Comparison of the chemical composition of three species of smartweed (genus Persicaria) with a focus on drimane sesquiterpenoids Author Prota, N. Plant Research International, PO Box 619, 6700 AP Wageningen, The Netherlands & Laboratory of Plant Physiology, Wageningen University, PO Box 658, 6700 AR Wageningen, The Netherlands Author Mumm, R. Author Bouwmeester, H. J. Author Jongsma, M. A. text Phytochemistry 2014 2014-12-31 108 129 136 http://dx.doi.org/10.1016/j.phytochem.2014.10.001 journal article 10.1016/j.phytochem.2014.10.001 1873-3700 10490544 2.2. Accumulation pattern of secondary metabolites during flower and leaf development in P. hydropiper To investigate the relative abundance of the secondary metabolites possibly involved in defence throughout development, three stages of flowers and four of leaves were analysed for their chemical composition. Table 4 shows how all drimane sesquiterpenoids were most abundant in Flower stage B. The leaf content of these chemicals was 13% or less, compared to flowers, and the trend was that smaller leaves contained more drimanes. This is possibly due to a relative decrease in cavity density per surface area as leaves expand, as is the case with trichomes of some plant species ( Ascensao and Pais, 1987 ). By MANOVA differences between flowers and leaves were significant for all compounds, but generally not between stages. However, a paired t -test across all compounds per stage did reveal some significant differences between stages ( Table 4 ). The accumulation patterns of different compounds throughout flower development were grouped together by means of a similarity analysis in order to point to common biosynthetic pathways, common storage tissues like cavities or common biological roles or physical properties. The similarity matrix in Fig. 2 shows how flower derived compounds of P. hydropiper cluster. The phylogenetic tree, obtained with the Unweighted Pair Group Method with Arithmetic mean (UPGMA) hierarchical clustering method, shows four distinct groups when half of the clustering distance is considered as a relevant cut-off point. The two minor groups seen at the left hand side and bottom of the matrix represent non-terpenoids (others) and diterpene neophytadiene isomers, with the exception of drimenol. Drimenol is an intermediate in the biosynthesis of polygodial and would not be expected to be stored in the cavities, which are not biosynthetically active, but rather in surrounding cells, which may produce the contents of the cavities. Therefore, an accumulation pattern of drimenol different from polygodial is expected. The three major drimane sesquiterpenes cluster together and belong to the largest group, which, next to bornyl acetate and dihydro-a- ionone, contains 12 sesquiterpenes. This could indicate that all these compounds are stored in the cavities ( Fig. 1 ). The other large cluster contains the monoterpene limonene with five other sesquiterpenes. They are characterised by lower levels in later flower stages ( Table 4 ). Possibly these are regulated differently, more volatile and/or not localised in the valvate glands. Table 2 Chemical composition of organic extracts of leaves a of three Persicaria species.

Compound	R.I.	P. hydropiper leaves b (%)	P. minor leaves b (%)	P. maculosa leaves b (%)
Monoterpenes	0.21	0.02	0.02
Bornyl acetate	1262	0.01	n.d.	n.d.
Limonene	1034	0.20	0.02	0.02
Sesquiterpenes	67.85	11.11	3.15
Drimanes	45.12	4.90	0.23
Polygodial c	1892	44.27	3.96	0.23
Isodrimenin	2039	0.65	0.14	n.d.
Drimenin	1973	0.19	0.80	n.d.
Drimenol	1788	n.d.	n.d.	n.d.
Non-drimanes	22.73	6.21	2.92
(E)-b- farnesene	1454	7.78	2.75	1.01
a- Humulene	1468	6.20	0.28	0.06
Caryophyllene	1432	6.10	2.58	0.02
a- Bisabolol	1693	1.15	0.41	n.d.
( E )-nerolidol	1563	0.50	0.05	n.d.
(E)-b- bergamotene	1492	0.32	0.04	n.d.
Aristolone	1716	0.31	0.08	n.d.
Caryophyllene oxide	1597	0.13	n.d.	n.d.
(E)-a- bergamotene	1439	0.11	0.01	n.d.
b- Bisabolene	1513	0.05	n.d.	n.d.
b- Elemene d	1397	0.03	0.01	0.62
a- Muurolene	1507	0.03	n.d.	0.88
Humulene epoxide II	1625	0.02	n.d.	n.d.
b- Bisabolol	1677	n.d.	n.d.	n.d.
d- Cadinene	1523	n.d.	n.d.	n.d.
b- Selinene	1500	n.d.	n.d.	0.33
Diterpenes	31.92	88.80	96.68
Neophytadiene isomer I	1836	21.01	59.06	62.72
Neophytadiene isomer III	1880	7.65	21.46	24.40
Neophytadiene isomer II	1861	3.26	8.28	9.56
Others	0.05	0.09	0.15
Heneicosane e	2100	0.03	0.06	0.12
Dihydro-a- ionone f	1419	0.01	n.d.	n.d.
Nonanal g	1105	0.01	0.03	0.03
Total abundance (TIC)	100	100	100
Relative abundance (TIC)	100	34.4	22.3

