Structure Elucidation of Lolitrem F, a Naturally Occurring Stereoisomer of the Tremorgenic Mycotoxin Lolitrem B, Isolated from Lolium perenne Infected with Acremonium lolii

Article abstract of DOI:10.1021/jf950396b

A new lolitrem, lolitrem F (2), was isolated from endophyte-infected perennial ryegrass. Its structure was shown by mass spectrometry and one- and two-dimensional NMR spectroscopy to be a 31,35-cis-fused isomer of lolitrem B. Base-catalyzed epimerization of 2 and lolitrem B (1) afforded their 31-epimers (4 and 3, respectively). Comparison with spectral data for 1 and 31-epi-lolitrem B (3) established lolitrem F to be 35-epi-lolitrem B. Compounds 1, 2, and 4 were equally tremorgenic in a mouse bioassay, but 3 was nontremorgenic. Base-catalyzed exchange of H-31 was found to permit efficient incorporation of deuterium (and potentially, therefore, of tritium) into 1 and 3.

Full text of DOI:10.1021/jf950396b

2782
J. Agric. Food Chem. 1996, 44, 2782−2788
Str u ctu r e Elu cid a tion of Lolitr em F , a Na tu r a lly Occu r r in g
Ster eoisom er of th e Tr em or gen ic Mycotoxin Lolitr em B, Isola ted
fr om Loliu m per en n e In fected w ith Acr em on iu m lolii
Sarah C. Munday-Finch,† Alistair L. Wilkins,‡ Christopher O. Miles,*,† Richard M. Ede,‡ and
Ralph A. Thomson‡
New Zealand Pastoral Agriculture Research Institute Ltd., Ruakura Agricultural Research Centre,
Private Bag 3123, Hamilton, New Zealand, and Chemistry Department, The University of Waikato,
Private Bag 3105, Hamilton, New Zealand
A new lolitrem, lolitrem F (2), was isolated from endophyte-infected perennial ryegrass. Its structure
was shown by mass spectrometry and one- and two-dimensional NMR spectroscopy to be a 31,35-
cis-fused isomer of lolitrem B. Base-catalyzed epimerization of 2 and lolitrem B (1) afforded their
31-epimers (4 and 3, respectively). Comparison with spectral data for 1 and 31-epi-lolitrem B (3)
established lolitrem F to be 35-epi-lolitrem B. Compounds 1, 2, and 4 were equally tremorgenic in
a mouse bioassay, but 3 was nontremorgenic. Base-catalyzed exchange of H-31 was found to permit
efficient incorporation of deuterium (and potentially, therefore, of tritium) into 1 and 3.
Keyw or d s: Acremonium lolii; Lolium perenne; endophyte; lolitrem; tremor; ryegrass staggers;
neurotoxin; mycotoxin
INTRODUCTION
Ryegrass staggers is a neurotoxic disorder of livestock
grazing perennial ryegrass (Lolium perenne L.) that is
infected with the endophytic fungus Acremonium lolii
Latch, Christensen & Samuels. The structure of lolitrem
B (1), the major neurotoxin isolated from infected
ryegrass seed, was established in 1984 (Gallagher et al.).
Two complementary strategies for reducing the im-
pact of ryegrass staggers have been proposed: develop-
ment of L. perenne cultivars that contain endophytes
which do not produce lolitrem B and selective breeding
to produce livestock which are resistant to the neuro-
toxic effects of the lolitrems. Although the feasibility
of both strategies has been demonstrated (Campbell,
1986; Morris et al., 1995; Fletcher et al., 1993), the
probability of a successful outcome would be enhanced
by knowledge of the biosynthetic pathway to lolitrem B
and of the biological activities of the major intermediates
on that biosynthetic pathway. To this end, we recently
began a systematic study of the chemical nature and
biological activities of the lolitrems, including the minor
lolitrems.
F igu r e 1. Structures of lolitrem B (1), lolitrem F (2), 31-epi-
lolitrem B (3), and 31-epi-lolitrem F (4).
Here we report the isolation of a new minor lolitrem,
voltage of 60 V and with methanol-water (1:1) as the carrier.
Flash chromatography (Still et al., 1978) was performed on
silica gel (Merck, Art. 9385). The lolitrem content of fractions
obtained during purification was assessed by HPLC, based on
the method of Gallagher et al. (1985) as modified by Miles et
al. (1994), on a 4.6 mm × 25 cm, 5 µm Zorbax silica gel column
with acetonitrile-dichloromethane (1:4, 3:17, or 1:9, as ap-
propriate) as eluent (1.8 mL min-1). Eluting compounds were
detected with a Shimadzu RF-530 Fluorescence Spectromoni-
tor (excitation at 268 nm, emission detection at 440 nm) and
a Hewlett-Packard 1040M diode array UV detector connected
in series. Semipreparative HPLC purification was performed
on an RCM-100 radial compression separation system (Waters)
fitted with a silica gel Radial-PAK cartridge (8 mm × 10 cm,
10 µm) (Waters), with acetonitrile-dichloromethane (1:9) as
the eluent (3.0 mL min-1), and eluting compounds were
detected with an LC-85B spectrophotometric detector (Perkin-
Elmer). The tremorgenic activities of lolitrems B and F and
of 31-epi-lolitrem B and 31-epi-lolitrem F, were assessed by
intraperitoneal injection (as solutions in dimethyl sulfoxide
lolitrem F (2), from L. perenne seed, and the base
catalyzed epimerization of 1 to give 3 and of 2 to give 4
(see Figure 1). We also report on the tremorgenic
activities of 1-4 and describe a method for isotopic
labeling of lolitrems.
