E/Z-Thesinine-O-4′-α-rhamnoside, pyrrolizidine conjugates produced by grasses (Poaceae)

Article abstract of DOI:10.1016/j.phytochem.2008.03.017

Based on direct infusion mass spectrometry we identified a novel alkaloid as a major component of perennial ryegrass (Lolium perenne). Initial mass spectral data suggested it to be a pyrrolizidine conjugate. As this class of alkaloids has not been described before from grasses, we isolated it to elucidate its structure. The isolated alkaloid proved to be a mixture of two stereoisomers. The structures of the two compounds as determined by 1D and 2D NMR spectroscopy, were E-thesinine-O-4′-α-rhamnoside (1) and Z-thesinine-O-4′-α-rhamnoside (2). These identifications were supported by the characterisation by GC-MS and optical rotation of (+)-isoretronecanol as the necine base released on alkaline hydrolysis of these alkaloids. 1 and 2 together with the aglycone and a hexoside were also detected in tall fescue (Festuca arundinacea). This is the first report of pyrrolizidine alkaloids produced by grasses (Poaceae).

Full text of DOI:10.1016/j.phytochem.2008.03.017

Phytochemistry 69 (2008) 1927–1932
Contents lists available at ScienceDirect
Phytochemistry
journal homepage: www.elsevier.com/locate/phytochem
E/Z-Thesinine-O-40-a-rhamnoside, pyrrolizidine conjugates produced
by grasses (Poaceae)
Albert Koulman a,1, Claudine Seeliger a, Patrick J.B. Edwards b, Karl Fraser a, Wayne Simpson a,
Linda Johnson a, Mingshu Cao a, Susanne Rasmussen a, Geoffrey A. Lane a,
*
a AgResearch Grasslands Research Centre, Tennent Drive, Private Bag 11008, Palmerston North, New Zealand
b Centre for Structural Biology, Institute of Fundamental Sciences, Massey University, Private Bag 11222, Palmerston North, New Zealand
a r t i c l e i n f o
a b s t r a c t
Article history:
Based on direct infusion mass spectrometry we identified a novel alkaloid as a major component of
perennial ryegrass (Lolium perenne). Initial mass spectral data suggested it to be a pyrrolizidine conjugate.
As this class of alkaloids has not been described before from grasses, we isolated it to elucidate its struc-
ture. The isolated alkaloid proved to be a mixture of two stereoisomers. The structures of the two com-
pounds as determined by 1D and 2D NMR spectroscopy, were E-thesinine-O-40-a-rhamnoside (1) and Z-
thesinine-O-40-a-rhamnoside (2). These identifications were supported by the characterisation by GC-MS
and optical rotation of (+)-isoretronecanol as the necine base released on alkaline hydrolysis of these
alkaloids. 1 and 2 together with the aglycone and a hexoside were also detected in tall fescue (Festuca
arundinacea). This is the first report of pyrrolizidine alkaloids produced by grasses (Poaceae).
Ó 2008 Elsevier Ltd. All rights reserved.
Received 17 October 2007
Received in revised form 17 March 2008
Available online 6 May 2008
Keywords:
Lolium perenne
Festuca arundinacea
Poaceae
Pyrrolizidine alkaloids
1. Introduction
2007) an ion for an unknown alkaloid of m/z 434 was observed
to predominate in many L. perenne extracts independent of the
Perennial ryegrass (Lolium perenne) and tall fescue (Festuca
arundinacea) are important pasture grasses providing the main
feed for grazing livestock in many countries e.g. USA, Australia,
and New Zealand. These two grasses have been the subject of in-
tense chemical investigation because of their toxic effects on graz-
ing mammals, particularly in relation to the syndromes commonly
known as ryegrass staggers and fescue foot. This led to the discov-
ery of a number of grass alkaloids such as perloline, a diazaphe-
nanthrene alkaloid (Grimmett and Waters, 1943; Jeffreys, 1964;
Bush and Jeffreys, 1975), which however were not responsible
for the toxicoses. During the 1970s these livestock toxicoses were
associated with the occurrence of endophytic fungi in L. perenne
and F. arundinacea, and attention moved to fungal alkaloids, lead-
ing to the discovery of the fungal metabolites lolitrem B (Gallagher
et al., 1984), and ergovaline (Lyons et al., 1986), primarily respon-
sible for the animal toxicity. Since then almost all recent chemical
attention regarding pasture grasses has been focused on fungal
secondary metabolism and no further alkaloids produced by the
host grasses have been described.
presence of endophytic fungi. This predominant ion showed some
variation between tissues and treatments (Fig. 1), but was not
identified amongst those most significantly affected by the pres-
ence of endophyte in the grass (Cao et al., 2008.) and therefore
the elucidation of its chemical identity was not an initial priority.
