C. Klinchan et al. / Tetrahedron Letters 55 (2014) 6204–6207
6205
2.00 Å between N3 and the hydrogen atom of the b-hydroxyl group
stayed within the range of hydrogen bonding. This interaction is
thought to pull the adenine ring toward the b-hydroxyl group in
the optimized structure of the 20-ketone hydrate (Fig. 2, panel A).
On the other hand, the optimized structure of the 30-ketone
hydrate did not show any hydrogen bond interaction between
the adenine base and the ribose (Fig. 2, panel B). Presumably, the
observed internal hydrogen bonding between N3 and the b-hydro-
xyl group at the C20 position in the 20-ketone hydrate might be the
driving force for the formation of the ketone hydrate at this posi-
tion, while the 30-isomer did not benefit from the same type of
stabilization.
Figure 1. The non-hydrolyzable substrate analogs of GatCAB.
azide. In addition, previous reports in which bromoadenosines
were successfully utilized as electrophiles are limited to nitrogen
or oxygen nucleophiles, which reacted intramolecularly to afford
cyclic intermediates.11,12 We next shifted our attention to another
double inversion strategy by utilizing an oxidation/reduction/sub-
stitution reaction sequence to first alter the stereochemistry of the
20-/30-hydroxyl groups and then to replace the activated hydroxyl
groups with a nitrogen nucleophile (Scheme 2).
The reduction of compound 4 with excess sodium triacetoxy-
borohydride, generated in situ from NaBH4 and AcOH at 0 °C, pro-
vided the ribo (minor) and xylo (major) diastereomers, as the
hydride preferentially attacked the less sterically hindered
a face.
On the other hand, when similar conditions were applied to com-
pound 11, only the desired stereoisomer was obtained in high
yield. In this case, the stereoselectivity was presumably induced
by the steric hindrance of the adenine base located adjacent to
the reaction center. After simple aqueous work-up, compounds 5
and 12 were used in the subsequent step without further purifica-
tion. It is worth mentioning that the small amount of the undesired
ribo isomer was eliminated during product purification in the next
synthetic step.
The synthesis started with partial protection of the adenosine
hydroxyl groups using tert-butyldimethylsilyl chloride to give the
20,50- and 30,50-bis-O-tert-butyldimethylsilyladenosines in 45% (2)
and 38% (3) yields, respectively.13 The inversion of the configura-
tions at the C20 and C30 carbons on the ribose moiety was carried
out by an oxidation/reduction sequence. Initially, the oxidation
was envisaged through the reaction between protected nucleo-
sides and chromium trioxide (CrO3) as previously reported.14–17
In our case, however, the reaction proceeded with low product
yield when the reported protocol was conducted, presumably
due to the formation of a chromium-nucleoside complex, as sug-
gested previously.14 In addition, the chromatographic step was
problematic, presumably because the chromium–nucleoside com-
plex became trapped on the silica gel. It is worth mentioning that
the same observation has been noted previously.18 These compli-
cations impeded significantly the large-scale preparation of the
keto-adenosine intermediates. With the problems related to chro-
mic acid oxidation, along with its toxicity, alternative oxidants
were investigated. The first alternative was explored using the
Pfitzner–Moffatt (DMSO/DCC) oxidation protocol.19 In this case,
the desired product was generated in low yield. We then moved
to the use of 2-iodoxybenzoic acid (IBX).20,21 Unfortunately, the
oxidation with this reagent did not proceed cleanly and the desired
product was obtained in low yield. Finally, Dess–Martin period-
inane (DMP) oxidized effectively both compounds 2 and 3 to afford
compounds 4 and 11, respectively, in good yields. The reactions
were thus practical for scale-up, which certainly facilitated the
subsequent modification steps.15,22
For activation of the hydroxyl group at the 30 position of the
20,50-bis-O-tert-butyldimethylsilyl adenosine, the previously
reported conditions employing triflic anhydride (Tf2O) resulted in
a low product yield.23 When the 20-hydroxyl isomer 12 was sub-
jected to Tf2O in the presence of three equivalents of DMAP, we
obtained the corresponding triflate 13 in good yield (64%). We also
observed that the triflation reaction was sensitive to the solvent.
No reaction was observed when THF was substituted for methy-
lene chloride. Nevertheless, compound 5 failed to give the triflate
product under these conditions. Instead, a smooth and rapid con-
version of compound 5 into the triflate 6 was achieved using TfCl.
In addition, we found that the reaction time could be shortened to
less than one hour. Triflates 6 and 13 were then subjected to nucle-
ophilic substitution with NaN3 in DMF to afford azides 7 and 14 in
good to excellent yields.
Deprotection of the silyl groups in compounds 7 and 14 was
performed using NH4F in MeOH at 60 °C.24 The resulting alcohols
8 and 15 were obtained in good yields. Subsequently, the azido
alcohols 8 and 15 were reduced using Pd/C and hydrogen gas.
Without further purification, the resulting amines were directly
coupled with the corresponding Cbz-protected L-amino acid deriv-
atives to provide amide products 9A, 9B, 16A, and 16B in moderate
yields. It is worth mentioning that the amino group on the adenine
ring posed no threat to the coupling reaction. The final step
involved hydrogenolysis to obtain the desired products (10A,
10B, 17A, and 17B), which were conveniently purified using C-18
reverse phase chromatography.
In conclusion, non-hydrolyzable substrate analogs for Asp-
tRNAAsn/Glu-tRNAGln amidotransferase (GatCAB) were synthesized
with overall yields from 2% to 14%. The synthetic route reported
herein does not require protection/deprotection of the amino
group on the adenine moiety, and Dess–Martin periodinane was
proven to be a suitable oxidant for the conversion of both the 20-
and 30-hydroxyl groups of the ribose into the keto-adenosine inter-
mediates, and applicable for large scale synthesis. The DFT calcula-
tions suggest the origin of the formation of the 20-ketone hydrate
via the internal hydrogen bond network, which was not observed
in the 30-keto isomer. Investigations on the interactions between
the GatCAB enzyme and these non-hydrolyzable substrate analogs
are in progress.
It is interesting to note that the 20-ketoadenosine derivative 11
exists as an equilibrium mixture of the ketone and its hydrate
form, which was spectroscopically confirmed using high-resolu-
tion mass spectrometry and NMR spectroscopy. The equilibrium
was not observed with the 30-ketoadenosine derivative 4, which
existed as a stable ketone. Our results agreed with a previous
report in which the oxidation was conducted using the Pfitzner–
Moffatt reagent.19
Due to these intriguing observations, we turned to computa-
tional chemistry for possible explanations. The DFT calculations
using the B3LYP level of theory were set up for 20- and 30-keto-
adenosine derivatives (both the ketone and hydrate form). The silyl
protecting group was omitted in order to simplify the calculations.
Although the total energy of the 20- and 30-ketone hydrates was not
dramatically different, the geometries of these structures revealed
an interesting piece of evidence. The N3 atom of the 20-ketone
hydrate was located very close to the hydroxyl group on the b-face
at C20 (the b-hydroxyl group) of the ribose ring. The distance of