R.I. = retention index; n.d. = not detected. a Average of three replicates of each of four developmental stages ( N = 3 × 4, Table 3). Standard errors were on average 28% of a given value. b Relative content as percentage of the total content (sum of TIC peak areas) of a given species. Multiplication of the percentages in each column with the relative abundance of the species yields the relative contents between species. c Polygodial contains a fraction of 9- epi -polygodial, which cannot be estimated reliably by GC–MS. d b- Elemene is the heat-induced Cope rearrangement product formed from germacrene A in the GC–MS injection path ( De Kraker et al., 1998 ). e Alkane. f Carotenoid derivative. g Fatty acid derivative. Fig. 2. Structures of the five drimane sesquiterpenoids observed in the extracts of the analysed plants: (1) drimenol; (2) polygodial; (3) 9- epi -polygodial; (4) drimenin; (5) isodrimenin. 2.3. Volatile emissions from P. hydropiper In our investigation, we also analysed the headspace of P. hydropiper . We focussed only on this species, as it had a far higher content of secondary metabolites in the extracts. The volatile blend emitted by leaves and flowers of P. hydropiper contained eight major compounds which were, identified like the compounds in the extracts, by matching the library hits with the known and observed retention indices ( Table 5 ). The six sesquiterpenes in the headspace more or less reflect the relative contents in the extracts of flowers and leaves. No drimanes were detected in the headspace, presumably because their predicted boiling point ( ~ 323 °C for polygodial) is at least 50° higher compared to that of other sesquiterpenes (e.g. 261 °C for zingiberene), which results in a 228-fold lower vapour pressure and thus volatility (0.000107 vs. 0.0245 mm Hg at 25 °C for polygodial and zingiberene, respectively) (Values were generated using the US Environmental Protection Agency’s EPISuite™; information retrieved at www.chemspider.com). The monoterpene limonene was much more dominant in the headspace of flowers than in the extract, presumably due to its high volatility. Table 3 Species-specific content of polygodial a in different tissues. Table 5 Volatile compounds identified in the headspace of flowers and leaves of P. hydropiper .

Species	Flowers b Lg g FW — 1	Leaves b Lg g FW — 1
Persicaria hydropiper	6198 ± 820	500 ± 38
Persicaria minor	32 ± 6	17 ± 6
Persicaria maculosa	0.07 ± 0.03	0.7 ± 0.5

Compound	R.I.	Flowers a (%)	Leaves a (%)
b- Caryophyllene	1446	32.3	43.6
a- Humulene	1477	18.0	19.6
( E )-b- Farnesene	1450	17.0	21.8
Limonene	1040	15.4	1.2
a- Bisabolol	1650	7.1	1.7
Decanal	1211	6.6	9.6
a- Muurolene	1506	2.8	1.6
Humulene epoxide II	1611	0.9	0.9

a Polygodial contains a fraction of 9- epi -polygodial which cannot be estimated reliably by GC–MS. b Values with standard error ( N = 3). a The composition is expressed as percentage of the total peak area of the volatile blend. N = 3. Table 4 Abundance of GC–MS detectable compounds during the development of flowers and leaves in P. hydropiper .