However, because it was a major constituent of the L. perenne
metabolome, we decided that it was prudent to characterise it.
2. Results and discussion
Infusion of an extract into the ESI source of the mass spectrom-
eter and collision-induced fragmentation and high resolution Fou-
rier Transformation mass spectrometry of the major ion (m/z 434)
provided information about the parent species and neutral loss and
product ion fragments. The [MH]+ ion (m/z 434.2175 = C23H32NO7,
calculated mass 434.2173) underwent a neutral loss of 146.0576
(C6H10O4; calculated mass 146.0579) leaving a product fragment
of m/z 288.1597 (C17H22NOþ; calculated mass 288.1594) that
3
underwent a further neutral loss of 146.0370 (C9H6O2; calculated
mass 146.0368), leaving a product fragment of m/z 142.1227
(C8H16NO+; calculated mass 142.1226). We determined the UV
kmax to be 305 nm, by LCMSPDA which together with the neutral
loss fragment of molecular formula C9H6O2 suggested that the
chromophore was a coumaryl moiety and the rest of the molecule
did not have any chromophores. Based on the accurate mass we as-
signed the first neutral loss fragment as deriving from a putative
Recent metabolomic investigations of the L. perenne endophyte
symbiosis have revealed the presence of novel grass alkaloids.
Based on direct infusion mass spectrometry (Koulman et al.,
* Corresponding author. Tel.: +64 6 356 8019; fax: +64 6 351 8032.
E-mail address: Geoff.Lane@agresearch.co.nz (G.A. Lane).
1
Current address: MRC Collaborative Centre for Human Nutrition Research (HNR),
Elsie Widdowson Laboratory, 120 Fulbourn Road, Cambridge, CB1 9NL, UK.
0031-9422/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.phytochem.2008.03.017

1928
A. Koulman et al. / Phytochemistry 69 (2008) 1927–1932
Fig. 1. Variation in intensities of the ion of m/z 434 in DIMS of extracts of different tissues of Lolium perenne uninfected (E ꢀ I: uninfected immature tissue, E ꢀ B; uninfected
blade tissue, E ꢀ S: uninfected sheath tissue) or infected with Neotyphodium lolii (Lp19) endophyte (E + I: infected immature tissue, E + B; infected blade tissue, E + S: infected
sheath tissue) (n = 3, mean stdev).
Table 1
rhamnoside, and the remaining product ion fragment as a putative
The 1H (500.1 MHz) and 13C (125.7 MHz) NMR data of E-thesinine-O-40-a-rhamnoside
(1) (in D2O)
alkaloid. The LCMS also showed that there were two partially re-
solved compounds with the same fragmentation pattern and sim-
C
13C
1H
2.79
a1.9
b2.1
a3.26
b3.45
a2.93
b3.75
a1.83
b2.15
a1.82
b2.05
4.236
a4.23
b4.32
J (HZ)
COSY
H2BC
HMBC
ilar UV spectrum suggesting that we were dealing with a mixture
of two stereo isomers (1/2). The aglycone part of the molecule cor-
responded in its molecular formula to thesinine, a saturated pyrr-
olizidine conjugate. However as the reported occurrence of
pyrrolizidine alkaloids has thus far been limited to a select number
of plant families (Reimann et al., 2004), none of which are related
to grasses, we considered more evidence of the chemical structure
was required.
The mixture of stereoisomers was extracted from L. perenne
herbage (110 g dry wt) with MeOH using a Soxhlet extractor. The
extract was made alkaline using NH3 and partitioned between
water and CH2Cl2. The organic phase was concentrated and sub-
jected to two silica column separations. This resulted in 62 mg of
yellow–brown oil, which comprised the two stereoisomers with
other little interfering material based on LCMSPDA (k 305 nm). This
was subjected to NMR spectroscopy in solution in either a CDCl3/
d4–MeOH mixture or D2O.
The proton and carbon spectrum in the CDCl3/d4–MeOH mix-
ture gave signals corresponding to those published for the agly-
cone part of thesinine-glucoside (Herrmann et al., 2002),
confirming the presence of a thesinine conjugate. However the
spectrum also showed clear evidence of a second distinct compo-
nent. To determine the complete structure of both compounds,
we chose D2O as NMR solvent because it allowed us to determine
the long-range coupling between the coumaryl part and the alka-
loid part in the HMBC spectrum, which failed in the CDCl3/d4–
MeOH mixture (Herrmann et al., 2002). The NMR data for the
two components of the mixture are reported in Tables 1 and 2.
