S. Furumochi et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
3
A
B
O
O
O
O
a
b, c
d
OH
OH
O
EtO
EtO
EtO
I
5
5
5
5
O
17
20
18
21
19
9
O
f
e
EtO
OH
SCoA
5
5
O
O
O
j
g
k
h, i
MeO
HO
OH
MeO
OH
6
6
6
26
25
27
O
O
OH
SCoA
6
6
28
10
Figure 2. Syntheses of (A) (E)-pentadeca-12,14-dienoyl CoA (9) and (B) hexadec-15-ynoyl CoA (10). (a) BH3ÁTHF, rt, 5 h; (b) TEMPO, NaOCl/CH2Cl2, rt, 1.5 h, 67%; (c) CHI3,
CrCl2/1,4-dioxane, rt, 1.5 h, 30% in 2 steps; (d) vinylboronic acid pinacol ester, Pd(dppf)Cl2, Ag2O/10:1 THF–H2O, rt, 2.5 h, 82%; (e) 1.0 M aq NaOH, rt, 5 h; 1.0 M aq HCl, 81%; (f)
CoA, PyBOP, K2CO3/7:3 THF–H2O, rt, 4 h, 59%; (g) 0.6 M trimethylsillyldiazomethane/THF, rt, 30 min; (h) TEMPO, NaOCl/CH2Cl2, rt, 30 min; (i) dimethyl (1-diazo-2-oxopropyl)
phosphonate/MeOH, rt, 8 h, 23% in 3 steps; (j) 1.0 M aq NaOH, 4 h; 1.0 M aq HCl, 46%; (k) CoA, PyBOP, K2CO3/7:3 THF–H2O, 83%.
(calculated for C74H106N5O19 as [MÀH]À: m/z 1368.7488, Figs. 3B
and S11) but was not detected by a fluorescence detector con-
nected to LC (LC-FLD). Two major peaks were clearly apparent in
the LC-FLD chromatogram and were attributed to DMEQ-TAD
(closed arrow, Fig. 3B) and the compound 24 (open arrow,
Fig. 3B) from Diels–Alder reaction of the carboxylic acid 21 and
DMEQ-TAD. This observation suggested that most of the acyl CoA
9 was hydrolyzed in situ to the carboxylic acid 21 and functioned
as the dienophile (Figs. 3B and S12). Many minor peaks were also
apparent in the LC-FLD chromatogram. DMEQ-TAD might have
undergone nucleophilic attack by biological compounds, thus pro-
viding many undesired products that hampered identification of
the Diels–Alder adduct.
As shown in the present study, fluorescent derivatives of bio-
logical compounds, including proteins, peptides, small molecules,
have been used to evaluate their mode of action. This trend
towards the use of fluorescent derivatives has been accompanied
by the emergence of chemical biology, in which a bioorthogonal
click reaction between an azide unit and an alkyne unit is a key
reaction for labeling molecules of interest. Like phosphorylation
and dephosphorylation, acylation is an important modification
of biological compounds and enables binding to proteins and
promotes signal transductions.17,18 For example, cholesterol
acylated at the 3b-OH binds to low/high density lipoproteins
(LDL/HDL),19 and all-trans-retinol (vitamin A) must be acylated
for conversion to 11-cis-retinol.20 Chemical biology has also
targeted lipid metabolism to provide various chemical probes.21
We therefore examined hexadec-15-ynoyl CoA (10)22 and
15-azidopentadecanoyl CoA (11)23 as alternatives to (E)-pentadeca-
12,14-dienoyl CoA (9) as the substrate in the in vitro acylation
reaction. 15-Azidopentadecanoyl CoA (11) was synthesized
according to the literature23 and hexadec-15-ynoyl CoA (10)
was synthesized via a new route as follows (Fig. 2B). Hexadec-
15-ynoic acid (25) was converted to methyl ester 26 with
trimethylsilyldiazomethane. The methyl ester 26 was oxidized
with TEMPO to the corresponding aldehyde, which was reacted
with Ohira–Bestmann reagent24 to afford hexadec-15-ynoic acid
methyl ester (27), then 27 was hydrolyzed under basic condi-
tions to give hexadec-15-ynoic acid (28). The carboxylic acid
28 was converted to acyl CoA 10 in the presence of CoA, PyBOP
and K2CO3.
Hexadec-15-ynoyl CoA (10) and 15-azidopentadecanoyl CoA
(11), together with hexadecanoyl CoA (7), were used as substrates
for the in vitro acylation reaction and converted OA (1) to the cor-
responding DTX3 29, 30 and 15 in 0.43%, 0.10% and 0.31% yield,
respectively. The relative yields of 29 and 30 were 1.39 and 0.32,
respectively (Table 1). The relative yield of 29 was comparable to
that of 15, presumably because the acyl CoAs 7 and 10 have the
same carbon chain length (C16), and unsaturation in the carbon
chain of the acyl CoA may not affect enzymatic activity. On the
other hand, the relative yield of 30 (C15) was about one third that
of 14 (C14) or 15 (C16). If an azide unit is counted as being three car-
bons in size, then acyl CoA 11 can be considered as a mimic of acyl
CoA 8 because the acyl chain of 11 is composed of 18 atoms (15 C
and 3 N). In fact, the yield of DTX3 (16) from OA (1) and acyl CoA 8
was comparable to the yield of DTX3 (30) from OA (1) and acyl CoA
11. As described earlier, DMEQ-TAD might have acted as an accep-
tor during Michael reaction for endogenous compounds and left
un-identified fingerprints in the LC-FLD chromatogram. In contrast,
the alkyne unit in 29 and the azide unit in 30 should react specif-
ically according to the intended click chemistry. Thus, the acyl
CoAs 10 and 11 would be favored over the acyl CoA 9.
The detection of 23 showed that the sensitivity of MS is superior
to that of FLD for this family of compounds. An alternative assay for
DTX3 (3) using FLD should provide increased efficiency of the
in vitro acylation reaction. In fact, Suzuki et al. reported that the
ratio of the acylated form (DTX3, 3) to the non-acylated form
(OA (1) or DTX1 (2)) was more than 10 in scallops.8 We believed
that the in vitro acylation reaction of OA (1) was not optimized;
indeed, the administered acyl CoA was hydrolyzed in situ and thus
unavailable as the substrate for the OA-acylating enzyme. We
therefore attempted to prevent hydrolysis of the acyl CoA to
improve the efficiency of the in vitro acylation reaction. However,
the inhibitors for acyl CoA hydrolase tested (clofibrate,25 gemfi-
brozil,26 b-NADP,27 and ADP28) did not increase the production of
DTX3 (3). It is possible that there is a physiological or ecological
factor(s) involved in the generation of OA-transforming enzyme
in the digestive gland of scallops. DTX3 (3) diminishes binding