MATERIALS AND METHODS
Gen er a l. Electron-impact mass spectra (EI-MS) were
obtained on a Kratos MS-80 RFA instrument, using a direct
insertion probe. Electrospray mass spectra (ES-MS) were
obtained in positive ion mode on a VG Platform II (Fisons)
equipped with a MassLynx data system, with an applied cone
* Author to whom correspondence should be ad-
dressed (e-mail milesc@agresearch.cri.nz).
† Ruakura Agricultural Research Centre.
‡ The University of Waikato.
S0021-8561(95)00396-7 CCC: $12.00
© 1996 American Chemical Society

Structure Elucidation of a Tremorgenic Mycotoxin
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2783
Ta ble 1. 1Ha a n d 13C NMR (δ, CDCl3) Assign m en ts for Lolitr em B (1), Lolitr em F (2), 31-epi-Lolitr em B (3), a n d
31-epi-Lolitr em F (4)
lolitrem B (1)
31-epi-lolitrem B (3)
lolitrem F (2)
31-epi-lolitrem F (4)
13C
1Hb
13C
1Hb
13C
1Hb
13C
1Hb
C-2 (s)
C-3 (s)
C-4 (s)
C-5 (t)
C-6 (t)
C-7 (d)
C-9 (d)
152.8
50.7
42.4
27.5
28.0
71.5
71.3
71.1
61.2
67.8
78.1
30.3
20.5
50.1
29.2
118.7
126.1
124.0
137.1
120.4
110.4
142.0
16.0
19.0
74.8
28.3
16.6
196.6
60.0
80.0
79.3
49.9
28.3
30.7
25.1
25.1
29.4
92.7
122.0
139.7
18.7
25.7
152.6
50.7
42.4
27.6
28.0
71.5
71.3
71.2
61.2
67.8
78.1
30.3
20.5
50.1
29.2
118.9
125.4
123.7
136.6
120.3
110.6
142.2
16.1
19.0
74.8
28.5
16.7
197.2
57.7
82.7
82.1
47.9
25.5
27.8
33.5
24.4
29.2
92.7
122.0
139.7
18.7
25.7
152.6
50.7
42.4
27.6
28.0
71.5
71.3
71.1
61.3
67.7
78.1
30.3
20.5
50.7
29.0
118.9
125.3
123.6
136.5
120.2
110.6
142.0
16.1
19.0
74.8
28.4
16.6
197.1
57.6
82.7
82.1
47.9
25.7
33.5
27.8
29.2
24.4
92.7
122.0
139.7
18.7
25.7
152.5
50.7
42.7
27.6
28.0
71.5
71.3
71.1
61.3
67.8
78.1
30.3
20.6
49.9
29.0
d
2.70, 1.37
2.25, 1.75
4.33
3.57
3.92
2.72, 1.36
2.26, 1.76
4.33
3.57
3.92
2.73,c 1.36c
2.29,c 1.75c
4.33
3.57
3.92
d
d
4.33
3.57
3.92
3.63
C-10 (d)
C-11 (d)
C-12 (s)
C-13 (s)
C-14 (t)
C-15 (t)
C-16 (d)
C-17 (t)
C-18 (s)
C-19 (s)
C-20 (s)
C-21 (s)
C-22 (d)
C-23 (d)
C-24 (s)
C-25 (q)
C-26 (q)
C-27 (s)
C-28 (q) (eq)
C-29 (q) (ax)
C-30 (s)
C-31 (d)
C-32 (s)
C-34 (s)
C-35 (d)
C-36 (t)
C-37 (q)
C-38 (q)
C-39 (q)
C-40 (q)
C-43 (d)
C-44 (d)
C-45 (s)
C-46 (q)
C-47 (q)
NH (s)
3.63
3.63
3.62
1.56, 1.42
1.93, 1.63
2.84
1.56,c 1.42c
1.96, 1.62
2.84
1.56,c 1.42c
1.99,c 1.64c
2.85
d
d
d
d
2.62, 2.92
2.63, 2.93
2.63,2.90
d
d
d
7.87
7.22
7.83
7.21
7.84
7.22
120.3
110.4
d
7.87
7.22
1.284
1.153
1.286
1.151
1.303
1.150
16.0
19.0
74.7
28.3
16.6
196.5
60.0
80.0
79.3
49.8
28.4
25.1
30.7
29.4
25.0
92.7
122.0
140.0
18.7
25.8
1.306
1.152
1.298
1.298
1.300
1.297
1.297
1.297
1.301
1.298
2.78
3.34
3.35
2.78
2.66
2.66
2.67
d
2.96, 3.43
1.536
1.321
1.254
1.385
5.54
3.16, 3.38
1.118
1.586
1.424
1.360
5.54
3.30, 3.19
1.585
1.105
1.361
1.432
5.54
3.34, 2.99
1.322
1.528
1.389
1.255
5.54
5.30
5.30
5.30
5.32
1.732
1.746
8.00
1.731
1.746
8.01
1.731
1.744
8.06
1.732
1.746
8.00
a
Chemical shifts reported to more than the conventional number of decimal places are meant to convey the relative positions of
b
c
closely separated resonances and do not imply enhanced accuracy for the data. Proton resonances in the format HR, Hâ. Stereochemical
d
assignments of pairs of methylene resonances (HR and Hâ) assigned by analogy with 1. The position of these resonances could not be
determined due to the signal-to-noise ratio of the spectrum.