Based on the coupling constants in the 1H NMR spectrum we were
able to determine that the purified fraction comprised a pair of
stereoisomers differing in configuration around the coumaryl dou-
ble bond. Thus the olefinic protons of the coumaryl group showed
couplings consistent with either a Z (J = 12.3 Hz) or E (J = 16.1 Hz)
configuration, and these were observed in a near 1 to 1 ratio. The
30/50 aromatic ring proton signals for the two stereoisomeric coum-
aryl moieties were overlaid, but two distinct sets of 20/60 proton
doublets (J = ca. 8.5 Hz) were observed (Tables 1 and 2). An appar-
ent isolated doublet at 5.54 ppm was assigned as the anomeric
proton of the sugar group. Careful examination of the spectrum
1
2
39.2
m
ol
ol
m
m
m
m
ol
ol
ol
ol
ol
9/8
1,3a/b
3b
2a/3b
2a/b
6a
6a
5a/b,7a/b
3b,5a, 7b
8
8
1/7
1
2,8,9
1,3
2a/b, 9a/b 7a 6a
ol
25.4
54.0
56.2
25.4
25.3
3
5
6
7
2
2a/b, 5a
6
3a/b,6a,7a
5,7
6,8
ol
ol
8
9
69.1
63.2
1,7
1
1,2b,3a/b,5a,7a
1,2a,5a
ol
dd(10.6,7.3)
10
128.5
130.1
117.1
157.4
145.7
115.4
169.0
97.9
69.8
70.0
71.9
69.5
70,80
2/60
3/50
40
7.54
7.1
d(8.6)
ol
3/50
2/60
3/50
2/60
3/50,70
2/60
100,3/50,2/60
70
7.62
6.38
d(16)
d(16)
80
70
80
70
80
70
9,70,80
90
100
200
300
400
500
600
5.54
4.09
3.92
3.45
3.67
1.14
d(1.3)
bs
dt(9.7,2.6)
t(9.7)
dt(6.4,9.7)
d(1.3)
200
200
100,300
200,400
300,500
400,600
500
100, 300
200, 400
300, 500
400, 500
600
400,100
ol
200,300,600
600
16.6
400
and further evidence from the 2D spectra revealed that this signal
comprised two doublets each with a very small coupling constant
(J = 1.3 Hz), assigned to the anomeric protons of the two stereoiso-
mers. Through a selective TOCSY experiment at 5.54 ppm with
150 ms mixing time we were able to obtain a proton spectrum of
the sugar group confirming that this was a rhamnoside type sugar
with a methyl group signal at 16.6 ppm in the 13C spectrum. As the
value of the coupling constant of H-100 was less than 7 Hz, it was
determined to be a-rhamnose (Agrawal, 1992). By overlaying the
sugar signals with the complete proton spectra we chose a well-
isolated multiplet at 2.646 ppm for a second selective TOCSY
experiment. This gave
a series of signals between 1.5 and
4.5 ppm accounting for all the protons of the pyrrolizidine part of

A. Koulman et al. / Phytochemistry 69 (2008) 1927–1932
1929
an ester bridge to E coumaryl-40-O-a-rhamnoside (1) and Z coum-
aryl-40-O-a-rhamnoside (2) (Fig. 2a), and the MS fragmentation
pattern could be assigned as in Fig. 2b.
Table 2
The 1H (500.1 MHz) and 13C (125.7 MHz) NMR data of Z-thesinine-O-40-a-rhamnoside
(2) (in D2O)
C
13C
1H
J (HZ)
COSY
H2BC
HMBC
Confirmatory evidence of structures 1 and 2 including evidence
for the absolute configuration as shown in Fig. 2 was provided by
alkaline hydrolysis. Two p-coumaric acid 4-O–rhamnoside isomers
were released, as identified by LCMSMS, together with the alkaloid
base which was identified as isoretronecanol by GC-MS compari-
son with trachelanthamidine and isoretronecanol prepared from
an alkaloid extract of a Phalaenopsis hybrid orchid by alkaline
hydrolysis (Frölich et al., 2006) and as (+)-isoretronecanol by its
positive optical rotation.
