H.D. Boswell et al. / Phytochemistry 52 (1999) 871±878
875
(Hashimoto, Nakajima, Ongena & Yamada, 1992), the
2.3. The esteri®cation of the products of the reductase
activities by tropine:- and pseudotropine: acyl-CoA
transferases
TRI and TRII enzymes extracted from B. candi-
da  aurea show a considerable ¯exibility of activity,
reducing a range of model compounds lacking the tro-
pane skeleton (Table 2). Similarly, N-ethylnortropi-
none, N-(2-¯uoroethyl)nortropinone and N-iso-
propylnortropinone were all found to be good sub-
strates for both reductases in vitro (Table 2).This is
consistent with the metabolism of these analogues in
vivo. While at ®rst sight the observed in vitro activity
with TRII is not consistent with the virtual absence of
N-alkylnortropan-3b;-ol products in the fed cultures, it
must be borne in mind that in all cultures from the
tribe Datureae (containing Brugmansia and Datura )
which have been analysed, the activity of TRI greatly
exceeds that of TRII. In the present experiments, TRI
exceeds TRII in D. stramonium by ca. 25-fold
(Portsteen et al., 1992) and in B. candida  aurea by
ca. 8-fold. Therefore a preponderance of the products
with the 3a con®guration is likely to be favoured, as is
observed to be the case in vivo. Thus, the overall prop-
erties of the enzymes Ð substrate speci®city combined
with speci®c activity Ð are broadly consistent with the
observed in vivo accumulation of products.
The N-alkyltropan-3-ols produced endogenously by
the reduction of N-alkylnortropinones undergo con-
siderable further metabolism (Boswell et al., 1999).
While the major products are the aromatic phenyllac-
toyl and tropoyl esters, signi®cant amounts of the ali-
phatic acetyl and tigloyl esters accumulate. Thus, as
with reduction, esteri®cation mimics the metabolism of
the natural compounds. This metabolism has also been
examined in vitro at the enzymatic level.
Two separate acyl transferase activities have been
identi®ed in D. stramonium (Rabot et al., 1995; Robins
et al., 1991, 1994) that transfer an acyl unit from an
acyl-CoA thioester to tropine (TAT) or pseudotropine
(PAT) respectively. As with TRI and TRII, these
enzymes show absolute speci®city in vitro for the endo/
3a and exo/3b con®gurations respectively but consider-
able ¯exibility in the structure of the substrate (Rabot
et al., 1995; Robins et al., 1991). N-Substituted tropan-
3-ols and norpseudopelletierines, the products of TRI
and TRII on the respective ketones, were tested in
vitro as substrates for the B. candida  aurea activities
(Table 3). Substrate solutions, containing a mixture of
both the 3a- and 3b-ols were incubated with a partially
puri®ed extract of B. candida  aurea roots containing
both TAT and PAT activities. Products were separated
and quanti®ed by GC and their identities determined
by GC/MS. TAT showed a very constrained substrate
speci®city range, only N-ethyltropan-3a-ol (besides tro-
pine itself) proving to be a good substrate. In contrast,
all three N-substituted tropan-3-ols proved to be good
substrates for PAT, 3b-tigloylated and 3b-acetylated
products being found for each substrate.
The observed properties of these acyl transferases in
vitro with N-alkylnortropinones are broadly consistent
with the accumulation of acetyl- and tigloyl-esters
observed following in vivo feedings. With tropine and
pseudotropine, tigloyl-CoA proved the better substrate
in vitro, as found previously for the D. stramonium
enzymes (Rabot et al., 1995; Robins et al., 1991,
1994), although a constant ratio was not observed.
Nevertheless, in vivo, the spectrum of alkaloids (both
natural and analogues) is dominated by the acetyl pro-
ducts (Boswell et al., 1999) and with all three N-substi-
tuted tropan-3-ols, only a single acetyl ester product
was found. On the basis of the PAT and TAT enzy-
matic data alone, this might be expected to be 3b-iso-
mer. In vivo, however, the tropine:pseudotropine ratio
is very high and the normal ratio of acetyltropine:ace-
tylpseudotropine is large. Thus, by analogy, the acetyl
ester products are much more likely to be the 3a-iso-
mers (see Boswell et al., 1999). Tigloylated products, in
contrast, were found with both 3a- and 3b-con®gur-
The norpseudopelletierines, in contrast, are found to
be poor substrates for both TRI and TRII (Table 2).
This observation is consistent with the lack of metab-
olism of cadaverine beyond norpseudopelletierine in
vivo. It is not, however, consistent with the eective
metabolism of pseudopelletierine and N-ethylnorpseu-
dopelletierine to a mixture of 3a-ol and 3b-ol products
which was observed in vivo (see Tables 4 and 6 of
Boswell et al., 1999). This apparent contradiction
between in vivo and in vitro ®ndings might be
explained either by the possibility that the time-inte-
grated activities, though low, are nevertheless sucient
to account for the conversions in vivo or, alternatively,
by the possibility that the enzymes of the B. candi-
da  aurea hybrid (in which the in vivo work was car-
ried out) show
a
somewhat broader substrate
speci®city than those of D. stramonium.
None of the N-alkylnortropinones showed signi®-
cant inhibitory activity with TR I or TR II when incu-
bated at 1 mM in the presence of 5 mM tropinone
(Portsteen, 1994). As the Km values (Table 2) are at
least double those for the endogenous substrate, this is
not unexpected. In the in vivo experiments (Boswell et
al., 1999), however, substrates were fed at 1 mM, a
concentration likely to have given an intracellular con-
centration well in excess of that of tropinone, which is
usually present at only low levels (probably due to
high levels of TRI). The lack of inhibition of TRI by
analogues is consistent with the observed lack of inhi-
bition of the accumulation of normal alkaloids in cul-
tures fed with these analogues (Tables 5 and 7 of
Boswell et al., 1999).