Solution-phase preparation of a 560-compound library of individual pure mappicine analogues by fluorous mixture synthesis

Article abstract of DOI:10.1021/ja026947d

Solution-phase mixture synthesis has efficiency advantages and favorable reaction kinetics. Applications of this technique, however, have been discouraged by the difficulty in obtaining individual, pure final products by using conventional separation and

Full text of DOI:10.1021/ja026947d

Published on Web 08/08/2002
Solution-Phase Preparation of a 560-Compound Library of
Individual Pure Mappicine Analogues by Fluorous Mixture
Synthesis
Wei Zhang,*,† Zhiyong Luo,† Christine Hiu-Tung Chen,† and Dennis P. Curran*,‡
Contribution from Fluorous Technologies, Inc., 970 William Pitt Way,
Pittsburgh, PennsylVania 15238, and Department of Chemistry and Center of Combinatorial
Chemistry, UniVersity of Pittsburgh, Pittsburgh, PennsylVania 15260
Received May 17, 2002
Abstract: Solution-phase mixture synthesis has efficiency advantages and favorable reaction kinetics.
Applications of this technique, however, have been discouraged by the difficulty in obtaining individual,
pure final products by using conventional separation and identification processes. Introduced here is a
new strategy for mixture synthesis that addresses the separation and identification problems. Members of
a series of organic substrates are paired with a series of fluorous tags of different chain lengths. The tagged
starting materials are then mixed and taken through a multistep reaction process. Fluorous chromatography
is used to demix the tagged product mixtures on the basis of the fluorine content of the tags to provide the
individual pure components of the mixture, which are detagged to release the final products. The utility of
fluorous mixture synthesis is demonstrated by the preparation of a 560-membered library of analogues of
the natural product mappicine. A seven-component mixture is carried through a four-step mixture synthesis
(two one-pot and two parallel steps) to incorporate two additional points of diversity onto the tetracyclic
core. Methods for analysis and purification of the intermediates are established for the quality control of
the mixture synthesis.
Scheme 1. The Conceptual Basis of Fluorous Mixture Synthesis
Introduction
Mixture synthesis techniques increase efficiency in direct
proportion to the number of compounds mixed. Despite this
inherent efficiency, solution-phase mixture synthesis is not
considered to be a viable route to individual, pure target
molecules. Solid-phase mixture techniques (split-pool) are now
well established, but they mix beads (or containers), not
compounds.1 Solution-phase mixture techniques usually do not
target the isolation of the individual components and instead
use deconvolution methods to identify only the most active
compounds.2 These are then resynthesized individually.
Conducting reactions on mixtures of organic molecules is
straightforward. However, analyzing, identifying, and ultimately
separating the mixture components have until now seemed liked
insurmountable barriers. We suggest that these barriers can be
surmounted by a strategy based on separation tags.3 Substrates
are tagged with a series of different tags, and the tagged
substrates are mixed and taken through a series of reactions.
Central to this conceptual approach is the availability of at least
one separation technique that is targeted toward tag differentia-
tion. This technique can then be used at any time to demix
(separate based on tag) the mixture for analysis, identification,
and purification.
* To whom correspondence should be addressed. E-mail: D.P.C.,
curran@pitt.edu; W.Z., w.zhang@fluorous.com.
Fluorous mixture synthesis is the first solution-phase mixture
technique that allows the isolation of the individual pure
components at the end of the mixture exercise. Scheme 1
outlines the conceptual basis of fluorous steps. A series of n
initial substrates (S1 - Sn) are tagged with different fluorous
tags (F1 - Fn). The fluorous separation tags bear a homologous
series of fluoroalkyl groups. The tagged substrates (S1F1 - SnFn)
are mixed to give a single mixture (M1) that is then taken
through a series of mixture steps. The mixture steps can be either
one-pot reactions to give new mixtures (for example, M1 f
M2) or split-parallel reactions with a set of x building blocks
† Fluorous Technologies, Inc.
‡ University of Pittsburgh.
(1) (a) Lam, S.; Lebl, M.; Krchnak, V. Chem. ReV. 1997, 97, 411. (b) Furka,
A. In Combinatorial Peptide and Non Peptide Libraries; Jung, G., Ed.;
VCH: Weinheim, 1996; p 111. (c) Nicolaou, K. C.; Xiao, X. Y.;
Parandoosh, Z.; Senyei, A.; Nova, M. P. Angew. Chem., Int. Ed. Engl.