1
2
38.8
2.64
1.80
1.97
3.36
3.15
3.68
2.86
1.75
2.01
1.66
1.85
4.03
4.13
4.25
m
m
m
m
9/2a
1,3a/b
3b
2a/3b
2a/b
6a
2,8,9
1,3
2a/b, 9a/b 7a 6a
ol
25.4
53.8
56.0
25.4
25.3
3
5
6
7
2
2a/b, 5a
m
dt (5.3,11.0)
6
3a/b,6a,7a
6a
m
m
m
m
5a/b,7a/b
3b,5a, 7b
8
8
1/7
5,7
6,8
ol
ol
Thesinine, isolated from borage (Borago officinalis, Boragiaceae)
(Dodson and Stermitz, 1986), has been characterised as isoretro-
necanol linked through an ester bridge to E-p-coumaric acid. Thes-
8
9
69.0
63.2
m
1,7
1
1,2b,3a/b,5a,7a
1,2a,5a
dd (8.0,10.7)
dd (6.1,1.2)
1
inine-40-O-b-
D-glucoside has also been isolated and characterised
10
129.6
130.9
116.6
156.0
143.8
118.0
168.5
80,70
from borage (Herrmann et al., 2002). The presence of Z-thesinine
or Z-thesinine glucoside in borage (or any other species) has not
been reported.
2/60
3/50
40
7.41
7.1
d (8.5)
ol
3/50
2/60
3/50
2/60
3/50,70
2/60
100,3/50,2/60
70
7.1
5.92
d (12.1)
d (12.3)
80
70
80
70
The discovery of a pyrrolizidine alkaloid in grasses as a plant
product is very remarkable. Pyrrolizidine alkaloids which have an
ether-bridge between the C2 and C7, and a nitrogen C9-substituent
are well-known from F. arundinacea where they occur as the prod-
ucts of endophytic fungi (Neotyphodium species) and there is evi-
dence they have an important role as insect deterrents
(Wilkinson et al., 2000). However pyrrolizidine alkaloids have been
reported in only a few plant families, none closely related to
grasses, and the biosynthesis of fungal and plant pyrrolizidines is
quite different (Blankenship et al., 2005; Faulkner et al., 2006).
Within the monocots, they are known from only one other family,
the Orchidaceae, and in that case also only 1,2-saturated pyrrolizi-
dine alkaloids have been reported (Frölich et al., 2006). Plant pyrr-
olizidine alkaloids as a class are infamous for their toxicity towards
mammals and insects, but this toxicity is observed only for those
pyrrolizidine alkaloids with C1–C2 unsaturation and a hydroxy
group at C7 (Hartmann, 1999). As the grass pyrrolizidine alkaloid
conjugates identified in this study have neither, their toxicity to
mammals is likely to be limited.
Supporting evidence for this view was provided by a limited
survey of several grass cultivars by LCMSMS which showed these
alkaloids were also present in L. perenne cultivars Nui D, Impact
and Samson as well as F. arundinacea cultivars Jesup and Kentucky
31. The LCMSMS analysis showed that most of the grass samples
that contained 1 and 2, also contained hexose conjugates of 1
and 2 (likely a glycoside), and F. arundinacea also accumulates
the aglycone, thesinine. All these compounds gave a m/z 288 prod-
uct ion fragment in the MS2, which fragments further as described
for 1/2. We have sought ions in the LCMSMS chromatograms that
might suggest the presence of 1,2-unsaturated version of isoretro-
necanol, either free, or conjugated to a coumaryl group and sugars
(see material and methods, for selected masses), but thus far have
not found any evidence for such species.
When we investigated grass plants of the same cultivar from the
greenhouse we found wide variations in the level of 1/2, and with
variation in the ratios of 1 to 2 between 0.2 and 1 (data not shown).
Clonal material grown in a climate room under carefully controlled
conditions did not show such a large variation (see Fig. 1; Cao et al.,
2008), suggesting that environmental factors have a strong influ-
ence on the accumulation of E/Z-thesinine-rhamnoside.
As these pyrrolizidines are abundant in commercial cultivars of
grasses in wide use in agriculture it is unlikely that they cause any
severe toxicity for mammals. It has been suggested that 1,2 satu-
rated pyrrolizidines might have deterring effects on insect herbi-
vores, but there is no direct evidence for this (Frölich et al.,
2006). However, their common occurrence in commercial cultivars
suggests that plant breeders may have inadvertently selected for
80
70
9,70,80
90
100
200
300
400
500
600
97.6
69.8
70.0
71.9
69.5
16.6
5.55
4.09
3.92
3.45
3.67
1.15
d (1.3)
bs
dt (9.7,2.6)
t (9.7)
dt (6.4,9.7)
d (6.0)
200
200
100, 300
200, 400
300, 500
400, 600
500
100, 300
200, 400
300, 500
400, 500
600
400, 100
ol
200, 300, 600
600
400
2. A third selective TOCSY experiment was performed at a free dou-
ble doublet at 4.335 ppm, which yielded the proton signals of the
pyrrolizidine part of 1.
