will give a rough estimate of therelative composition of quebracho PAs. At worst we believe thatmass discrimination will underestimate the relative amount ofhigher oligomers present. 13C isotope ions become an importantfactor with oligomers and these were taken into account in ourESI quantification.

The absence of significant fragments smaller than m/z 561(dimer) in Fig. 5a (negative mode ESI) allows us to assume thatquebracho extract contains almost no flavan-3-ol monomers. Thisis in agreement with the conclusion by Roux and Evelyn (1960)that catechin and ent-fisetinidol-4-ol is virtually absent in thecentral heartwood of old quebracho trees.

A calculation of the composition of quebracho extract based onthe intensities of peaks in the ESI (negative mode, Fig. 5a) gave anumber average degree of polymerization (aDP) of 3.1 (Table 3).This is more conservative than the values of 4.5, 6.25, and 6.74determined for sulfited quebracho extract with gel permeationchromatography (Covington et al., 2005), MALDI-TOF (Paschet al., 2001), and 13C NMR (Thompson and Pizzi, 1995), respec-tively. These values agree with Mouls et al. (2011) observation thata PA extract with an aDP of 6.7 determined by thiolysis gave anaDP value of 4.9 with ESI (about 1.8 lower).

6. Conclusion

Phytochemistry and established synthetic organic chemistryperspectives were combined with a mass spectrometry investiga-tion (ESI, APCI, and product ion scans as fingerprints) to probethe chemical composition of the PAs in commercial hot water sol-uble (unsulfited) quebracho extract. Comparison of the fragmenta-tion spectrum of the m/z 289 fragment in the product ion scans ofdimers and trimers, with the fragmentation spectra of authenticsamples of catechin and robinetinidol, assigns this fragmentunequivocally to catechin. The starter unit in our sample was thusalways catechin and the extender unit always ent-fisitinedol. Thequebracho extract sample does not contain detectable quantitiesof robinetinidol, either in the monomeric form or as extender unitin prorobinetinidin type oligomers. Quebracho PAs thus consist ofa homologous series of one molecule of catechin (starter unit)linked to one, two, three, etc. ent-fisetinidol extender units. Thisconclusion is further supported by the isolation of exclusively cat-echin and ent-fisetinidol-4b-ol monomers from quebracho heart-wood, the structure of the dimers and trimers determined byNMR from the same heartwood, and biomimetic synthesis ofdimers, trimers, and a tetramer from catechin and ent-fisetinidol-4b-ol.

The relatively simple nature of quebracho extract and theabsence of large oligomers (the biggest oligomer observed was aheptamer) allowed us to estimate the composition of quebrachoextract with ESI. Quebracho PAs consist roughly of 33% dimers,37% trimers, 21% tetramers, 8% pentamers, and 1% heptamers.

Larger polymers no doubt exist, but probably in small quantities.A conservative aDP of 3.1 was calculated from the intensity ofMS fragments. Taking Mouls’s results, that the aDP of small oligo-mers is underestimated by about 1.8 with ESI relative to thethiolysis-HPLC method, the aDP of quebracho should be about 4.9.

The relatively poor solubility of hot water soluble quabraco ex-tract, as compared to mimosa extract (soluble in cold water anddoes not require sulfitation for complete extraction) is attributedto the absence of robinetinidol extender units. Robinetinidol hasone more aromatic OH than fisetinidol which increases water sol-ubility via hydrogen bonding.

Acknowledgments

Thanks are due to Prof. H. Pasch for recording the ESI spectra(Figs. 5a–c) of hot water soluble quebracho extract.

Mimosa Extract Company (Pty) Ltd. and the Technology andHuman Resources for Industry Programme (THRIP) for financialsupport.

References

Akin, D., 1982. Forage cell wall degradation and q-coumaric, ferulic, and sinapicacids. Agron. J. 74, 424–428.