(DMSO)-water, 9:1) into mice (female CF1, weight 25 ( 5 g,
13-18 weeks old) (Miles et al., 1992). All experiments
involving animals were considered and approved by the
Ruakura Animal Ethics Committee. Lolitrem B was obtained
by methods described by Miles et al. (1994) and did not contain
a colorless solid. 1H and 13C NMR data for lolitrem F (2) are
reported in Table 1. EI-MS: m/z 686 (42%), 685.3968 (M+,
685.3981 for C42H55NO7, 84), 671 (26), 670 (60), 601 (13), 472
(12), 471 (25), 456 (23), 349 (14), 348 (52), 335 (14), 143 (25),
84 (93), 83 (100). The UV absorbance spectrum of 2 is shown
in Figure 2.
1
impurities detectable by H NMR, HPLC, or EI-MS.
Nu clea r Ma gn etic Reson a n ce Sp ectr oscop y. One- and
Ba se-Ca ta lyzed Ep im er iza tion of Lolitr em B. Lolitrem
B (1) was dissolved in ethanolic NaOH (0.1 M, 30 mL) and
allowed to stand for 48 h at 4 °C. HPLC analysis (Figure 2)
revealed the presence of new compounds. The reaction
mixture was diluted with saturated NaCl, and the products
were extracted with dichloromethane (4 × 50 mL). The extract
was dried (MgSO4) and the solvent removed in vacuo to afford
a colorless oil. Isolation of the products was achieved by
repeated flash chromatography with acetonitrile-dichlo-
romethane (3:47) or ethyl acetate-petroleum spirit (1:4) as the
eluent, followed by semipreparative HPLC. One minor and
two major products were recovered. One of the major products
was identified as unchanged 1 by HPLC and 1H NMR
spectroscopy. The other major product was 31-epi-lolitrem B
(3), identified by one- and two-dimensional NMR spectroscopy
(Tables 1 and 2). λmax (MeCN): 265 nm (ꢀ ) 47 000 ( 2000).
EI-MS: m/z 686 (36%), 685.3968 (M+, 685.3981 for C42H55NO7,
75), 601 (22), 586 (23), 471 (52), 456 (43), 349 (37), 348 (100),
335 (35), 290 (29). The UV absorbance spectrum of 3 is shown
1
two-dimensional H (300.13 MHz) and 13C (75.47 MHz) NMR
spectra were determined from deuteriochloroform (CDCl3)
solutions with a Bruker AC-300 instrument as described
previously (Munday-Finch et al., 1995).
Isola tion of Lolitr em F . Fractions enriched in lolitrem
F were obtained during isolation of lolitrem B (Miles et al.,
1994). Crystallization of lolitrem B from acetonitrile-dichlo-
romethane (Miles et al., 1994) removed most of the lolitrem
B, concentrating lolitrem F in the mother liquor. Further
purification of lolitrem F was achieved by repeated flash
chromatography of the mother liquor with acetonitrile-
dichloromethane (1:19 or 1:9) or ethyl acetate-petroleum spirit
(1:4) as the eluent. Reaction with acetic anhydride in pyridine
(1:1) for 1 h followed by flash chromatography removed
acylatable contaminants. Ketosterol impurities were removed,
when necessary, by brief treatment with NaBH4 followed by
flash chromatography (Miles et al., 1994). Final purification
was achieved by semipreparative HPLC to give lolitrem F as

2784 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Munday-Finch et al.
F igu r e 2. HPLC chromatograms of (A) epimerized lolitrem B and; (B) epimerized lolitrem F, with acetonitrile-dichloromethane
(1:4) as eluent and with detection by absorbance at 265 nm. (C) UV absorbance spectra of lolitrem B (1), lolitrem F (2), 31-epi-
lolitrem B (3), and 31-epi-lolitrem F (4) obtained from the chromatograms by means of a diode array detector.
Ta ble 2. Selected 1H NMR Cou p lin g Con sta n tsa (Hz) for Lolitr em B (1), Lolitr em F (2), 31-epi-Lolitr em B (3), a n d
31-epi-Lolitr em F (4)
lolitrem B (1)
31-epi-lolitrem B (3)
lolitrem F (2)
31-epi-lolitrem F (4)
ring A/B protons
H-31R
2.78 (d, 14.3)
3.35 (d, 7.4)
H-31â
3.34 (d, 7.3)
3.16 (dd, 12.1, 16.6)
3.38 (dd, 6.3, 16.6)
2.78 (d 14.1)
3.34 (dd, 3.9, 15.7)
2.99 (dd, 11.8, 15.7)
H-36R
H-36â
2.96 (dd, 12.3, 15.9)
3.43 (dd, 3.9, 15.9)
3.30 (dd, 6.9, 16.6)
3.19 (dd, 11.9, 16.6)
aryl protons
H-22
H-23
7.87 (d, 8.7)
7.22 (d, 8.7)
7.83 (d, 8.7)
7.21 (d, 8.7)
7.84 (d, 8.6)
7.22 (d, 8.6)
7.87 (d, 8.6)
7.22 (d, 8.6)
ring G/I protons
H-7R
4.33 (br t, 9.0)
3.57 (d, 9.4)
3.92 (d, 9.4)
5.54 (d, 6.7)
4.33 (br t, 9.0)
3.57 (d, 9.5)
3.92 (d, 9.5)
5.54 (d, 6.7)
4.33 (br t, 9.1)
3.57 (d, 9.5)
3.92 (d, 9.5)
5.54 (d, 6.7)
4.33 (br t, 8.9)
3.57 (d, 9.4)
3.92 (d, 9.5)
5.54 (d, 6.7)
H-9R
H-10â
H-43â
a
In the format δ (multiplicity, coupling constant). Abbreviations: d, doublet; t, triplet; br, broad.
