L. Johns et al. / Phytochemistry 53 (2000) 607±611
609
The remaining chiral centre in I, II and indeed III±
V, is at C-7, which is not amenable to study by the
foregoing oxidative approach. However, this could be
partly de-lineated by examination of 22-S and 22-RS
samples of II by 13C NMR using chiral shift reagents
in conjunction with a soluble silver b-diketonate com-
plex, using a re®nement of a previous method (Wenzel
& Sievers, 1982). In order to test the validity of the
method, two model compounds were ®rst investigated.
Thus, all the individual 13C resonances of myrcene
(VII), which contains mono-, di- and tri-substituted
alkene moieties (cf II), were shifted ꢀDd 11:2 to
1.3 ppm) upon addition of praseodymium tris[3-hep-
ta¯uoropropylhydroxymethylene)-(+)-camphorate/AgI
6,6,7,7,8,8,8-hepta¯uoro-2,2-dimethyl-3,5-octanedio-
nate (Pr(hfc)3/Ag(fod)) to the solution. This veri®ed
that coordination of the shift reagent to all three
double bonds via the silver b-diketonate had taken
place. In the absence of the AgI b-diketonate, no res-
onance shifts were observed, as expected. Similar res-
onance changes were observed when 3RS-methylpent-
1-ene (VIII) was employed as the model substrate
ꢀDd 16:6 to 0.8 ppm). Further, the in¯uence of
the chiral shift reagent resulted in spectroscopic separ-
ation of the two enantiomers of VIII ꢀDdꢀRS 0:1 to
0.3 ppm). Thus, both the dynamic coordination of the
shift reagent to a tri-unsaturated alkene and separation
of alkene enantiomers were veri®ed. When the method
was applied to a 22-RS sample of II, coordination of
the shift reagent was con®rmed by the observation of
shifts of similar magnitude ꢀDd 14:2 to 3.6 ppm)
to those seen for the two model compounds. However,
no further splitting of resonances was observed,
suggesting that C-7 must be homochiral (R or S ) and
not racemic (RS ). Finally, Pr(hfc)3/Ag(fod) also com-
plexed to the 22-S isomer of II ꢀDd 2:5 to 2.1
ppm), and again no enantiomeric splitting was
observed, again suggesting homochirality at C-7. In
the absence of any chiral shift reagent, the Dd between
resonances observed for diastereomeric 22RS alkene II
was of the same order (0.1±0.4 ppm) as that observed
for the induced resonance shifts caused by complexa-
tion of the shift reagent with the 3RS-methylpent-1-
ene.
before R (Fig. 2). GC studies of alkenes from numer-
ous Haslea cultures then con®rmed the variability in
the con®guration at C-22 in I±III with S, R and RS
con®gurations produced in HBIs isolated from ran-
domly sampled cultures (Fig. 2). Whilst these ®ndings
require further investigation, the stereochemical studies
may aid elucidation of the biosynthetic pathways to
the sesterterpenoids. For instance, others have
suggested (Ourisson & Nakatani, 1994) that a plausible
precursor to the alkenes is the diphosphate ester of a
C25 pentaunsaturated alcohol (e.g. VI). Whilst other
precursors are possible, it is interesting that loss of
diphosphate from VI or other similar esters, would
produce hexaene V and progressive saturation would
yield IV±I, all of which we have identi®ed previously
in H. ostrearia (Belt et al., 1996; Johns et al., 1999;
Wraige et al., 1999). Saturation at C-22 would give
rise to 22RS (non-stereospeci®c), or by stereospeci®c
reduction to S or R isomers (IV) whereas reduction of
the 9(10) bond in II to I would involve a stereospeci®c
step leading to the 10S con®guration reported above.
In comparison with stereospeci®c saturation of other
acyclic isoprenoids, the con®guration at C-10 in I (viz
corresponding to a 3S con®guration of the tetrahydro-
geranyl group) is contrasts with the observed R stereo-
chemistry of the C-3, (and 7, 11 positions) in phytanyl
(3,7,11,15-tetramethylhexadecyl-) moieties of the gly-
We acknowledge that it is possible that the reson-
ances of the isomers of the HBI alkene were still not
separated under these conditions, but we consider it
unlikely given the results of our experiments with
model compounds such as myrcene and methylpen-
tene. The 7 (R or S ), 10S, 22S and 22RS con®gur-
ations reported above were established by these
somewhat laborious methods for three samples of I
and II from two dierent diatom cultures. However,
non-enantioselective GC (Carbowax) revealed that
22RS diastereomers of these authenticated compounds
could be separated and that the elution order was S
Fig. 2. Partial gas chromatogram (carbowax) indicating separation
of C-22 diastereomers of HBI triene (II) isolated from dierent cul-
tures of Haslea ostrearia. Chromatographic conditions are given in
the text.