1995, 34, 2289.
(2) (a) Houghten, R. A.; Pinilla, C.; Appel, J. R.; Blondelle, S. E.; Dooley, C.
T.; Eichler, J.; Nefzi, A.; Ostresh, J. M. J. Med. Chem. 1999, 42, 3743. (b)
An, H.; Cook, P. D. Chem. ReV. 2000, 100, 3311. (c) Carell, T.; Wintner,
E. A.; Bashirhashemi, A.; Rebek, J. Angew. Chem., Int., Ed. Engl. 1994,
33, 2059. (d) Boger, D. L.; Chai, W. Y.; Jin, Q. J. Am. Chem. Soc. 1998,
120, 7220.
(3) Curran, D. P. Angew. Chem., Int. Ed. Engl. 1998, 37, 1175.
9
10.1021/ja026947d CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 10443-10450
10443

A R T I C L E S
Zhang et al.
Scheme 3. Plan of Mixture Synthesis
to generate x numbers of new mixtures (for example, M2 f
M13 - Mx3). The one-pot mixture reaction is an “internal
parallel” approach with no change of library size, while the split-
parallel mixture reaction proceeds in “double parallel” fashion
that has powerful library amplification capability. At the end
of the synthesis, all of the mixtures are separated chromato-
graphically over fluorous silica gel.4 This process is called
“demixing”, and each mixture containing n tagged products
(Px F1 - Px Fn) elutes in order of increasing tag size. The
chromatography serves both to separate the mixture components
(from each other and from other impurities) and to identify the
products. Detagging provides a total of nx number of final
1
n
1
n
products (P1 - Px ).
The natural product (S)-mappicine 1 was originally isolated
from Mappa foetida.5 Its analogue mappicine ketone 2, also
known as nothapodytine B,6 was isolated from Nothapodytes
foetida and is active on herpes viruses (HSV) and human
cytomegalovirus (HCMV) at a range of 3-13 µM.7 One of the
most efficient synthetic routes to the construction of the
mappicine skeleton is radical tandem cyclization of isonitriles
developed by Curran and co-workers.8 Incorporation of three
diversity points (R1, R2, R3) into the parallel solution-phase
synthesis of 64 mappicine analogues 3 has also been ac-
complished in the same group.9 Subsequent attempts to transfer
the synthesis to the solid phase, however, were abandoned after
extensive effort because the yields of the key reactions could
not be optimized to the levels needed for practical utility.10 Now
the problems associated with the solid-phase synthesis have been
successfully bypassed by a homogeneous fluorous solution-
phase synthesis. We report herein the first example of the use
of fluorous mixture synthesis to make a challenging natural
product library.
with fluorocarbon-bonded silica gel.12 Following this validation
of fluorous mixture synthesis at the proof-of-principle level, we
have now extended both the synthetic scale and the scope to
make a 560-membered library to demonstrate the practical utility
of fluorous mixture synthesis. The focal points of this work are
(1) to conduct a multistep library synthesis starting from a
mixture of seven components, (2) to assess the reactivity of
each component of the mixture in multistep mixture synthesis,
(3) to establish a general protocol to analyze intermediates and
to separate impurities in mixture mode, (4) to evaluate the
reliability of the semipreparative scale demixing with fluorous
HPLC columns, and (5) to produce milligram quantities of the
final product for appropriate characterization.
Results and Discussion
Mixture Plan. The plan for the synthesis of a 560-member
mappicine library is outlined in Scheme 3. Seven pyridinyl
(4) Curran, D. P. Synlett 2001, 1488.
(5) Govindachari, T. R.; Ravindranath, K. R.; Viswanathan, N. J. Chem. Soc.,
Perkin Trans. 1 1974, 1215.
(6) (a) Wu, T. S.; Chan, Y. Y.; Leu, Y. L.; Chern, C. Y.; Chen, C. F.
Phytochemistry 1996, 42, 907. (b) Pirillo, A.; Verotta, L.; Gariboldi, P.;
Torregiani, E.; Bombardelli, E. J. Chem. Soc., Perkin Trans. 1 1995, 583.
(7) (a) Pendrak, I.; Barney, S.; Wittrock, R.; Lambert, D. M.; Kingsbury, W.
