J Chem Ecol (2011) 37:57–70 DOI 10.1007/s10886-010-9887-y
Qualitative Variation in Proanthocyanidin Composition of Populus Species and Hybrids: Genetics is the Key Ashley N. Scioneaux & Michael A. Schmidt & Melissa A. Moore & Richard L. Lindroth & Stuart C. Wooley & Ann E. Hagerman
Received: 31 July 2010 / Revised: 4 November 2010 / Accepted: 12 November 2010 / Published online: 30 November 2010 # Springer Science+Business Media, LLC 2010
Abstract The literature on proanthocyanidins (tannins) in ecological systems is dominated by quantitative studies. Despite evidence that the qualitative characteristics (subunit type, polymer chain length) of these complex polyphenolics are important determinants of biological activity, little is known about genetic and environmental controls on the type of proanthocyanidins produced by plants. We tested the hypothesis that genetics, season, developmental stage, and environment determine proanthocyanidin qualitative characteristics by using four Populus “cross types” (narrowleaf [P. angustifolia], Fremont [P. fremontii], F1 hybrids, and backcrosses to narrowleaf). We used thiolysis and HPLC analysis to characterize the proanthocyanidins, and found that genetics strongly control composition. The narrowleaf plants accumulate mixed procyanidin/prodelphinidins with average composition epicatechin 11 -epigallocatechin 8 catechin2-catechin(terminal). Backcross genotypes produce mixed procyanidin/prodelphinidins similar to narrowleaf, while Fremont makes procyanidin dimers, and the F1 plants contain procyanidin heptamers. Less striking effects were noted for genotype × environment, while season and A. N. Scioneaux : M. A. Schmidt : M. A. Moore : A. E. Hagerman (*) Department of Chemistry & Biochemistry, Miami University, Oxford, OH 45056, USA e-mail:
[email protected] R. L. Lindroth : S. C. Wooley Department of Entomology, University of Wisconsin-Madison, Madison, WI 53706, USA Present Address: S. C. Wooley Department of Biological Sciences, California State University Stanislaus, 1 University Circle, Turlock, CA 95382, USA
developmental zone had little effect on proanthocyanidin composition or chain length. We discuss the metabolic and ecological consequences of differences in condensed tannin qualitative traits. Key Words Populus . Condensed tannin . Genetic variation . Polyphenolic . Proanthocyanidin . Tannin composition . Thiolysis
Introduction Secondary plant products including phenolics have important roles in regulating the interactions between plants and other organisms, in determining the responses of plants to abiotic stresses, and in governing ecosystem processes. As a consequence, investigation of how genetics, development, and environment control production of secondary compounds is essential for understanding plant physiology and ecology. Ideally, studies of secondary product accumulation include both qualitative and quantitative measures. Qualitative assessments that reveal production of specific compounds are important (Yarnes et al., 2008), since bioactivity is usually associated with specific compounds, not classes of compounds. Quantitative assays are useful for establishing metabolic flux and utilization of resources such as C and N to make broad classes of compounds (Arnold and Schultz, 2002). Studies of secondary compound accumulation often target phenolics because of their wide distribution and diverse functions. In general, phenolic compounds can be divided into the low molecular weight phenolics and the polymeric polyphenolics (tannins). The low molecular weight phenolics include numerous compounds such as phenolic acids, glycosides, esters, and flavonoids, while the
58