in Figure 2. The minor product was identical to 2 by HPLC
solved in methanol (10 mL) and the solvent removed in vacuo;
this process was performed three times to exchange out labile
deuterium (e.g. 13-OD) present in 1 and 3. The 1H NMR
spectra of [31-2H]1 and -3 were characterized by the absence
of H-31 resonances and by modified multiplicity patterns for
the H-35 resonance (Supporting Information). In addition, the
C-31 resonance was not detected in the DEPT135 spectrum
of deuterio-1. EI-MS: for [31-2H]1, m/z 687 (18%), 686.4035
(M+ 686.4041 for C42H542HNO7, 38), 672 (17), 671 (37), 490 (13),
475 (20), 474 (19), 473 (38), 472 (92), 471 (16), 459 (22), 458
(39), 457 (93), 456 (12), 399 (12), 350 (29), 349 (92), 149 (100);
for [31-2H]3, m/z 687 (14%), 686 (27), 672 (13), 671 (27), 490
(18), 475 (30), 474 (21), 473 (42), 472 (98), 471 (24), 459 (27),
458 (42), 457 (100), 456 (17), 414 (13), 400 (15), 399 (21), 350
(29), 349 (96), 336 (41). DEPT135 13C NMR spectrum of [31-
2H]1 (CDCl3): δ 16.0 (q, C-25), 16.7 (q, C-29), 18.7 (q, C-46),
19.0 (q, C-26), 20.5 (t, C-15), 25.08 (q, C-39), 25.13 (q, C-38),
25.7 (q, C-47), 27.6 (t, C-5), 28.0 (t, C-6), 28.3 (t, C-36), 28.3
(q, C-28), 29.2 (t, C-17), 29.4 (q, C-40), 30.3 (t, C-14), 30.6 (q,
C-37), 49.8 (d, C-35), 50.1 (d, C-16), 61.2 (d, C-11), 71.2 (d,
C-10), 71.3 (d, C-9), 71.5 (d, C-7), 92.7 (d, C-43), 110.4 (d, C-23),
120.4 (d, C-22), 122.0 (d, C-44).
and 1H NMR spectroscopy.
Ba se-Ca ta lyzed Ep im er iza tion of Lolitr em F . Epimer-
ization of 2 was performed as for lolitrem B (above). Two
major products (see Figure 2) were isolated, one of which was
identified by HPLC and 1H NMR spectroscopy as being
unchanged 2. The other product was identified by NMR
spectroscopy (Table 1) as 31-epi-lolitrem F (4). EI-MS: m/z
685.3954 (M+, 685.3981 for C42H55NO7, 8%), 670 (5), 659 (10),
551 (20), 524 (14), 368 (21), 348 (22), 313 (22), 239 (27), 236
(23), 98 (58), 83 (71), 69 (100). The UV absorbance spectrum
of 4 is shown in Figure 2.
Isotop ic La belin g of Lolitr em B. To lolitrem B (1.3 mg)
in tetrahydrofuran (1 mL) was added NaOD in D2O (0.1 M, 1
mL, 99 atom % D). The reaction was protected from light and
allowed to stand at room temperature for 20 h. It was then
added to saturated NaCl (100 mL), the products were extracted
with dichloromethane (3 × 25 mL), the extract was dried
(MgSO4), and the solvent was removed in vacuo to give a
colorless oil. Analysis by HPLC revealed the product to be an
86:14 mixture of 1 and 3, which was separated by semi-
preparative HPLC. The separated products were each dis-

Structure Elucidation of a Tremorgenic Mycotoxin
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2785
Ta ble 3. Lon g-Ra n ge 13C-1H NMR Cor r ela tion s
Deter m in ed for th e Meth yl Gr ou p P r oton s of Lolitr em F
(2)
1H signala (δ)
correlated 13C signals (δ)
1.105 (H-38)
1.150 (H-26)
1.297 (H-28)
1.297 (H-29)
1.303 (H-25)
1.361 (H-39)
1.432 (H-40)
1.585 (H-37)
1.731 (H-46)
1.744 (H-47)
33.5 (C-37), 57.6 (C-31), 82.7 (C-32)
27.6 (C-5), 42.4 (C-4), 50.7 (C-3), 78.1 (C-13)
16.6 (C-29), 71.3 (C-9), 74.8 (C-27)
28.4 (C-28), 71.3 (C-9), 74.8 (C-27)
50.7 (C-3 and C-16), 42.4 (C-4), 152.6 (C-2)
24.4 (C-40), 47.9 (C-35), 82.1 (C-34)
29.2 (C-39), 47.9 (C-35), 82.1 (C-34)
27.8 (C-38), 57.6 (C-31), 82.7 (C-32)
25.7 (C-47), 122.0 (C-44), 139.7 (C-45)
18.7 (C-46), 122.0 (C-44), 139.7 (C-45)
a
Chemical shifts reported to more than the conventional
number of decimal places are meant to convey the relative
positions of closely separated resonances and do not imply
enhanced accuracy for the data.
F igu r e 3. Low-energy conformer of lolitrem F (2), showing
selected NOEs observed for the ring A/B protons.