The structures of the three groups of both stereoisomers could
then be assigned combining information from the DQF-COSY,
HSQC, H2BC, HMBC and NOESY spectra. The spectra of the sugar
moiety showed the least difference between the two stereo iso-
mers. Only the anomeric signals were slightly separated and the
methyl group gave only one signal in the carbon spectrum. How-
ever, in the proton spectrum it appeared as a triplet as a result of
two overlapping doublets (J = 6.0 Hz). Based on the HMBC spec-
trum it was clear that the anomeric proton of the sugar is coupled
to C40 of the coumaryl group at 156 ppm (Table 2) and this linkage
was further confirmed by a nOe between the anomeric proton of
the rhamnose and the 20/60 protons of the aromatic ring in the
NOESY spectrum. The carbonyl of the coumaryl unit (C90) at
168.5 ppm is linked in the HMBC spectrum with the C9 protons
at 4.13 and 4.25 ppm in the alkaloid moiety. The C9 carbon and
its protons are linked to C1 at 39 ppm and the C1 proton. The ring
structure of this part of the molecule could be unambiguously
determined as a pyrrolizidine system by a heteronuclear 2-bond
correlations (H2BC) spectrum. This is a relatively new 2D NMR
spectroscopy method that allows two bond relationships between
protons and proton-bearing carbons to be detected, independent of
longer-range 1H–13C coupling constants (Nyberg et al., 2005). Thus
the connectivities C1–C2–C3, C1–C8–C7–C6–C5 and C1–C9 could
be established for both components (Tables 1 and 2). The COSY
spectrum established that the pyrrolizidine moiety of both compo-
nents had the same stereochemistry. In both cases a weak coupling
was observed between C1-H and both C8-H and C2-Ha but a strong
coupling between C1-H and C9-H. No coupling was observable be-
tween C1-H and C2-Hb. The NOESY spectrum shows a stronger nOe
between C1-H and C8-H than between C1-H and C9-H. This estab-
lished C1-H is syn to C8-H but gauche to the geminal C2 protons,
and hence the C1–C9 bond is trans to the bridgehead C8-H bond
in both components. This stereochemistry is that of isoretroneca-
nol. Thus we have isoretronecanol (or enantiomer) linked through

1930
A. Koulman et al. / Phytochemistry 69 (2008) 1927–1932
Fig. 2. (a) Structure of E/Z-thesinine-rhamnoside isolated from Lolium perenne. (b) Major pathway of fragmentation of protonated E/Z-thesinine-rhamnoside (m/z 434) in a
linear ion trap using 35% relative collision energy.
their presence which implies they may have favourable effects on
plant performance, for example, persistence. Further study is re-
quired to define the activity and role of these compounds.
ments (mixing time 150–200 ms) were used to identify individual
spin systems of glycoside and alkaloid groups. 1H–13C connectivi-
ties were from phase-sensitive heteronuclear single quantum
coherence (HSQC) and heteronuclear multiple bond correlation
(HMBC) experiments, and the 13C chemical shifts of were con-
firmed by direct measurement of the 13C spectrum. Double quan-
tum filtered correlation spectroscopy (DQF-COSY) and nuclear
Overhauser effect spectroscopy (NOESY) experiments (the latter
with a mixing time of 1 s) were then used to order coupling net-
works within glycoside residues and to confirm some long-range
HMBC connectivities. Standard instrument parameters were used
throughout. Two-dimensional (2D) experiments were performed
using the echo-antiecho method. Each spectrum contained 256
rows of 1024 points. Linear prediction followed by zero filling
was applied to the indirect dimension giving a final size of
1024 ꢁ 1024 points. 1H NMR spectra were referenced to residual
HOD at d4.70 ppm.
3. Experimental
3.1. General procedures
Gas chromatography-mass spectrometry was carried out on a
Shimadzu 2010 instrument (Shimadzu Corporation, Kyoto, Japan)
fitted with
a
capillary column (DB1-MS, 30 m ꢁ 0.25 mm
ID ꢁ 0.25 lm film thickness, J&W Scientific). Optical rotations were
measured on a NPL Automatic Polarimeter 143D (Thorn Bendix
Ltd., Nottingham, UK) equipped with an Hg lamp (589 nm filter)
and a 1 dm cell.