polyphenolics generally are classified as either hydrolyzable tannins (gallotannins and ellagitannins) or proanthocyanidins (condensed tannins) (Yoshida et al., 2005). Genetics, development, environment, and their interactions strongly influence production of low molecular weight, nontannin phenolics. Levels of phenolic glycosides in Populus are influenced by genotype and by developmental and environmental factors such as age (Donaldson et al., 2006), CO2 and nutrient levels (Lindroth et al., 2001), and defoliation or leaf damage (Stevens and Lindroth, 2005). Qualitative profiles of low molecular weight phenolics also are controlled by the interaction of genetics with developmental and environmental cues. Genetic control predominates over phenolic glycoside composition of parent and F1 offspring of Salix, with inheritance under directional dominance (Orians et al., 2000). In contrast, in three species of Betula, juvenile leaves had similar profiles of low molecular weight phenolics, but mature tissue had species-specific compositions, suggesting ontogenetic influence on genetically specified phenolic composition (Laitinen et al., 2005). Regulation of hydrolyzable tannin accumulation is similar to regulation for simple phenolics. Some oaks accumulate high levels of hydrolyzable tannin (bur, pin, or willow) while other species (cherrybark, Northern red) accumulate far smaller amounts (Foss and Rieske, 2003), suggesting an overall genetic constraint on the total quantity of hydrolyzable tannin that can be produced. Within that limit, both simple phenolics and complex ellagitannins are accumulated in patterns that are genetically controlled (Yarnes et al., 2006), with distinct tannin chemotypes identified in a population of Quercus gambelii x Q. grisea (Yarnes et al., 2008). During the growing season, oak leaves produce compounds sequentially in a pattern reflecting proposed biosynthetic relationships, with simple gallotannins appearing earlier in the growing season and ellagitannins accumulating later, thus demonstrating ontogenetic control over qualitative composition of hydrolyzable tannins (Salminen et al., 2004). Levels of accumulated hydrolyzable tannins are influenced by environmental conditions such as light and CO2, which independently modulate the amounts of gallotannins and ellagitannins produced in Acer saccharum (Agrell et al., 2000). In contrast to simple phenolics and hydrolyzable tannins, our understanding of the regulation of proanthocyanidin production is based almost entirely on quantitative assessments. Many studies have demonstrated that overall proanthocyanidin level is genetically controlled, and that environmental conditions may modify those levels. For example, in Populus (Osier and Lindroth, 2006), Salix (Orians et al., 2000), and Betula (Laitinen et al., 2000), the quantity of proanthocyanidin accumulated is genetically determined. Levels of proanthocyanidins in Populus, Salix,
J Chem Ecol (2011) 37:57–70
Betula, and Fagus are affected by environmental conditions such as levels of light, nutrients, CO2, and ozone (Lindroth et al., 2001; Lower and Orians, 2003; Karonen et al., 2006; Osier and Lindroth, 2006). Almost nothing is known about what factors dictate accumulation of specific proanthocyanidins. Subunit composition, interflavan bond position, and chain length are important variables in determining proanthocyanidin activity (Ayres et al., 1997; Kraus et al., 2003), but little is known about how these features are determined. Koupai-Abyazani et al. (Koupai-Abyazani et al., 1993a) reported that in Sainfoin, subunit composition of proanthocyanidins changed as the leaves matured, suggesting developmental influence over proanthocyanidin accumulation. However, technical difficulties in differentiating individual proanthocyanidins have prevented us from establishing whether qualitative differences in proanthocyanidins are controlled genetically, developmentally, or environmentally. This limitation has slowed our understanding of the specific functions that proanthocyanidins contribute to biological systems. We have conducted detailed qualitative studies of proanthocyanidin accumulation in Populus, examining the effects of genetics, season, developmental stage, and environmental factors. Wide-ranging “genes-to-ecosystems” studies of the Populus model system have provided insight into the relationships between plant genetics, chemistry and ecology (Whitham et al., 2006). Populus is an attractive taxon