Ta ble 4. Lon g-Ra n ge 13C-1H NMR Cor r ela tion s
Deter m in ed for th e Meth yl Gr ou p P r oton s of
31-epi-Lolitr em B (3)
To assess the stability of the label, deuterio-1 and -3 (<1
mg) were each placed in a sealed container with methanol (25
mL) and stored in the dark at 30 °C for 35 days. Samples of
each solution were taken regularly, and the solvent was
removed under a stream of dry nitrogen. The samples were
then stored in the dark at 4 °C until being redissolved in
methanol immediately prior to analysis by ES-MS. The degree
1H signala (δ)
correlated 13C signals (δ)
1.118 (H-37)
1.151 (H-26)
1.286 (H-25)
1.297 (H-29)
1.300 (H-28)
1.360 (H-40)
1.424 (H-39)
1.586 (H-38)
1.731 (H-46)
1.746 (H-47)
33.5 (C-38), 57.7 (C-31), 82.7 (C-32)
27.6 (C-5), 42.4 (C-4), 50.7 (C-3), 78.1 (C-13)
50.7 (C-3), 50.1 (C-16), 42.4 (C-4), 152.6 (C-2)
28.5 (C-28), 71.3 (C-9), 74.8 (C-27)
16.7 (C-29), 71.3 (C-9), 74.8 (C-27)
24.4 (C-39), 47.9 (C-35), 82.1 (C-34)
29.2 (C-40), 47.9 (C-35), 82.1 (C-34)
27.8 (C-37), 57.7 (C-31), 82.7 (C-32)
25.7 (C-47), 122.0 (C-44), 139.7 (C-45)
18.7 (C-46), 122.0 (C-44), 139.7 (C-45)
2
of H incorporation in 1 and 3 was estimated by comparison
of the intensities of the peaks at m/z 686 (MH+ for 1 and 3)
and 687 (MH+ for [31-2H]1 and -3) after making allowance for
the contribution of the undeuterated MH+1+ peak to m/z 687.
Molecu la r Mod elin g. The energy-minimized three-di-
mensional structures of lolitrems B and F (depicted in part in
Figure 3), and their C-31 epimers, were determined on an Iris
Indigo computer (Silicon Graphics) running MacroModel ver-
sion 4.5 (Chemistry Department, Columbia University, New
York, NY) with the supplied MM2* constants, energy mini-
mization, and Monte Carlo search routines.
a
Chemical shifts reported to more than the conventional
number of decimal places are meant to convey the relative
positions of closely separated resonances and do not imply
enhanced accuracy for the data.
complete carbon and proton assignment reported in
Table 1 was derived from an analysis of one- and two-
dimensional NMR data, including COSY, NOE-differ-
ence, HMQC, and HMBC (Table 3) data in a manner
analogous to that used previously for related lolitrem
derivatives (Miles et al., 1992, 1994; Munday-Finch et
al., 1995).
Lolitrem F is thereby established as being an isomer
of lolitrem B (1) possessing a cis-fused (rather than a
trans-fused, as in 1) ring A/B junction, but it was not
possible to determine whether lolitrem F had structure
2 or structure 3 solely from spectroscopic data. How-
ever, we anticipated that treatment of 1 with base would
result in epimerization at C-31, to give a mixture of 1
and 3, and that comparison of the chemical and spectral
properties of 3 with those of lolitrem F would allow the
structure of the latter compound to be assigned as being
either 2 or 3.
RESULTS AND DISCUSSION
Str u ctu r e of Lolitr em F . The mass spectrum of
lolitrem F (2) was very similar to that of lolitrem B (1).
In particular, the mass of the molecular ion was
consistent with the same atomic composition as lolitrem
B (C42H55NO7), and the prominent ion at m/z 348 was
consistent with the presence in 2 of the ring A-E
structure that is present in 1. The UV spectrum of 2
was very similar to that of 1, but λmax was shifted 2 nm
toward longer wavelengths (Figure 2), suggesting the
presence of similar chromophores in the two compounds.
As was the case for lolitrem B (1), the 13C NMR
spectrum of lolitrem F (2) comprised 15 singlet, 11
doublet, 6 triplet, and 10 quartet resonances. With the
exception of C-16 and C-19, the H and 13C resonances
1
attributable to rings D-I of 2 occurred within (0.06
ppm or (0.2 ppm, respectively, of the equivalent
resonances of 1 (Table 1), indicating that 1 and 2 are
identical in rings D-I and that the difference between
them lies in rings A and B. In particular, we observed
that the C-31 resonance of 2 occurred at 57.6 ppm
(compared to 60.0 in 1) and that this resonance cor-
related with the proton (H-31) resonating at 3.35 ppm
(J H-31-H-35 ) 7.4 Hz). In 1, H-31 resonates at 2.78 ppm
(J H-31-H-35 ) 14.3 Hz). The diminished value of
J H-31-H-35 in 2 suggested the existence of a cis-
relationshipsrather than a trans-relationship, as in
1sbetween H-31 and H-35. Consistent with this con-
clusion, irradiation of the H-31 resonance of 2 in an
NOE-difference experiment enhanced H-35 (2.67 ppm),
H-37 (1.585 ppm), and H-39 (1.361 ppm). Other struc-
turally significant NOEs are depicted in Figure 3. The
As expected, epimerization of 1 generated a mixture
of 1 and 3, from which 31-epi-lolitrem B (3) was isolated
by column chromatography. The assignments of the
NMR spectra of 3 were obtained by methods analogous
to those used for lolitrem F (above) and are reported in
Table 1. Long-range 13C-1H correlated (HMBC) data
1
are given in Table 4. The H and 13C NMR resonances
of 3 occurred within (0.03 and (0.2 ppm, respectively,
of the equivalent (note that C-37 of 2 is regarded as
equivalent to C-38 of 3 because they have identical
orientations with respect to rings A-D; the same
considerations apply to C-39/C-40 and H-36R/H-36â)
resonances of lolitrem F (2), except for the equatorial
H-36 resonances (which differed by 0.08 ppm) and the
C-16 resonance. The UV spectrum of 3 was qualita-
tively identical to that of lolitrem F and was slightly

2786 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Munday-Finch et al.