D. J. Org. Chem. 1994, 59, 2623. (b) Pendrak, I.; Wittrock, R.; Kingsbury,
W. D. J. Org. Chem. 1995, 60, 2912.
Scheme 2
(8) (a) Curran, D. P.; Liu, H. J. Am. Chem. Soc. 1991, 113, 2127. (b) Josien,
H.; Curran, D. P. Tetrahedron 1997, 53, 8881. Works from other groups:
(c) Boger, D. L.; Hong, J. Y. J. Am. Chem. Soc. 1998, 120, 1218. (d)
Comins, D. L.; Saha, J. K. J. Org. Chem. 1996, 61, 9623. (e) Das, B.;
Madhusudhan, P. Tetrahedron 1999, 55, 7875. (f) Mekuoar, K. G., Y.;
Loue, S.; Greene, A. J. Org. Chem. 2000, 65, 5212. (g) Yadav, J. S.; Sarkar,
S.; Chandrasekhar, S. Tetrahedron 1999, 55, 5449. (h) Fortunak, J. M. D.;
Mastrocola, A. R.; Mellinger, M.; Wood, J. L. Tetrahedron Lett. 1994, 35,
5763. (i) Bowman, W. R.; Bridge, C. F.; Brookes, P.; Cloonan, M. O.;
Leach, D. C. J. Chem. Soc., Perkin Trans. 1 2002, 58.
(9) de Frutos, O.; Curran, D. P. J. Comb. Chem. 2000, 2, 639.
(10) Frauenkron, M.; Cheong, J. H.; Gabarda, A. E.; Curran, D. P. University
of Pittsburgh, unpublished results.
(11) (a) Luo, Z. Y.; Zhang, Q. S.; Oderaotoshi, Y.; Curran, D. P. Science 2001,
291, 1766. (b) For a full paper describing the quasiracemic synthesis of
mappcine and pyridovericin, see: Zhang, Q.; Rivkin, A.; Curran, D. P. J.
Am. Chem. Soc. 2002, 124, 5774.
(12) At the time of this preliminary work, preparative fluorous HPLC columns
were not available, and most of the mixtures were not preparative demixed
and detagged.
Preliminary work to construct a 100-membered tagged
mappicine library has recently been communicated.11a Four
tagged alcohols M-5 were mixed and carried through four steps
of mixture synthesis to provide tagged mappicines M-4 (Scheme
2). A total of 25 four-component mixtures were made, and the
mixtures were demixed analytically by an HPLC column packed
9
10444 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Library of Individual Pure Mappicine Analogues
A R T I C L E S
Scheme 4. Modified Synthesis of Iodopyridine 13
Scheme 5. Synthesis of Propargyl Bromides 14 and Isonitriles 15
Scheme 6. Preparation and Tagging of Alcohols 16
alcohols bearing different R1 groups are first attached to seven
different tags and then mixed together. The mixture M-5
undergoes two consecutive one-pot reactions to generate a
mixture of tagged pyridones M-7. The new mixture is then
divided to eight portions for N-alkylation with eight different
propargyl bromides in parallel. The resulting eight mixtures M-8
are each split into 10 portions for parallel free radical reactions
with 10 isonitriles to generate 80 mixtures of tagged mappicines
M-4. Each mixture containing seven components is then
demixed by HPLC to give a total of 560 individual, pure
mappicine analogues after detagging. This four-step mixture
synthetic process consists of two one-pot and two parallel
reactions, and allows combinatorialization of three sets of
building blocks. The yield for the mixture synthesis is calculated
on the basis of the average molecular weight of a compound
mixture.
Synthesis of Starting Materials, Building Blocks, and Tags.
The key starting material for the mixture synthesis is iodopy-
ridine 13. We quickly found that the previous small-scale
synthesis9 was not reliable on a larger scale. A major problem
was the deoxygenation of aldehyde 11, which gave only 30%
yield of 13 in gram-scale reactions (Scheme 4). After some
experimentation, we found that a two-step sequence involving
primary alcohol 12 reliably gave 70% overall yield of iodopy-
ridine 13 from aldehyde 11. Using the modified procedure, we
obtained a total of 65 g (202 mmol) of compound 13 from 2,6-
dibromopyridine.