for studies of proanthocyanidin accumulation because its species do not produce hydrolyzable tannins. Populus does accumulate small phenolic glycosides, but via metabolic pathways distinct from the flavonoid metabolism that yields proanthocyanidins (Tsai et al., 2006). Recently Rehill et al. (2006) used quantitative assays to examine proanthocyanidin accumulation in four Populus “crosstypes” (two species and two hybrid types), across season, developmental stage, and location of garden plot. Similar to small phenolics, the quanitity of proanthocyanidins accumulated in leaf tissue was strongly genetically determined. The levels of proanthocyanidins in the F1 and backcross hybrids generally were intermediate to the parents (narrowleaf [P. angustifolia] and Fremont [P. fremontii]), suggesting additive inheritance. As noted for other phenolics, environmental factors including season and location modulated the levels of proanthocyanidins. In addition, significant variation in quantity of proanthocyanidin was associated with foliar developmental zone (i.e., mature and juvenile zones on individual trees) indicating developmental control of proanthocyanidin biosynthesis. Rehill et al. (2006) did not examine qualitative features of the Populus proanthocyanidins. We hypothesized that qualitative characteristics of the proanthocyanidins in Populus would be controlled mainly
J Chem Ecol (2011) 37:57–70
by genetics, just as composition of low molecular weight phenolics is mainly under genetic control. We predicted that subunit composition and chain length in F1 and backcross Populus would reflect the proanthocyanidins of their parents. We predicted that season, developmental stage, and environmental factors would have minimal effects on Populus proanthocyanidin composition. We tested these hypotheses by using leaf samples from four crosstypes of Populus collected across a growing season, at two developmental stages, and from gardens in three locations. Proanthocyanidins were extracted from the tissue, chemically derivatized with phenylmethanethiol (toluene-α-thiol), and analyzed by high performance liquid chromatography to establish subunit composition and average chain length.
Methods and Materials Plants We sampled leaf tissue from the four crosstypes of Populus described in Rehill et al. (2006): Fremont (P. fremontii), narrowleaf (P. angustifolia), F1 hybrids, and the backcrosses to narrowleaf. Genetic introgression in this system is unidirectional: hence the F1 hybrids cross only with narrowleaf, not with Fremont. All plant tissue was collected and processed as described in Rehill et al. (2006). Briefly, leaves were clipped from their petioles, flash-frozen between blocks of dry ice, and shipped to the University of Wisconsin, Madison, Wisconsin (U.S.A.). The frozen leaves were freeze-dried, then ground in a Wiley mill (#40 mesh), and stored at −20°C. Samples were sent to Miami University, Oxford, Ohio (U.S.A.) for qualitative analysis via thiolysis. To evaluate the effects of season on the composition of proanthocyanidins, leaf samples from mature branches of 7–10 trees from each of the four cross types were collected from the common garden at the Ogden Nature Center (ONC) in late May, late June, and late July 2003 (105 samples). The ONC common garden (41°14′48″ N, 111°59′ 59″ W; elevation 1300 m) represents Populus genotypes sampled from across a 13-km hybrid zone and from pure zones along the Weber River. Replicated genotypes of each of the four Populus cross types were planted in a random arrangement in the garden. To evaluate the role of developmental stage on composition of proanthocyanidins, in June 2007 leaves were collected from both the juvenile zone and the mature zone of four trees from each of five narrowleaf genotypes (1000, 1008, 1020, He-10, WC5) for a total of 40 samples. The developmental stage of the leaves was determined by the presence of catkins or catkin scars on the branches (Rehill et al., 2006). Leaves used for comparisons among developmental stages were collected only from trees growing in the ONC garden.