Ta ble 5. Lon g-Ra n ge 13C-1H NMR Cor r ela tion s
Deter m in ed for th e Meth yl Gr ou p P r oton s of
31-epi-Lolitr em F (4)
1H signala (δ)
correlated 13C signals (δ)
1.152 (H-26)
1.255 (H-40)
1.298 (H-29)
1.301 (H-28)
1.306 (H-25)
1.322 (H-37)
1.389 (H-39)
1.528 (H-38)
1.732 (H-46)
1.746 (H-47)
27.6 (C-5), 42.7 (C-4), 50.7 (C-3), 78.1 (C-13)
29.4 (C-39), 49.8 (C-35), 79.3 (C-34)
28.3 (C-28), 71.3 (C-9), 74.7 (C-27)
16.6 (C-29), 71.3 (C-9), 74.7 (C-27)
50.7 (C-3), 49.9 (C-16), 42.7 (C-4), 152.5 (C-2)
30.7 (C-38), 60.0 (C-31), 80.0 (C-32)
25.0 (C-40), 49.8 (C-35), 79.3 (C-34)
25.1 (C-37), 60.0 (C-31), 80.0 (C-32)
25.8 (C-47), 122.0 (C-44), 140.0 (C-45)
18.7 (C-46), 122.0 (C-44), 140.0 (C-45)
a
Chemical shifts reported to more than the conventional
number of decimal places are meant to convey the relative
positions of closely separated resonances and do not imply
enhanced accuracy for the data.
different from that of 1 (Figure 2). However, although
the spectral properties of lolitrem F were similar to
those of 3, HPLC analysis revealed that 3 did not coelute
with lolitrem F (Figure 2), and lolitrem F is therefore
assigned structure 2.
F igu r e 4. Structures of paxilline (5), lolitriol (6), paspalicine
(7), and paspalinine (8).
isomers of two (compared to one only, in the trans-fused
isomers) CH3 singlet resonances at δ < 1.16, small but
consistent differences in the chemical shifts of C-30 to
C-36 (Table 1), and λmax for the cis-isomers at slightly
longer wavelengths (by 2-3 nm) in both acetonitrile (see
Munday-Finch et al. (1995)) and acetonitrile-dichlo-
romethane (1:4) (Figure 2).
Molecular modeling of lolitrem F (2) revealed the
existence of twosrather than only one, as is the case of
the trans-fused analogue lolitrem B (1)slow-energy ring
A/B conformations. The two conformers were calculated
to differ in energy by 6.7 kJ mol-1. Some of the NOEs
observed within the A/B ring system of 2 (H-40-H-36R,
H-39-H-31R, H-39-H-37, and H-38-H-36â) were con-
sistent (calculated interatomic distances all < 2.3 Å)
with the low-energy, but not with the high-energy
(calculated interatomic distances all > 3.2 Å), conformer
(see Supporting Information). The low-energy con-
former and NOEs observed for the ring A/B region of 2
are depicted in Figure 3.
Biosyn th esis of Lolitr em F . The lolitrems are
thought to be produced by the ryegrass endophyte A.
lolii by modification of a paxilline-like biosynthetic
precursor (Weedon and Mantle, 1987; Miles et al., 1992,
1994; Penn and Mantle 1994; Gurney et al., 1994;
Munday-Finch et al., 1995). Conversion of paxilline (5)
(Figure 4), itself a metabolite of A. lolii (Weedon and
Mantle 1987), into lolitrem B (1) requires three steps:
epoxidation of the double bond at C-11 of 5; addition of
an isoprene unit across the oxygen atoms on C-10 (with
reduction of the carbonyl at C-10) and C-25 of 5 to form
the acetal moiety of 1; and addition of isoprene units to
C-20 and C-21 of 5, with cyclization to form the A and
B rings of 1. Although it is not known in what order
these transformations occur, the identification of lolitriol
(6) as a probable biosynthetic precursor of 1 (Miles et
al., 1992) suggests that formation of the acetal moiety
may be the final step in the biosynthesis of 1.
The identification of 2 as a metabolite of A. lolii
indicates a lack of stereospecificity in the enzymes
involved in the cyclization of the isoprene units during
the formation of the lolitrem A/B ring system. The
presence of the acetal moiety in 2 also suggests that
acetal formation during the biosynthesis of 1 and 2
occurs independently of the stereochemistry present at
C-35. This being so, 31R,35R-analogues of lolitriol,
lolitrem A, and lolitrem E might also be expected to be
present in endophyte infected ryegrass, but at much
lower levels than lolitrem F.
The assignment of structure 2 to lolitrem F was
supported by epimerization of lolitrem F (2) to a mixture
1
of 2 and 4. The H NMR spectral properties of 4 (see
Tables 1 and 2) were very similar to those of 1, with
the sole exception of the equatorial H-36 resonances
(which differed by 0.09 ppm). The 13C NMR assign-
ments determined for 4 were supported by HMQC and
HMBC data (see Table 5). The signal-to-noise ratio of
the HMQC spectrum of 4 was such that although we
were able to correlate the carbon and proton resonances
of the methine (CH) and methyl (CH3) groups, we were
not able to establish the resonances of individual
methylene (CH2) protons.