Fluorous tags (25 mmol each) were prepared by addition of
the lithium reagent derived from RfCH2CH2I to diisopropyl-
chorosilane to give RfCH2CH2(iPr)2SiH. In earlier work, this
silane was converted in situ to the silyl bromide (with dibromine)
prior to reaction with the requisite alcohol. Later, an improved
procedure involving in situ generation of the silyl triflate (with
triflic acid) was developed. Of the seven silanes, the precursors
(RfCH2CH2I) with even number of perfluorocarbons (C4F9,
C6F13, C8F17, and C10F21) were purchased with the spacer in
place. The odd tags (C3F7, C7F15, and C9F19) were more readily
available without spacers (RfI), but addition to ethylene is
readily accomplished to provide the requisite iodides. The C5F11
tag is absent because the precursors (C5F11I and C5F11CH2CH2I)
were not available at the time of the synthesis.
Tagging and Mixture Synthesis. Seven alcohols 16{1-7}
(13 mmol each) were prepared individually by quenching the
Grignard reagent derived from iodopyridine 13 with seven
different aliphatic aldehydes (Scheme 6).13 Silyl triflates were
generated from fluorous silanes and used in situ to protect
alcohols 16{1-7} to give 5{1-7}. The tags in an order of
The three sets of building blocks needed for the diversity
plan are easily obtained. All seven aldehydes (R1CHO, 15 mmol
each) are commercially available. Eight propargyl bromides 14-
{1-8} (3 mmol each) were selected from commercial sources
or prepared from the corresponding alcohols (Scheme 5A).13
A
total of 10 aryl isonitriles 15{1-10} (8 mmol each) were prepared
from the corresponding anilines (Scheme 5B).13 Of the 10
isonitriles, nine have substituents at the para position of the aryl
ring to avoid potential formation of regioisomers in the radical
annulation reaction. The 10th, 2-fluorophenyl isonitrile 15{8},
has an ortho substituent that is expected to direct the final
cyclization to the other ortho position.
(13) Bom, D. Ph.D. Dissertation, University of Pittsburgh, 2000.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10445

A R T I C L E S
Zhang et al.
Scheme 7. Mixture Synthesis of Tagged Mappicines
increasing fluorine content were matched to the corresponding
alcohols in an order of decreasing polarity of the R1 side chain.
In other words, the shortest tag (C3F7) was paired with the most
polar alcohol side chain (Me), and the longest tag (C10F21) was
paired with the least polar alcohol side chain (CH2CH2-c-C6H11).
In this way, the best HPLC separation of the mixture will be
achieved by matching the primary separation factor (fluorine
content) of fluorous silica gel with a secondary separation factor
(polarity).14 Seven alcohols with different R1 substituents were
coded with silyl tags as follows: Me/C3F7, Pr/C4F9, Et/C6F13,
s-Bu/C7F15, i-Pr/C8F17, cyclohexyl/C9F19, cyclohexylethyl/
C10F21. We deliberately mismatched the tags for the propyl,
ethyl, s-butyl, and isopropyl groups to test if tag dominance
overwhelms side-chain polarity.
Equimolar amounts of the seven tagged compounds 5{1-7}
(6.5 mmol each) were mixed and subjected to iodonative
desilylation with ICl (Scheme 7). The reaction was closely
followed by F-HPLC equipped with a Fluofix column to
optimize the reaction conditions. We found that slow addition
of ICl to a solution of 5{1-7} at 15 °C under sonication increased
the conversion of the starting materials from the original 50%
to more than 90%.9 The crude product 6{1-7} was demethylated
with boron tribromide, and the resulting mixture of seven
pyridones 7{1-7} was purified by standard silica gel column
chromatography (see below for a discussion of this type of
purification). The purified mixtures were then split into eight
portions, and each portion (2.1 mmol) was subjected to
N-alkylation with 1.25 equiv of propargyl bromides 14{1-8}.
After flash chromatographic purification on silica gel, each of
the eight mixtures of N-propargyl pyridones 8{1-7,1-8} was
split into 10 portions (0.15 mmol each) and irradiated under a
sunlamp with 3.0 equiv of aryl isonitriles 15{1-10} and a
catalytic amount of hexamethylditin. These 80 crude mixtures
were purified by rapid solid-phase extraction (SPE) with normal
silica gel. Unreacted N-propargyl pyridones 8{1-7,1-8} and
isonitriles 15{1-10} and other byproducts were washed off with
10% EtOAc/hexanes, and fluorous-tagged mappicine mixtures
4{1-7,1-8,1-10} were eluted with 15% MeOH/EtOAc. All 80
Figure 1. HPLC analysis of tagged mappicine 4{1-7,4,3}. Fluofix column
(4.6 × 250 mm, 5 µm), gradient 90% MeOH-H2O to 100% MeOH in 15
min and then 100% MeOH. Top trace: UV detection at 254 nm. Bottom
trace: mass spectrometer detection with a positive APCI ionization source.
