Communications
Table 2: Synthesis of nucleosides in batch and continuous flow.
We discovered that both the pyridinium cation and triflate
counterion play essential roles in this process. Other pyridi-
nium salts derived from alternative acids failed to show any
conversion under these conditions (Table 1, entries 7–10).
Moreover, triflate salts derived from alkylamines afforded
inferior results; for example, ethyldiisopropylamine provided
3a in only 22% yield (Table 1, entry 11).
For practical reasons the easy-to-handle salt of 2,6-di-tert-
butyl-4-methylpyridine was preferred, but for large-scale
preparative experiments (see below) we would suggest the
use of less expensive and commercially available pyridinium
salts, such as pyridinium triflate itself.
It should be noted that even though TMSOTf gave
satisfactory yield under microwave conditions (Table 1,
entry 2), the flow variant was not feasible to run due to
decomposition of the ribofuranose and subsequent clogging
of the reactor (see Supporting Information for details).
Moreover, even with the use of 1 equivalent of pyridinium
triflate, no conversion was observed after 20 h under other-
wise identical batch conditions at room temperature, indicat-
ing that the high temperature provided by flow synthesis
would be necessary for this new catalyst system to be of any
practicality.
A significant challenge of converting the batch conditions
into a continuous flow process related to the poor solubility of
persilylated heterocyclic bases in CH3CN and required care-
ful design of the reactor setup.[11] We found it was necessary to
heat the bases (2a–i) to 408C prior to flowing them into the
reactor in order to prevent their precipitation. In addition, the
T-mixer where both reagent streams meet had to be held at
100–1508C (the reaction temperature) in order to prevent
solid formation and subsequent clogging (see the Supporting
Information for details of the reactor setup).
Base
Method[a]
4a
(mol%)
T [8C]
t[b]
[min]
Yield[c]
[%]
1
2
3
4
5
6
7
8
9
2a
2b
2c
2d
2e
2 f
2g
2h
2i
MW
flow
MW
flow
MW
flow[d]
MW
flow[d]
MW
flow
MW
flow[d]
MW
flow[d]
MW
flow
MW
flow
5
5
5
5
5
5
10
10
5
5
5
10
5
15
5
5
5
5
150
150
120
120
120
120
100
100
120
120
120
100
150
120
150
150
150
150
3
5
1
1
0.5
0.5
10
10
1
3
5
20
10
5
5
5
3
5
99
96
99
99
96
99
85 (10, 3)
87 (9, 0)
99
99
85 (12, 3)
88 (11, 0)
83 (11, 0)
80 (6, 14)
95[e]
95[f]
99
95
[a] MW method: 0.2 mmol of 1, 0.22 mmol of 2a, 1.5 mL of CH3CN,
catalyst in 0.5–2 mL MW vial (Biotage Initiator). Flow method: Flow
reactions were run in 100 mL PFA (perfluoroalkoxy), 0.75 mm i.d. tubing
reactor unless specified otherwise. [b] Time of MW reaction refers to
hold time (temperature ramp time ca. 1 min and cooling time ca. 45 s);
time of flow reaction refers to residence time in the PFA tubing reactor.
[c] Yield of isolated product, numbers in parenthesis refer to N1,N3-
bisnucleoside by-product and recovered ribofuranose, respectively.
[d] 100 mL PFA, 0.5 mm i.d. tubing reactor. [e] Mixture of isomers
N9:N7 4.5:1. [f] Mixture of isomers N9:N7 4.7:1.
The results of preparative scale batch and continuous flow
experiments are shown in Table 2. The reaction tolerates a
variety of nucleobases including 5-substituted uracils (2a–f),
cytosine (2g), and the purine bases guanine (2h) and adenine
(2i). Notable features of the reaction include low catalyst
loadings (typically 5 mol%), short reaction times (0.5–
20 min), and high yields following off-line purification (80–
99%, with six out of nine examples ꢀ 95%).
Differences in the required reaction time for the various
nucleobases can be attributed to reversible s-complex
formation between the silylated bases and the pyridinium
catalyst (weak Brønsted acid).[12] This trend is clearly seen in
the series of substituted uracils 2a–c where reaction time and
temperature can be correlated to the basicity of the pyrimi-
dines (2a > 2b > 2c). In the case of uracils with small
substituents in the 5-position (Table 2, entries 4 and 6) and
cytosine (Table 2, entry 7), N1,N3-bisnucleoside by-products
were observed in both batch and flow experiments. It was also
found that the tubing diameter had significant impact on the
formation of bisalkylation by-product.
Interestingly, we observed that when tubing with a larger
inner diameter (0.75 mm) was used, the amount of bisalky-
lation by-product increased (and correspondingly the yield of
the desired product decreased) when compared to the related
batch experiment. These anomalous results could be over-
come by simply employing a reactor with a smaller inner
diameter (0.5 mm) which promotes a more uniform laminar
flow profile and hence increased accuracy of the residence
times (see Supporting Information for details).[13] In the case
of cytosine, guanine, and adenine (2g–i), residual bis(trime-
thylsilyl)acetamide (BSA) from the silylation step had no
negative influence in the glycosylation reaction.
To demonstrate the scalability of the reaction we directly
transferred the optimized conditions to the commercially
available flow system from Vapourtec (Scheme 1).[14] This
reactor comprises two independent HPLC pumps that deliver
substrate solutions from reservoirs into the reactor. Mixing
occurs in a T-joint connector and the resulting solution is then
pumped into a 2 mL PFA tubing (1 mm i.d.) reactor that is
heated in a convection chamber.
By maintaining all the key parameters from the small-
scale flow reactor (concentration, catalyst loading, residence
time, and temperature), we obtained 26 g of 3b (7.4 ghÀ1) in
3.5 h, without any further optimization. The yield of 3b
2156
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 2155 –2158