3.2. ESIMS
3.4. Plant material
Mass spectra were determined with a linear ion trap mass spec-
trometer (Thermo LTQ) using ESI in +ve mode. The spray voltage
was 5.0 kV and the capillary temperature 275 °C. The flow rates
of sheath gas, auxiliary gas, and sweep gas were set to 20, 5, and
10 (arbitrary units), respectively. Other parameters were opti-
mized automatically by infusing peramine in H2O:MeCN:HCO2H
(95:5:0.1, v/v/v) at a flow rate of 200 ll minꢀ1. UV spectra were ob-
tained using a Thermo Surveyor PDA detector (200–600 nm).
Accurate mass on the protonated compound and its fragments
from CID in an ion trap was obtained on a Thermo Finnigan LTQ
FT, which is a combination of a linear ion trap mass spectrometer
and a Fourier transform ion cyclotron resonance mass spectrometer.
Herbage of perennial ryegrass (Lolium perenne L.) Nui infected
with Epichloë festucae Fl1 was used for the preparative isolation,
and the other grass plants investigated were grown in individual
pots in the greenhouse from seed lines obtained from Margot Forde
Forage Germplasm Centre, AgResearch Grasslands, Palmerston
North, New Zealand. Leaf material from a Phalaenopsis hybrid orch-
id was provided by Accent on Flowers, Palmerston North, New Zea-
land. Plant samples were freeze dried and milled and stored in a
freezer (ꢀ20 °C) prior to extraction and isolation.
3.5. Extraction and isolation of compounds
3.3. NMR
Two batches of milled lyophilised herbage of perennial ryegrass
(ca. 55 g each) were extracted with hexane in a Soxhlet until the
pigmentation (chlorophyll plus lipids) was removed (ca. 4 h), fol-
lowed by an extraction with MeOH for 5 h. The MeOH fraction
Nuclear magnetic resonance (NMR) spectra were recorded
using a Bruker Avance 500 MHz spectrometer equipped with a
probe optimized for inverse detection. Selective TOCSY experi-

A. Koulman et al. / Phytochemistry 69 (2008) 1927–1932
1931
was concentrated to 100 ml under reduced pressure. About 50 ml
of H2O was added and the pH value was increased with 1% NH4OH
buffer to pH 10 and the extract was partitioned three times with
100 ml of CHCl3. The chloroform fraction was concentrated and
subjected to open CC using silica gel, collecting 25 ml fractions.
Elution was carried out with a series of solvents of increasing
polarity from 1 l of CHCl3:Me2CO:HCO2H (60/40/0.1), 0.5 l of
CHCl3:Me2CO:MeOH:HCO2H (25/10/65/0.1), 0.5 l of CHCl3:MeOH:
HCO2H (10/90/0.1) and ending with 0.5 l of Me2CO:MeOH:HCO2H
(10/90/0.1). To clean the column afterwards, it was eluted with
MeOH with 0.1% HCO2H. The fractions containing the compound
of interest (detected by LCMS) were combined, the pH adjusted
to 10 with 1% NH4OH buffer and the mixture partitioned for 3
times with 50 ml CHCl3 to transfer the compound to the CHCl3
phases. The CHCl3 phases were combined, concentrated and sub-
jected to open CC using silica gel, eluting with 1 l CHCl3:MeOH
(25/75) followed by 600 ml of MeOH with 0.1% HCO2H. This
yielded 62 mg of an enriched fraction.
3.7.1. E-Thesinine-O-40-a-rhamnoside (1)
Yellow oil; UV kmax MeCN/H2O (loge): 235, 305 nm (3.2). 1H and
13C NMR (500 MHz, D2O) see Table 1. HRESIMS, m/z [MH]+
434.2175 (Calcd. for C23H32NO7, 434.2173). ESIMS (positive ion
mode; m/z, rel int (%)) 434.2 [MH]+ ms2 434.20 @ 35% CE: 288.2
(100), 147.2 (2.6), 142.2 (4.1), 124 (0.6), ms3 434.20 @ 35% CE,
288.20 @ 35% CE: 147.0 (13.2), 142.12(17.3), 124.0 (100).
3.7.2. Z-Thesinine-O-40-a-rhamnoside (2)
Yellow oil; UV kmax MeCN/H2O (loge): 230, 290 nm (3.2). 1H and
13C NMR (500 MHz, D2O) see Table 2. HRESIMS, m/z [MH]+
434.2175 (Calcd. for C23H32NO7, 434.2173). ESIMS (positive ion
mode; m/z, rel int (%)) 434.2 [MH]+ ms2 434.20 @ 35% CE: 288.2
(100), 147.2 (0.5), 142.2 (1.5), ms3 434.20 @ 35% CE, 288.20 @
35% CE: 147.0 (12.4), 142.12(9.3), 124.0 (100).