59
To evaluate the effects of environmental conditions on composition of proanthocyanidin, leaves were collected from mature branches of the same genotypes planted at three different common garden locations: North Pit, ONC, and Taggart gardens, all planted between 1981 and 1992. From all three common gardens, leaves from 2 to 5 plants from each of the following genotypes were collected in June 2007: backcross genotype 10, narrowleaf genotypes 1000, 1008, and 1019 (41 samples). The ONC garden, described above, is dry most of the year, but a small creek adjacent to the garden provides water intermittently. The North Pit garden (41°08′00″ N, 111°54′05″ W; elevation 1395 m) is at the entrance to Weber Canyon on a gravelly slope, near the edge of a gravel pit at the mouth of Weber Canyon. The pit garden is dry, and is exposed to constant, strong winds coming out of Weber Canyon. Neither ONC nor North Pit has supplemental irrigation. The Taggart common garden (41°03′ 38″ N, 111°34′25″ W; elevation 1587 m) is located 6 km east of Morgan, UT, USA in the channel of the Weber River. Therefore, temperatures are cooler and water is constantly available at Taggart, while both the North Pit garden and the Ogden Nature Center garden have similar, warmer temperatures and little water. Proanthocyanidin Extraction For samples from narrowleaf, F1, and backcross samples, about 10 mg of freeze-dried, ground tissue were weighed into a 1.7 ml microfuge tube. The samples were mixed with 250 μl of chloroform:methanol (2:1, v/v) with constant stirring for 30 min to remove chlorophyll and pigments, centrifuged for two min at 9300 × g, and the supernatants were discarded. The remaining tissue was treated with 250 μl ethyl acetate with constant stirring for 15 min to remove the remaining pigments and free phenolics. Samples were centrifuged as above, and ethyl acetate was discarded. After a second ethyl acetate extraction, tissues were briefly dried under flowing nitrogen gas to remove any remaining ethyl acetate. Tissues were then extracted with 200 μl of methanol with constant stirring for 60 min. The extraction protocol was modified for Fremont samples to accommodate the very low level of proanthocyanidins in this cross type. Approximately 40 mg of tissue were weighed into 1.7 ml microfuge tubes and extracted with twice the volume of solvents compared with the other cross types (500 μl chloroform methanol, 500 μl ethyl acetate, 400 μl methanol). The remaining steps in the extraction were as described above. Thiolysis Each plant extract was reacted with phenylmethanethiol (toluene-α-thiol) to yield terminal flavan-3ols and the thiol derivatives of the extender flavan-3-ols (Guyot et al., 2001a). The reaction was started by mixing 100 μl of the plant extract with 10 μl of hydrochloric acid in methanol (32% by volume) and 24 μl of phenyl-
60
methanethiol in methanol (5% by volume). The reaction mixtures were incubated at 40°C for 30 min, and reactions were stopped by placing samples at −5°C for 5 min. Samples were then stored at −80°C until analysis, with all analyses conducted within 24 h of completing the reactions. In order to determine proanthocyanidin composition, we needed standards for the flavan-3-ol thiol derivatives. We prepared the standards by carrying out thiolysis (Guyot et al., 2001b) on the proanthocyanidins from well-characterized plant sources, isolating the thiol products by column chromatography (Hagerman and Butler, 1980) and establishing identity of each product by mass spectrometry (Souquet et al., 1996). We then used these standards to identify thiol adducts in our Populus extracts by retention time on HPLC. Epicatechin thiol was obtained from Sorghum bicolor grain (Gupta and Haslam, 1978), catechin thiol from catechin trimer C-2 (Kolodzieg, 1990), epigallocatechin thiol from sainfoin leaves (Koupai-Abyazani et al., 1993b), and gallocatechin thiol from Populus tremuloides leaves (Hagerman, unpublished). The thiolysis reaction products for standards and experimental Populus samples were run on reversed phase HPLC on an Agilent 1100 