The first peak is the solvent front.
mixtures were analyzed by automated LC-MS before loading
onto a semipreparative HPLC column for demixing.
A typical LC trace for the analytical demixing of 4{1-7,4,3}
is shown in Figure 1 (UV detection, upper trace; MS detection,
lower trace). Besides the solvent front, seven well-resolved peaks
were detected in the UV channel in this and each of the other
79 mixtures. As revealed by mass spectroscopy, the seven
compounds are the expected tagged mappicines, which eluted
in the order of increasing fluorine content of the tag despite the
deliberate reverse of four of the side chains. The molecular ions
of all of the expected 560 fluorous-tagged mappicines were
detected by LC-MS. This is a powerful demonstration that the
elution order of the products can be predicted from the initial
fluorous tagging of the starting substrates. To illustrate the
analytical separation of the mixture of seven compounds, the
retention times of 56 out of the 560 tagged mappicines are listed
in Table 1. We noticed that all of the seven peaks of mixtures
4{1-7,1-8,4} and 4{1-7,1-8,7} eluted about 2 min slower than
the rest of the library components. This is due to the presence
(14) Curran, D. P.; Oderaotoshi, Y. Tetrahedron 2001, 57, 5243.
9
10446 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Library of Individual Pure Mappicine Analogues
A R T I C L E S
Table 1. Retention Times (min) of 56 of the 560 Tagged
Mappicines 4a
Scheme 8. Purification of 8{1-7,3} by Flash Column
Chromatographya
4{1-7,y,z}
C F
C F
C F
C F
C F
C F
C F
10 21
3
7
4
9
6
13
7
15
8
17
9
19
y,z ) {1,1}
{2,2}
3.5
4.0
3.6
5.3
4.6
4.4
6.9
3.7
4.5
5.3
4.6
7.0
6.0
5.7
8.7
4.8
6.7
7.8
6.7
9.8
8.8
8.3
11.7
7.1
8.7
10.0
8.6
12.0
11.0
10.4
13.8
9.1
10.7
11.8
10.7
13.9
13.3
12.4
15.6
11.2
12.3
13.3
12.2
15.2
14.9
13.8
16.8
12.7
15.7
16.4
15.5
18.0
19.7
16.8
19.8
15.9
{3,3}
{4,4}b
{5,5}
{6,6}
{7,7}b
{8,8}
a Fluofix column (4.6 × 250 mm, 5 µm), gradient 90% MeOH-H2O to
100% MeOH in 15 min and then maintain 100% MeOH for 5 min, flow
rate 1 mL/min. b Note longer retention times due to the CF3 group.
Table 2. Fluorous HPLC of Tagged Intermediate Mixturesa
5{1-7}
(min)
6{1-7}
(min)
7{1-7}
(min)
1
R
Rf
{1} Me
{2} Pr
{3} Et
{4} s-Bu
{5} i-Pr
C3F7
C4F9
8.4
12.5
20.0
25.3
30.8
33.3
39.4
4.6
6.8
3.6
5.1
8.6
12.8
17.5
20.8
28.5
C6F13
C7F15
C8F17
C9F19
C10F21
11.7
17.1
21.9
25.3
32.4
{6} c-C6H11
{7} C2H4-c-C6H11
a Fluofix column (4.6 × 250 mm, 5 µm), gradient 90% MeOH-H2O to
100% MeOH in 15 min and then maintain 100% MeOH, flow rate 1 mL/
min.
of the extra CF3 group in the A-ring of mappicine in these
mixtures. This body of structure-retention information is a
valuable aid for the design of fluorous mixture synthesis in the
future.
a Fluofix column (4.6 × 250 mm, 5 µm), gradient 90% MeOH-H2O to
100% MeOH in 15 min, then 100% MeOH. bO-Alkylated byproducts
17{1-7,3} are pointed by the arrows.