3.7.3. E-Thesinine-O-40-a-rhamnoside-glycoside
ESIMS (positive ion mode; m/z, rel int (%)) 596.3 [MH]+ ms2
596.30 @ 35% CE: 434.2 (3.4), 288.2 (100), ms3 596.3 @ 35% CE,
288.20 @ 35% CE: 147.0 (14.4), 142.12 (19.3), 124.0 (100).
3.6. Identification of the necine base
Isolated alkaloid (ca. 10 mg) was dissolved in aqueous NaOH
(2.5%, 5 ml) and the mixture was allowed to stand at room temper-
ature with stirring for 10 m. LCMSMS analysis of an aliquot of the
reaction mixture with positive ESI detection as described below
showed the disappearance of the parent compound (m/z 434)
and with negative ESI detection the appearance of a pair of peaks
(m/z 309) putatively E and Z p-coumaric acid 4-O–rhamnoside.
Iso-PrOH (15 ml) was added to the mixture together with saturated
aqueous NaCl (10 ml) to aid partitioning. The organic layer was
separated, dried over anhydrous Na2SO4, and the iso-PrOH evapo-
rated under reduced pressure to give isoretronecanol as an oily res-
idue (1.1 mg). A necine base fraction was prepared for comparison
by base hydrolysis of an alkaloid extract from a Phalaenopsis hybrid
orchid, by an adaptation of the procedure of Frölich et al. (2006).
Aliquots of isoretronecanol and the Phalaenopsis necine base frac-
tion were dissolved in EtOAc and analysed by GC-MS under the
conditions described by Frölich et al. (2006). Isoretronecanol was
identified by comparison of the mass fragmentation pattern with
reference data (Wiley and NIST libraries) and by co-elution with
isoretronecanol and separation from trachelanthamidine from Pha-
laenopsis (Frölich et al., 2006).
3.7.4. Z-Thesinine-O-40-a-rhamnoside-glycoside
ESIMS (positive ion mode; m/z, rel int (%)) 596.3 [MH]+ ms2
596.30 @ 35% CE: 434.2 (3.4), 288.2 (100), ms3 596.3 @ 35% CE,
288.20@ 35% CE: 147.0 (12.4), 142.12 (8.3), 124.0 (100).
3.7.5. E- and Z-p-coumaric acid-4-O-a-rhamnoside
Analytical HPLC: Rt = 16.7 min and 18.5 min; ESIMS (negative
ion mode; m/z, rel int (%)) 309.1 [M–H]ꢀ, ms2 309.1 @ 35% CE:
163.0 (100), 119.1 (6), ms3 @ 35% CE: 119.1 (100) for both peaks.
3.7.6. (+)-Isoretronecanol
26
½aꢂD +55° (EtOH; c 0.055) Lit. [a]D +72° (Chapman and Hall Dic-
tionary of Natural Products, 2004); GC-MS: RI (DB-1MS), 1266; m/z
141 [MÅ]+.
3.7.7. Trachelanthamidine
GC-MS: RI (DB-1MS), 1239; m/z 141 [MÅ]+.
Acknowledgements
We acknowledge the support of the Foundation for Research
Science and Technology, Contract C10X0203 New opportunities
from Forage Plant Genomics and the AgResearch Board for Reinvest-
ment funding for metabolomics. We thank David Greenwood
(HortResearch and Auckland University) for his help with the
FTICRMS, and Professor Thomas Hartmann, of the Technical Uni-
versity, Braunschweig, Germany and an anonymous reviewer for
helpful suggestions.
3.7. Screening of grass extracts
Fresh plant material (between 100 and 200 mg fresh weight)
was collected in screw cap vials. For the extraction 1.5 ml of
MeOH:H2O (1/1) and two ceramic beads were added to each vial.
The vials were shaken vigorously in a BIO101/Savant FastPrep
FP120 (Qbiogene, Carlsbad, CA, USA) at 4.0 msꢀ1 for 45 s, after
which the tubes were rotated for 1 h at 30 rpm. After extraction
each vial was centrifuged at 13,000 rpm and 900 ll of the superna-
tant was transferred to an HPLC vial. Analytes were eluted through
a C18 Luna column (Phenomenex, Torrence, CA, USA) (150 ꢁ 2 mm,
5 lm) at a flow rate of 200 ll minꢀ1 using a Thermo Finnigan Sur-
veyor HPLC system with a solvent gradient (solvent A: H2O 0.1%
HCO2H; B: MeCN 0.1% HCO2H), starting with 3% B, 97% A for
5 min and then increasing to 23% B over 15 min followed by a col-
umn wash at 95% B. 1/2 were detected in SRM mode, selecting m/z
434.2 2.5, 35% relative collision energy, and quantifying the m/z
288.2 fragment ion. Furthermore the MS was set to collect frag-
mentation data in data dependent manner using a parent mass list
(m/z 140, 142, 156, 227, 286, 288, 382, 404, 432, 434, 444, 448,
450, 451, 561, 594, 596, 612) or otherwise the most intense ion.