system with ChemStation Rev. A.09.03 software (Agilent, Santa Clara, CA, USA). The column was an Agilent Zorbax RP-8 column, 4.6×150 mm with 5 μm packing. The chromatography employed gradient programming with a gradient comprised of 0.13% trifluoroacetic acid (TFA) in water (v/v) and 0.1% TFA in acetonitrile (v/v) as the mobile phase at a flow rate 0.5 ml/min. For the first 3 min of each run, the mobile phase was maintained with 15% of the organic phase. Linear gradients increased solvent strength to 20% organic phase at 8 min, and to 30% at 10 min. Between 28 and 32 min, a linear gradient brought the mobile phase to 70% organic. A linear gradient from 37 to 40 min returned the mobile phase to the initial condition of 15% organic, and the column was re-equilibrated from 40 to 45 min. The same solvents with a slightly different gradient program were used in a few preliminary analyses employing a Zorbax C-18 column. Two modes of detection were used: UV at 220 nm and electrochemical detection. The 12 channel CoulArray detector was employed over the voltage range from 180 to 840 mV with 60 mV increments between channels (ESA Chelmsford, MA, USA). The chromatograms were integrated automatically by using the ChemStation Rev.A.09.03 software for the UV data and Coularray for Windows Data Processing ver. 2.00 (ESA Chelmsford MA, USA) software for the electrochemical data. Baselines were set to accommodate changes in background absorbance with the gradient. To confirm the identity of the standards, HPLC-MS was run under the same conditions but substituting acetic acid for TFA in the mobile phase. The Agilent HPLC was
J Chem Ecol (2011) 37:57–70
interfaced with an Esquire LC ion trap instrument (Bruker Daltonics, Billerica, MA USA). The electrospray ionization source (ESI) was operated in negative mode. A nominal target mass was set to 1000 before fine tuning. The capillary, skimmer 1, and trap drive voltages were −100, −30, and 45 V, respectively. The ion charge control was on with a target of 30,000. The 300°C nitrogen dry gas flow rate was 4 L/min and the nebulization gas pressure was 11 psi. Each data point in the spectrum consisted of an average of 4 scans over a mass range of 120–1000 m/z. Acid Butanol Analysis To evaluate soluble and cell wall bound phenolics in the four cross types of Populus, 20 mg samples of tissue were extracted 3 times with 5 ml of aqueous acetone (3:7, water: acetone, v/v) in 50 ml falcon tubes. Subsamples (25 or 50 μl) of each extract (containing soluble proanthocyanidins) were reacted with 800 μl acid butanol (5% concentrated HCl in butanol, v/v) and 33 μl of iron reagent (2% ferric ammonium sulfate in 2 N aqueous HCl) for 10 min in a boiling water bath (Porter et al., 1986). The cell wall debris (containing bound proanthocyanidins), remaining after the exhaustive extraction, was reacted with 45 ml acid butanol and 1.8 ml ferric ammonium sulfate for 10 min in a boiling water bath with continuous stirring of the suspension. After heating, samples were cooled, and absorbances were determined at 550 nm. Statistics The experimental designs employed were orthogonal. Mixed model analysis of variance (ANOVA) models with least squares estimation were used to analyze compositional data for proanthocyanidins, via PROC MIXED (SAS version 9.1 for Windows). Effects were tested for three extenders (epicatechin, epigallocatechin, and catechin) and for chain length. Gallocatechin extender data were not statistically analyzed because this trace component was found only in relatively few backcross and narrowleaf samples from June 2003. Individual trees served as replicates for assessment of proanthocyanidins among genotypes, developmental zones, and genotype × zone interaction effects. Genotype was nested within garden with individual trees serving as replicates for the assessment of proanthocyanidins across genotypes and gardens (G × E). For comparisons of proanthocyanidins across months within and between crosstypes, we used repeated measures ANOVA with month and crosstype as fixed effects and location as the random effect.