out by the arrows).15 This crude mixture showed only two spots
on a normal silica gel TLC plate: a high Rf spot of O-alkylated
mixture 17{1-7,3} (minor) and a low Rf spot of N-alkylated
mixture 8{1-7,3} (major). Silica gel flash column chromatog-
raphy with hexanes-AcOEt (80:20) thus separated the O-
alkylated byproducts as a mixture and gave N-propargylation
products 8{1-7,3} as a significantly purer mixture (Scheme 8,
bottom HPLC trace). The relative ratio of seven components
of 8{1-7,3} in the crude mixture had no significant change after
chromatography purification. The technique of silica gel flash
column chromatography was also used in the purification of
other intermediates (see Experimental Section). In principle,
whenever tagged products differ collectively from tagged
byproducts and other impurities in polarity, silica gel flash
column chromatography can be applied for purification in a
mixture mode.
Demixing and Detagging. Selection of an optimum column
is critical to the success of HPLC demixing. Three commercially
available analytical columns with perfluoroalkyl and perfluo-
roaryl stationary phases16 have been evaluated (Fluofix,
SiMe2C2H4C(CF3)2CF2CF2CF3; Fluophase-RP, SiMe2C2H4C6F13;
Fluophase-PFP, SiMe2C6F5). HPLC retention times for a typical
mixture of seven tagged mappicines 4{1-7,6,2} are shown in
Figure 2. For the purpose of comparison, the separation
result on a normal reverse-phase C18 column (Symmetry) is also
listed.
Intermediate Analysis, Characterization, and Purification.
Standard mixture synthesis without separation tags is difficult
because mixture components cannot easily be analyzed, char-
acterized, and purified. In fluorous mixture synthesis, fluorous
chromatography provides a powerful tool to monitor the mixture
synthesis because the fluorine-fluorine interaction dominates
the separation. The polarity and other features of the substrate
contribute a secondary effect to the separation.14 Table 2 lists
three sets of HPLC retention times obtained from different
mixtures of tagged intermediates. The polarity gradually in-
creases from trimethylsilyl pyridines 5 to iodopyridines 6 and
then to pyridones 7; the retention times of corresponding
molecules with the same tag are gradually decreased. However,
within each mixture family, the seven components kept the same
elution order and were well separated from each other. Those
purposely “mismatched” R1/Rf pairs (Pr/C4F9, Et/C5F11, s-Bu/
C6F13, and i-Pr/C7F15) did not cause compound overlapping or
crossing in the HPLC analysis.
Fluorous HPLC provides a tool to analyze intermediate
mixtures. Purification of tagged intermediates from tagged
byproducts, however, requires a nontag-based separation strategy
because the mixed components should not be demixed until the
end of the mixture synthesis. We discovered that standard silica
gel chromatography addressed the purification problem because
the separation is based on the polarity of the substrates and is
relatively insensitive to the attached tags. Scheme 8 shows an
example of purification of propargylation products. Fluorous
HPLC of the crude product indicated the presence of about 10%
byproduct mixture. On the basis of our previous experience,
these byproducts are O-propargylated pyridines 17{1-7,3}
(Scheme 8, top HPLC trace, O-alkylation mixtures are pointed
The Fluophase-RP and Fluofix columns have demonstrated
comparably good separation with a time window of 12 min
(15) Liu, H.; Ko, S. B.; Curran, D. P. Tetrahedron Lett. 1995, 36, 8917.
(16) New FluoroFlash fluorous HPLC columns (SiMe2C2H4C8F17) are now
available for both analytical and preparative scales from Fluorous Tech-
nologies, Inc., www.fluorous.com.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10447

A R T I C L E S
Zhang et al.
Figure 3. Semipreparative demixing of tagged mappicines 4{1-7,6,2}.
Fluophase-RP column (20 × 250 mm, 5 µm), gradient 88% MeOH-12%
H2O to 100% MeOH in 28 min, then to 100% THF in 7 min, flow rate 12
mL/min.
Scheme 9. Detagging and SPE Purification
about 0.07 mmol (45-65 mg) of seven-component mixtures
were dissolved in 300-350 µL of THF and were loaded onto
the column. One-half of 80 mixture samples were demixed on
a Waters HPLC system, and the fractions were collected
manually. The other one-half were demixed on a Gilson serial
HPLC with an automated fraction collector. The two methods
were equally successful, and 560 individually pure tagged
mappicines were obtained in an average amount of 0.005-0.01
mmol (3-6 mg) each. A typical chromatogram of demixing of
4{1-7,6,2} is shown in Figure 3.