References
Agrawal, P.K., 1992. NMR spectroscopy in the structural elucidation of
oligosaccharides and glycosides. Phytochemistry 31, 3307–3330.
Blankenship, J.D., Houseknecht, J.B., Pal, S., Bush, L.P., Grossman, R.B., Schardl, C.L.,
2005. Biosynthetic precursors of fungal pyrrolizidines, the loline alkaloids.
ChemBioChem 6, 1016–1022.
Bush, L.P., Jeffreys, J.A.D., 1975. Isolation and separation of tall fescue and ryegrass
alkaloids. Journal of Chromatography A 111, 165–170.
Cao, M., Koulman, A., Johnson, L.J., Lane, G.A., Rasmussen, S., 2008. Advanced data-
mining strategies for the analysis of direct-infusion ion trap mass spectrometry
data from the association of Lolium perenne with its endophytic fungus
Neotyphodium lolii. Plant Physiology 148, 1501–1514.
Dodson, C.D., Stermitz, F.R., 1986. Pyrrolizidine alkaloids from borage (Borago
officinalis) seeds and flowers. Journal of Natural Products 49, 727–728.
Faulkner, J.R., Hussaini, S.R., Blankenship, J.D., Pal, S., Branan, B.M., Grossman, R.B.,
Schardl, C.L., 2006. On the sequence of bond formation in loline alkaloid
biosynthesis. ChemBioChem 7, 1078–1088.

1932
A. Koulman et al. / Phytochemistry 69 (2008) 1927–1932
Frölich, C., Hartmann, T., Ober, D., 2006. Tissue distribution and biosynthesis of 1,2-
saturated pyrrolizidine alkaloids in Phalaenopsis hybrids (Orchidaceae).
Phytochemistry 67, 1493–1502.
Gallagher, R.T., Hawkes, A.D., Steyn, P.S., Vleggaar, R., 1984. Tremorgenic
neurotoxins from perennial ryegrass causing ryegrass staggers disorder of
livestock – structure elucidation of Lolitrem B. Journal of the Chemical Society –
Chemical Communications 9, 614–616.
Koulman, A., Tapper, B.A., Fraser, K., Cao, M., Lane, G.A., Rasmussen, S., 2007. High-
throughput direct-infusion ion trap mass spectrometry: a new method for
metabolomics. Rapid Communications in Mass Spectrometry 21, 421–428.
Lyons, P.C., Plattner, R.D., Bacon, C.W., 1986. Occurrence of peptide and clavine
ergot alkaloids in tall fescue grass. Science 232, 487–489.
Nyberg, N.T., Duus, J.Ø., Sørensen, O.W., 2005. Heteronuclear two-bond correlation:
Suppressing heteronuclear three-bond or higher NMR correlations while
enhancing two-bond correlations even for vanishing 2JCH. Journal of the
American Chemical Society 127, 6154–6155.
Reimann, A., Nurhayati, N., Backenkohler, A., Ober, D., 2004. Repeated evolution of
the pyrrolizidine alkaloid-mediated defense system in separate angiosperm
lineages. Plant Cell 16, 2772–2784.
Wilkinson, H.H., Siegel, M.R., Blankenship, J.D., Mallory, A.C., Bush, L.P., Schardl, C.L.,
2000. Contribution of fungal loline alkaloids to protection from aphids in a
grass-endophyte mutualism. Molecular Plant-Microbe Interactions 13, 1027–
1033.
Grimmett, R.E., Waters, D.F., 1943. A fluorescent alkaloid in ryegrass (Lolium perenne
L.) II. Extraction from fresh ryegrass and separation from other bases. New
Zealand Journal of Science and Technology, Section B 24, 151–155.
Hartmann, T., 1999. Chemical ecology of pyrrolizidine alkaloids. Planta 207, 483–
495.
Herrmann, M., Joppe, H., Schmaus, G., 2002. Thesinine-40-O-b-
D-glucoside the first
glycosylated plant pyrrolizidine alkaloid from Borago officinalis. Phytochemistry
60, 399–402.
Jeffreys, J.A.D., 1964. The alkaloids of perennial rye-grass (Lolium perenne L.). Part I.
Perloline. Journal of the Chemical Society 859, 4504–4512.