The trans-fused ring A/B system of lolitrem B (1) has
been assigned as having 31R,35â stereochemistry (Ede
et al., 1994). The foregoing discussion establishes that
lolitrem F is a 31,35-cis-isomer of lolitrem B (1), but was
not 31-epi-lolitrem B (3); lolitrem F must therefore be
35-epi-lolitrem B (2). Although it is conceivable that 3
could be generated from the much more abundant 1
during its extraction and purification, it is unlikely that
2 could be an artifact of the isolation procedure. Indeed,
2, but never 3 or 4, was routinely detected during HPLC
analysis of freshly prepared extracts of ryegrass.
Lolitrem F (2) is therefore a natural product and is
presumably produced by the fungal endophyte in a
manner similar to the other lolitrems (Munday-Finch
et al., 1995).
Biologica l Activity. The tremorgenic activities of
1-4 were determined in a standard mouse bioassay
(Figure 5). Compounds 2 and 4 were found to have
potencies and time courses of action similar to those of
lolitrem B (1). The 31-epimer of lolitrem B (3) did not,
however, cause detectable tremors. Because 1 causes
discernible tremors at 1 mg kg-1 under these assay
conditions (Miles et al., 1992), 3 is at least 4 times less
tremorgenic than are 1, 2, and 4. The slightly reduced
activities of 2 and 4, relative to 1, may be due to the
Characteristic differences were observed between the
NMR and UV spectra of lolitrems possessing cis- and
trans-fused A/B rings. Particularly diagnostic were the
chemical shift and coupling constant of H-31 (Table 2),
1
the presence in the H NMR spectrum of the cis-fused

Structure Elucidation of a Tremorgenic Mycotoxin
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2787
F igu r e 5. Mean tremor score vs time postinjection for groups of mice dosed with (A-C) 1 (O) (n ) 5 mice) and (A) 4 (b) (n ) 4),
(B) 2 (b) (n ) 5), and (C) 3 (b) (n ) 5) at 4 mg kg-1. Error bars indicate the standard error of the mean.
presence of small amounts of undetected impurities, as
only 1 was sufficiently abundant to be purified by
fractional crytallization.
isotopic label but also shifted the position of the 1-3
equilibrium in favor of lolitrem B; HPLC indicated that
the equilibrium ratio of 1:3 was 86:14 in tetrahydrofu-
ran-water and 55:45 in ethanol-water. Higher yields
of labeled 1 are therefore obtained when THF, rather
than ethanol, is used as the cosolvent. Appropriate
modifications to this methodology should permit forma-
tion of [31-3H]1.
The label on C-31 appears to be stable to exchange
in neutral hydroxylic solvents, as no loss of 2H from [31-
2H]1 or -3 was observed upon standing in methanol for
35 days. Furthermore, the absence of detectable levels
of 3 or 4 in ryegrass seed or extracts stored for up to 10
years suggests that enolizationswhich would cause loss
of the label at C-31 by exchangesis unlikely to be rapid
under physiological conditions.
Molecular modeling of 1-4 revealed the trans-fused
isomers (1 and 4) to be relatively planar in rings A-H.
In contrast to this, the cis-fused isomers (2 and 3)
possess an A ring which protrudes from the plane of
the molecule (see Figure 3). In 3, the A ring protrudes
onto the R-face, whereas in 2 it protrudes onto the
â-face. It would appear that the stereochemistry of the
A/B ring junction does not influence the tremorgenic
activity of the lolitrems, unless the A ring is oriented
onto the R-facesin which case activity is abolished.
Compounds 1-4 are very similar in their physical
properties, so the lack of tremorgenic activity by 3 may
be due to the preventionsby the presence of the A ring
(and/or its appended methyl groups) on the R-face of the
moleculesof effective binding to the necessary receptor
site(s). This would suggest that interaction of such
receptors with the R-face of the lolitrems may be
important in causing the observed neurotoxic effects.
This hypothesis is consistent with the observation
(Springer and Clardy, 1980; Gallagher et al., 1980),
made on the closely related indole-diterpenoids paspal-
icine (7) and paspalinine (8) (Figure 4), that the pres-
ence of an R-oriented hydroxyl at C-13 is required for
tremorgenic activity.
The resistance of sheep to ryegrass staggers has been
shown to be a heritable trait (Campbell, 1986; Morris
et al., 1995), but the biochemical basis of this resistance
is unknown. Furthermore, the biochemical basis of the
tremorgenic effects of the indole-diterpenoid neurotox-
ins, and of the unusually prolonged tremors caused by
lolitrems A, B, and F, is also unclear. Biochemical
studies of the indole-diterpenoids are severely con-
strained by the difficulty of obtaining radiolabeled toxin;
this is especially so for the lolitrems, which are produced
at low (Miles et al., 1992, 1994; Penn et al., 1993), or
undetectable (Weedon 1987; Weedon and Mantle, 1987),
levels by A. lolii when grown in culture.
Thus, practical methods are now available for produc-
tion of lolitrem B (Miles et al., 1994) and for its efficient
conversion into an isotopically-labeled form, opening the
way for fundamental studies into the mode of action of
the lolitrem neurotoxins.
ACKNOWLEDGMENT
We thank D. E. McNaughton and B. Clarke for
obtaining EI-MS, W. J . J ackson for ES-MS, J . M. Allen
for high-resolution EI-MS of [31-2H]1, and B. L. Smith
for assistance with experiments involving animals.
Su p p or tin g In for m a tion Ava ila ble: Figures showing
1
ES-MS and H NMR spectra of deuteriated 1 and 3 and a table
listing calculated interatomic distances (Å) for low- and high-
energy conformers of lolitrem F (2) (3 pages). Ordering
information is given on any current masthead page.
LITERATURE CITED
Campbell, A. G. Selection strategies for animal disease resis-
tance. N. Z. Agric. Sci. 1986, 20, 169-171.