Compound	Rank a	FA b,c	FB c (%)	FC c (%)	LA c (%)	LB c (%)	LC c (%)	LD c (%)
Monoterpenes
Limonene	15	6.86 × 10 4	109	82	16	26	12	9
Bornyl acetate	22	1.05 × 10 4	121	87	5	7	3	1
Sesquiterpenes
Drimanes
Polygodial d	1	2.45 × 10 7	202	158	12	12	9	6
Isodrimenin	7	3.92 × 10 5	213	165	12	11	8	5
Drimenin	10	2.19 × 10 5	180	101	6	6	3	3
Drimenol	25	3.69 × 10 3	341	131	6	0	0	0
Non-drimanes
Caryophyllene	2	3.63 × 10 6	94	61	10	14	6	5
a- Humulene	3	2.70 × 10 6	95	64	15	20	8	5
(E)-b- farnesene	4	1.94 × 10 6	101	80	22	37	11	15
a- Bisabolol	5	6.41 × 10 5	112	88	14	15	5	3
a- Muurolene	6	5.44 × 10 5	138	92	1	3	0	0
Aristolone	9	2.55 × 10 5	190	117	8	8	5	10
Caryophyllene oxide	12	1.40 × 10 5	115	81	4	9	3	3
( E )-b- bergamotene	13	1.15 × 10 5	95	69	23	21	9	5
( E )-nerolidol	16	5.47 × 10 4	79	50	53	93	20	10
b- Bisabolene	17	5.04 × 10 4	134	100	5	9	2	1
( E )-a- bergamotene	19	4.32 × 10 4	96	70	20	21	8	5
b- Elemene e	20	2.16 × 10 4	100	55	6	13	2	2
Humulene epoxide II	21	1.60 × 10 4	114	90	6	14	7	8
d- Cadinene	23	8.41 × 10 3	169	102	0	3	0	0
b- Bisabolol	26	3.09 × 10 3	268	135	0	0	0	0
Diterpenes
Neophytadiene I	8	3.49 × 10 5	108	154	392	315	409	298
Neophytadiene III	14	1.13 × 10 5	110	171	410	362	465	383
Neophytadiene II	18	4.91 × 10 4	111	169	400	361	450	371
Others
Heneicosane f	11	1.88 × 10 5	5	6	1	1	1	0
Dihydro-a- ionone g	24	4.92 × 10 3	292	193	11	14	7	5
Nonanal h	27	4.11 × 10 2	208	445	244	71	249	231
Average	100	171	133	18	19	15	10
p <0.05i	ac	b	bc	e	e	f	–

Columns with the same letter are not significantly different a The ‘‘Rank’’ column gives a value according to the abundance of each compound: the lower the value the more abundant the metabolite. b For flower stage 1 ( FA ) all TIC areas are given expressed in arbitrary units. Injections of 1 LL of a 50 mg/mL DCM extract of fresh tissue was used in all samples. c All other flower stages ( FB and FC ) and all leaf stages ( LA to LD ) are expressed as percentages relative to FA . d Polygodial contains a fraction of 9- epi -polygodial, which cannot be estimated reliably by GC–MS. e b- Elemene is the heat-induced Cope rearrangement product formed from germacrene A in the GC–MS injection path ( De Kraker et al., 1998 ). f Alkane. g Carotenoid derivative. h Fatty acid derivative. i Significant differences between tissues based on overall contents by paired t -test. Tissue LD was only done in duplicate and therefore not part of the analysis. Fig. 3. Similarity matrix of the compounds, obtained using the Log 2 -transformed peak areas of the different P. hydropiper flower samples throughout development. The similarity of the abundance pattern between two given compounds assumes values between — 1 and 1, with 1 meaning a 100% identical pattern. The cells at the diagonal are always intensely red as they represent the comparison of the abundance pattern of one compound with itself. 3. Conclusions In this study, we compared the chemical profiles of three species of smartweed. In all three species, the flowers contained higher amounts and more types of compounds compared with leaves of the same plant. The vast majority of the compounds identified were sesquiterpenes, five of which were drimanes. We observed that P. hydropiper produced the highest amounts of sesquiterpene secondary metabolites, while its congeners, by comparison, accumulated at least 100–500 times less in flowers, and 15– 100 times less in leaves. This differential investment in making secondary metabolites might reflect differences in defence strategies or needs between the three species of smartweed. Interesting, in that respect, is the observation that in P. maculosa , in both leaves and flowers, there seems to be selective loss (or lack of selected gain) of drimanes and most sesquiterpenes except for (E)-b- farnesene and germacrene A . According to the Optimal Defence Theory (ODT), the chemical-defence needs of any part of the plant are determined by value and vulnerability ( McCall and Fordyce, 2010 ). In this frame, young developing tissues are very vulnerable to predation by herbivores. This could explain the higher quantity of polygodial found in young leaves as well as in the tepals of those flower heads with a young developing fruit, although it does not explain why three closely related species, occurring in the same habitats have diverged so much with respect to drimane-based defences. Quite possibly the vulnerability of the other two species to whatever attacks P. hydropiper is less, so that there has been not sufficient value to drimane accumulation to balance the cost associated with producing it. A pattern similar to that of P. hydropiper accumulation of polygodial (much higher in flowerheads than in leaves) has been observed in the unrelated shrub Pseudowintera colorata (Raoul) , where berries contained roughly 100 times more polygodial and 9-deoxymuzigadial compared to leaves ( Larsen et al., 2007 ). Those authors speculate, that the presence of those two pungent sesquiterpene dialdehydes might exert a protective function against non-specialist herbivores. Considering the broad action of polygodial against mammals, insect pests and microbial pathogens, we would argue that ecological studies with genotypes that are differential in the accumulation of drimanes can provide the necessary evidence for this.