Bate-Smith, E., 1954. Astringency in foods. Food Proc. Packag. 23, 124–127.Covington, A.D., 2009. Tannning Chemistry: The Science of Leather. The Royal

Society of Chemistry, Cambridge, pp. 304–309.Covington, A.D., Lilley, T.H., Song, L., Evans, C.S., 2005. Collagen and polyphenols:

new relationships and new outcomes. Part 1. Flavonoid reactions for newtanning processes. J. Am. Leather Chem. Ass. 100, 325.

Elliot, R.J., Sackwild, V., Richards, J.J., 1982. Quantitative frontier orbital theory: partI. Electrophilic aromatic substitution. Mol. Struct. 86, 301–314.

Fechtal, M., Riedl, B., 1993. Use of eucalyptus and Acacia mollissima bark extract-formaldehyde adhesives in particleboard manufacture. Holzforschung 47, 349.

Foo, L.Y., Porter, L.J., 1978. Prodelphinidin polymers: definition of structural units. J.Chem. Soc. Perkin Trans. I, 1186–1190.

Forest, K., Wan, P., Preston, C.M., 2004. Catechin and hydroxybenzhydrols as modelsfor the environmental photochemistry of tannins and lignins. Photochem.Photobiol. 3, 463–472.

Hagerman, A.E., Robbins, C.T., Weerasuriya, Y., Wilson, T.E., Mcarthur, C.J., 1992.Tannin chemistry in relation to digestion. Range Manage. 45, 57–62.

Haslam, E., 1974. Polyphenol–protein interactions. Biochem. J. 139, 285–288.Haslam, E., 1988. Twenty-second procter memorial lecture: vegetable tannins –

renaissance and reappraisal. J. Soc. Leather Technol. Chem. 72, 45.Haslam, E., 1997. Vegetable tannage: where do the tannins go? J. Soc. Leather

Technol. Chem. 82, 45.Haslam, E., 2005. Polyphenols, Collagen and Leather In: Practical Polyphenolics.

From Structure to Molecular Recognition and Physiological Action. CambridgeUniversity Press, Cambridge, p. 378.

Hofmann, T., Glabasnia, A., Schwarz, B., Wisman, K.N., Gangwer, K.A., Hagerman,A.E., 2006. Protein binding and astringent taste of a polymeric procyanidin,1,2,3,4,6-penta-O-galloyl-b-D-glucopyranose, castalagin, and grandinin. J. Agric.Food Chem. 54, 9503–9509.

Kennedy, J.A., Taylor, A.W., 2003. Analysis of proanthocyanidins by highperformance gel permeation chromatography. J. Chromatogr. A 995, 99–107.

Koupai-Abyazani, M.R., Muir, A.D., Bohm, B.A., Towers, G.H.N., Gruber, M.Y., 1993.The proanthocyanidin polymers in some species of Onobrychis. Phytochemistry34, 113–117.

P.B. Venter et al. / Phytochemistry 73 (2012) 95–105 105

Le Roux, E., Doco, T., Sarni-Manchado, P., Lozano, Y., Cheynier, V., 1998. A-typeproanthocyanidins form pericarp of Litchi chinensis. Phytochemistry 48, 1251–1258.

Mouls, L., Mazauric, J.P., Sommerer, N., Fulcrand, H., Mazerolles, G., 2011.Comprehensive study of condensed tannins by ESI mass spectrometry:average degree of polymerisation and polymer distribution determinationfrom mass spectra. Anal. Bioanal. Chem. 400, 613–623.

Pasch, H., Pizzi, A., Rode, K., 2001. MALDI-TOF mass spectrometry of polyflavonoidtannins. Polymer 42, 7531–7539.

Pizzi, A., 1978. The chemistry and development of tannin-based adhesives forexterior plywood. J. Polym. Sci. 22, 2397–2399.

Reed, J.D., Krueger, C.G., Vestling, M.M., 2005. MALDI-TOF mass spectrometry ofoligomeric food polyphenols. Phytochemistry 66, 2248–2263.