Results Validation of Analytical Methods Thiolysis is a wellaccepted method for establishing composition and chain
J Chem Ecol (2011) 37:57–70
61
length of proanthocyanidins. The acid-catalyzed oxidative cleavage of 4,6 or 4,8 interflavan bonds yields a carbocation intermediate that reacts with the nucleophilic phenylmethanethiol to yield a thiol-derivitized flavan-3-ol. Under optimized conditions (Guyot et al., 2001a), the polymer is degraded completely to thiol adducts for all extender groups. The terminal unit is released as the underivitized flavan-3-ol. Commercially available flavan-3-ols (Fig. 1) were used to determine retention times for the epimeric pairs of the possible terminal units (Table 1). The order of elution was the same for either the C18 or C8 column, and was consistent with the trihydroxylated (delphinidin-type) B ring (gallocatechin or epigallocatechin) increasing the polarity of the compounds relative to the dihydroxylated (cyanidin-type) B ring (catechin or epicatechin). We prepared the thiol adducts of the four common flavan-3-ols (catechin thiol, epicatechin thiol, gallocatechin thiol, and epigallocatechin thiol (Fig. 1)) from plant tannins of known composition, and confirmed the identity of each by negative ion mode electrospray MS (Table 1). On the C8 column the retention times of the thiols were inverted within the epimeric pairs compared to the free flavan-3-ols, so that although catechin (2R, 3S) eluted earlier than Fig. 1 Proanthocyanidin precursor and thiolysis reaction products. In the presence of acid and benzyl mercaptan, proanthocyanidins (1) react to yield the terminal group as a free flavan-3-ol (2) and the extenders as thiol adducts (3). The dihydroxylated flavan-3-ols (catechin, epicatechin) are characteristic of procyanidins, and the trihydroxylated flavan-3-ols (gallocatechin, epigallocatechin) are characteristic of prodelphinidins
epicatechin (2R, 3R), catechin thiol eluted later than epicatechin thiol (Table 1). In contrast, using a C18 column we found that the epimeric pairs of the flavan-3-ols and their corresponding thiol adducts eluted in the same order (Table 1). With either type of column, the prodelphinidintype thiols (gallocatechin, epigallocatechin) eluted earlier than the procyanidin-type compounds (catechin, epicatechin), similar to the order for the simple flavan-3-ols. It clearly is important to identify carefully each member of the epimeric pair for any HPLC system since small changes in chromatographic conditions can change the relative retention times. We quantitated the flavan-3-ol terminal units and the thiol extender units based on detection at 220 nm and peak integration. We ran standard curves for the commercial flavan-3-ols and for the isolated thiols, establishing that the UV-response is equivalent on a molar basis for these four flavan-3-ols and their thiol adducts. For plant tissues that contained very low levels of proanthocyanidins, the UV peaks were too small to integrate accurately. For these samples, we used electrochemical detection to achieve a four-fold increase in sensitivity. Using a coularray detector, each compound was detected in a voltage window that gave the best linear response. The voltage windows were 240– R OH HO
O
OH OH
OH
S
3 R HO
S
OH
E O
+
H
OH OH R
OH
extender group thiol adduct epicatechin thiol: R = H, cis catechin thiol: R = H, trans epigallocatechin thiol: R = OH, cis gallocatechin thiol: R = OH, trans
OH HO
R
O
OH OH
OH HO
O
OH
OH
1
OH OH
proanthocyanidin E = extenders R = H, cyanidin-type R = OH, delphinidin-type
2 terminal group flavan-3-ol epicatechin: R = H, cis catechin: R = H, trans epigallocatechin: R = OH, cis gallocatechin: R = OH, trans
62
J Chem Ecol (2011) 37:57–70
Table 1 Retention timesa and ms data for flavan-3-ols and thiols produced by reaction of proanthocyanidins with phenylmethanethiol in acidic conditions Compound
Source
Gallocatechin (2R, 3S) Epigallocatechin (2R, 3R) Catechin (2R, 3S) Epicatechin (2R, 3R) Gallocatechin thiol (2R, 3S) Epigallocatechin thiol (2R, 3R) Catechin thiol (2R, 3S) Epicatechin thiol (2R, 3R)
Commercial Commercial Commercial Commercial Aspen proanthocyanidin Sainfoin proanthocyanidin C2 trimer Sorghum proanthocyanidin
C18 (min)
C8 (min)
[M-H−] m/z
MS2 m/z
4.0 4.3 6.3 7.0 12.9 12.6 15.2 14.5
3.5 4.1 5.0 6.6 17.3 18.3 26.3 27.6
426.9 426.9 410.9 410.9
302.9 302.9 287.1 287.1
a
Gradient HPLC with trifluoroacetic acid-modified aqueous and acetonitrile mobile phases was carried out on reversed phase columns with either C18 or C8 stationary phases. Detection was at 220 nm.