Ede, R. M.; Miles, C. O.; Meagher, L. P.; Munday, S. C.;
Wilkins, A. L. Relative stereochemistry of the A/B rings of
the tremorgenic mycotoxin lolitrem B. J . Agric. Food Chem.
1994, 42, 231-233.
Fletcher, L. R.; Garthwaite, I.; Towers, N. R. Ryegrass staggers
in the absence of lolitrem B. In Proceedings of the Second
International Symposium on Acremonium/Grass Interac-
tions; Hume, D. E., Latch, G. C. M., Easton, H. S., Eds.;
AgResearch: Palmerston North, New Zealand, 1993; pp
119-121 [ISSN no. 1052-5181].
Gallagher, R. T.; Finer, J .; Clardy, J .; Leutweiler, A.; Weibel,
F.; Acklin, W.; Arigoni, D. Paspalinine, a tremorgenic
metabolite from Claviceps paspali Stevens et Hall. Tetra-
hedron Lett. 1980, 235-238.
Isotop ic La belin g of Lolitr em B. The epimeriza-
tion reaction used to generate 3 and 4 involves base-
catalyzed exchange of the weakly acidic H-31 proton of
the lolitrems with the hydroxylic protons present in the
solvent; epimerization of 1 in the presence of tritiated
water ought, therefore, to generate a readily separable
mixture of [31-3H]1 and [31-3H]3.
We explored the feasibility of this approach with
deuteriated, rather than tritiated water, and obtained
very satisfactory results. Mass spectrometry indicated
1
>98% incorporation of deuterium into 1 and 3, and H
and 13C NMR spectroscopy did not reveal incorporation
at any site other than at the 31-position. The use of
tetrahydrofuran as the cosolvent for the isotope ex-
change reactionsrather than ethanol, as in the epimer-
ization reactionsnot only prevented dilution of the
Gallagher, R. T.; Hawkes, A. D.; Steyn, P. S.; Vleggaar, R.
Tremorgenic neurotoxins from perennial ryegrass causing
ryegrass staggers disorder of livestock: Structure elucida-
tion of lolitrem B. J . Chem. Soc., Chem. Commun. 1984,
614-616.

2788 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Munday-Finch et al.
Gallagher, R. T.; Hawkes, A. D.; Stewart, J . M. Rapid deter-
mination of the neurotoxin lolitrem B in perennial ryegrass
by high-performance liquid chromatography with fluores-
cence detection. J . Chromatogr. 1985, 321, 217-226.
Gurney, K. A.; Mantle, P. G.; Penn, J .; Garthwaite, I.; Towers,
N. R. Loss of toxic metabolites from Acremonium lolii, the
endophyte of ryegrass, following mutagenesis. Naturwis-
senschaften 1994, 81, 362-365.
Miles, C. O.; Wilkins, A. L.; Gallagher, R. T.; Hawkes, A. D.;
Munday, S. C.; Towers, N. R. Synthesis and tremorgenicity
of paxitriols and lolitriol: Possible biosynthetic precursors
of lolitrem B. J . Agric. Food Chem. 1992, 40, 234-238.
Miles, C. O.; Munday, S. C.; Wilkins, A. L.; Ede, R. M.; Towers,
N. R. Large-scale isolation of lolitrem B and structure
determination of lolitrem E. J . Agric. Food Chem. 1994, 42,
1488-1492.
Penn, J .; Garthwaite, I.; Christensen, M. J .; J ohnson, C. M.;
Towers, N. R. The importance of paxilline in screening for
potentially tremorgenic Acremonium isolates. In Proceedings
of the Second International Symposium on Acremonium/
Grass Interactions; Hume, D. E., Latch, G. C. M., Easton,
H. S., Eds.; AgResearch: Palmerston North, New Zealand,
1993; pp 88-92 [ISSN no. 1052-5181].
Springer, J . P.; Clardy, J . Paspaline and paspalicine, two
indolemevalonate metabolites from Claviceps paspali. Tet-
rahedron Lett. 1980, 231-234.
Still, W. C.; Kahn, M.; Mitra, A. Rapid chromatographic
technique for preparative separations with moderate resolu-
tion. J . Org. Chem. 1978, 43, 2923-2925.
Weedon, C. M. Biosynthesis of indole-diterpenoids and other
isoprenoids by the ryegrass endophyte, Acremonium loliae.
Ph.D. Thesis, University of London, 1987.
Morris, C. A.; Towers, N. R.; Wheeler, M.; Amyes, N. C. A note
on the genetics of resistance or susceptibility to ryegrass
staggers in sheep. N. Z. J . Agric. Res. 1995, 38, 367-371.
Munday-Finch, S. C.; Miles, C. O.; Wilkins, A. L.; Hawkes, A.
D. Isolation and structure elucidation of lolitrem A, a
tremorgenic mycotoxin from perennial ryegrass infected
with Acremonium lolii. J . Agric. Food Chem. 1995, 43,
1283-1288.
Weedon, C. M.; Mantle, P. G. Paxilline biosynthesis by
Acremonium loliae; a step towards defining the origin of
lolitrem neurotoxins. Phytochemistry 1987, 26, 969-971.
Received for review J une 28, 1995. Revised manuscript
received October 16, 1995. Accepted May 7, 1996.X
J F950396B
Penn, J .; Mantle, P. G. Biosynthetic intermediates of indole-
diterpenoid mycotoxins from selected transformations at
C-10 of paxilline. Phytochemistry 1994, 35, 921-926.
X Abstract published in Advance ACS Abstracts, Au-
gust 1, 1996.