Rigaud, J., Perez-Ilzarbe, J., Ricardo da Silva, J.M., Cheynier, V., 1991. Micro methodfor the identification of proanthocyanidin using thiolysis monitored by high-performance liquid chromatography. J. Chromatogr. A 540, 401–405.

Roux, D.G., 1992. Introduction. In: Hemingway, R.W., Laks, P.E. (Eds.), PlantPolyphenols. Plenum Press, New York, pp. 7–39.

Roux, D.G., Evelyn, S.R., 1960. The distribution and deposition of tannins in theheartwoods of Acacia mollissima and Schinopsis spp. Biochem. J. 76, 17–23.

Roux, D.G., Paulus, E., 1962. Polymeric leuco-fisetinidin tannins from the heartwoodof Acacia mearnsii. Biochem. J. 82, 320–324.

Roux, D.G., Ferreira, D., Hundt, H.K.L., Malan, E., 1975. Structure, stereochemistry,and reactivity of natural condensed tannins as basis for their extendedindustrial application. Appl. Polym. Symp. 28, 335–353.

Shen, Z., Haslam, E., Falshaw, C.P., Begley, M.J., 1986. Procyanidins and polyphenolsof Larix gmelini bark. Phytochemistry 25, 2629–2635.

Takahata, Y., Ohnishi-Kameyama, M., Furuta, S., Takahashi, M., Suda, I., 2001. Highlypolymerized procyanidins in brown soybean seed coat with a high radical-scavenging activity. J. Agric. Food Chem. 49, 5843–5847.

Taylor, A.W., Barofsky, E., Kennedy, J.A., Deinzer, M.L., 2003. Hop (Humuluslupulus L.) proanthocyanidins characterized by mass spectrometry, acidcatalysis, and gel permeation chromatography. J. Agric. Food Chem. 51,4101–4110.

Thompson, D., Pizzi, A., 1995. Simple 13C NMR methods for quantitativedeterminations of polyflavonoid tannin characteristics. J. Appl. Polym. Sci. 55,107.

Thompson, R.S., Jacques, D., Haslam, E., Tanner, R.N.J., 1972. Plantproanthocyanidins. Part I. Introduction; the isolation, structure, anddistribution in nature of plant procyanidins. J. Chem. Soc. Perkin Trans. 1,1387–1399.

Vivas, N., Nonier, M.F., Vivas de Gaulejac, N., Absalon, C., Bertrand, A., Mirabel, M.,2004. Differentiation of proanthocyanidin tannins from seeds, skins and stemsof grapes (Vitis vinifera) and heartwood of quebracho (Schinopsis balansae) bymatrix-assisted laser desorption/ionization time-of-flight mass spectrometryand thioacidolysis/liquid chromatography/electrospray ionization massspectrometry. Anal. Chim. Acta 513, 256–274.

Viviers, P.M., Kolodziej, H., Young, D.A., Ferreira, D., Roux, D.G., 1983. Synthesis ofcondensed tannins. Part 11. Intramolecular enantiomerism of the constituentunits of tannins from the Anacardiaceae:Stoicheometric control in directsynthesis: derivation of 1H nuclear magnetic resonance parameters applicableto higher oligomers. J. Chem. Soc. Perkin Trans. 1, 2555–2562.

Wright, J.S., Johnson, E.R., DiLabio, G.A., 2001. Predicting the activity of phenolicantioxidants: theoretical method, analysis of substituent effects, andapplication to major families of antioxidants. J. Am. Chem. Soc. 123, 1173–1183.

Young, D.A., Ferreira, D., Roux, D.G., 1985. Synthesis of condensed tannins. Part 15.Structure of natural ‘angular’ profisetinidin tetraflavanoids: asymmetricinduction during oligomeric synthesis. J. Chem. Soc. Perkin Trans. 1, 2529–2535.