360 mV (catechin), 180–300 mV (catechin thiol), 360– 480 mV (epigallocatechin thiol), and 240–360 mV (epicatechin thiol). The other flavan-3-ols and thiol adducts were not present in our plant samples, so we did not optimize their electrochemical detection. We confirmed the validity of the EC detection by comparing the compositions determined by EC to the compositions determined by UV for narrowleaf samples that contained large amounts of proanthocyanidins, easily detected with either method. We converted peak areas obtained from either UV or electrochemical detection to moles using the standard curve, and then calculated moles relative to the moles of terminal flavan-3-ol, giving average composition for the proanthocyanidin from each sample. The sum of the mole ratios provided an estimate of the average degree of polymerization. Direct thiolysis was carried out on a methanol extract of proanthocyanidins from the dried, ground leaf tissue. It was essential to extract pigments and free flavan-3-ols before extracting proanthocyanidins to prevent interference in terminal unit determination. Plant tissue was extracted first with 2:1 (v/v) chloroform in methanol to remove pigments and then with ethyl acetate to remove free flavan-3-ols. HPLC analysis of the ethyl acetate fractions showed that the characteristic broad peak due to proanthocyanidins was never present in the ethyl acetate fraction, consistent with the poor solubility of proanthocyanidins in ethyl acetate. Methanol extracts of plant tissues not extracted with ethyl acetate contained catechin, but the pre-extraction with ethyl acetate removed all of the free catechin from the tissue so the methanol extracts contained only the broad, unresolved peak characteristic of proanthocyanidins on RP-HPLC. After thiolysis the products were qualitatively and quantitatively analyzed by HPLC. As expected, thiolysis reaction products did not include the typical proanthocyanidin peak but comprised peaks with retention times consistent with the underivitized flavan-3-ol, catechin; and with the various thiol derivatives of flavan-3-ols. The thiols produced from Populus
proanthocyanidins included epicatechin thiol and epigallocatechin thiol, with lower levels of catechin thiol. Gallocatechin thiol was rarely detected. There was no evidence for branched proanthocyanidins (4,6 linkages), A-type linkages, or galloyl esters. All of these variations produce thiolysis products that are chromatographically distinct from the standard 4-thiol derivatives of the flavan-3-ols, and all the products from thiolysis reactions of Poplulus leaves corresponded to the standard derivatives. Acid butanol analysis was used to evaluate the distribution of proanthocyanidins between soluble and cell wallbound fractions in the four cross types. Distribution between soluble and bound forms was consistent across cross types, with less than 10% of the total proanthocyanidin associated tightly with the cell walls in any of the plants. Because there was no quantitative variation in cell wall-bound proanthocyanidins, and because methods for qualitative analysis of cell wall-bound proanthocyanidins have not been developed, we restricted the remainder of our analyses to the soluble proanthocyanidins. Effect of Genetics on Proanthocyanidin Composition and Chain Length The strongest effect on proanthocyanidin composition was cross type (Table 2). Leaf proanthocyanidins from narrowleaf and backcross were similar in composition (Table 3, Fig. 2). The proanthocyanidins in these two cross types were made up of about 40–50% prodelphinidin-type (epigallocatechin) extenders. A typical narrowleaf proanthocyanidin in June comprises epicatechin11-epigallocatechin8catechin2-catechin(terminal) with less than one gallocatechin per chain (Table 3). In June, backcross proanthocyanidin has a slightly higher degree of polymerization and averages one gallocatechin per chain (Table 3). The Fremont proanthocyanidin was distinctly different from the proanthocyanidins from the other cross types (Table 3, Fig. 2). In contrast to narrowleaf or backcross proanthocyanidins, the Fremont samples yielded very small
J Chem Ecol (2011) 37:57–70
63
Table 2 Statistical comparisons of genetic, seasonal, developmental and environmental effects on proanthocyanidin composition in populusa Effect
DF
Epicatechin
Epigallocatechin
Catechin
Chain Length
F
P
F
P
F
P
F
P
Cross type Seasonb Cross type × Season Developmental zonec Genotype
3,38 2,54 6,54 1,14 4,14
28.02 1.48 1.42 0.54 2.17