Figure 2. HPLC separation of tagged mappicines 4{1-7,6,2}. Gradient 90%
MeOH-H2O to 100% MeOH in 15 min and then maintain 100% MeOH.
between C3F7 and C10F21 peaks under a standard HPLC
condition. The Fluophase-PFP column has low resolution on
C7F15 and C8F17 peaks, and the time window between C3F7 and
C10F21 peaks is only 8 min. The reverse C18 column does not
separate C7F15 and C8F17 tagged mappicines at all, presumably
because of the polarity mismatching of the tag and side chain
(s-Bu/C7F15 vs i-Pr/C8F17). Fluofix and Fluophase-RP are
fluorous columns that separate the mixture mainly on the basis
of fluorine content. On these columns, nonfluorous compounds
eluted out together with the solvent front. The pentafluorophenyl
group does not have a sheath of fluorines such as the perfluo-
roalkyl groups and is not considered to be fluorous.17 Indeed,
Fluophase-PFP behaved more like a normal reverse-phase C18
column than like a fluorous column. Its separation is sensitive
to the polarity (or size) of the substrates, and demixing is not
guaranteed by the fluorine content. On these reverse-phase
columns, nonfluorous byproducts can be separated to some
extent from the solvent front.
On the basis of separation results obtained from analytical
columns, a semipreparative Fluophase-RP column (20 × 250
mm, 5µm) was selected for demixing of 80 tagged mappicine
mixtures. A modified MeOH-H2O-THF solvent system was
used as the mobile phase. The strong fluorophilic solvent THF
was introduced at the end of each gradient program to facilitate
the elution of the molecules containing C10F21 tags and also for
column conditioning between injections. Samples containing
The fluorous silyl protecting groups were cleaved with HF-
pyridine in THF (Scheme 9). The short tags (C3F7 and C4F9)
can be easily detached within 1 h at 60 °C, while longer tags
required extended heating times up to 10 h. The crude products
were partitioned between AcOEt and H2O. The concentrated
organic layers were loaded onto reverse-phase SPE cartridges
and eluted with MeOH/H2O (80/20) or THF/H2O (50/50).
Mappicines eluted first. Cleaved tags (silanols) followed when
the cartridges were washed with MeOH or THF. The 560 SPEs
were conducted manually in groups of 24 in a parallel manifold.
Finally, the products were quantified by weighing, and the
amount of 560 mappicines was distributed as follows: 315
samples (56%) were between 1 and 2 mg, 180 samples (32%)
were less than 1 mg, and 65 samples (12%) were greater than
2 mg.
Product Characterization. The fluorous mixture synthesis
provides quality control at each reaction step by F-HPLC
analysis and flash column chromatographic purification. The
most critical purification occurs at the step of demixing because
that separates the mixture components and also offers product
purification. Because the structures of all 560 compounds were
characterized by LC-MS before demixing, the final analyses
were focused on the assessment of product purity. Several 19F
NMR analyses were carried out to ensure that there were no
tag residues after detagging and SPE. HPLC analysis with UV
detection of 112 (20% of the library) randomly selected samples
indicated that the product purity was greater than 90%.
(17) Gladysz, J.; Curran, D. P. Special issue on fluorous chemistry, Tetrahedron
2002, 58, 3823.
9
10448 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002

Library of Individual Pure Mappicine Analogues
A R T I C L E S
Additional product characterizations included MS/LC-MS and
1H NMR/LC NMR analyses (see Supporting Information).
All of the mappicine samples had the expected substituents
with one exception. HPLC analysis of those 56 mappicine
analogues 4{1-7,1-8,10} with the MeS- substituent at the A
ring gave more than one peak. MS analysis revealed that one
of the peaks was the expected sulfide, and the others (usually
one peak, sometimes two) were the sulfoxides as a mixture of
diastereomers. We suspect that detagging with HF-pyridine at
high-temperature conditions may have caused the oxidation.
Using a different desilylation agent such as CsF under milder
conditions might prevent the formation of sulfoxides.
structurally diverse, it will be helpful to give some thought to
tag/analogue pairings. This is because large differences in the
size and/or polarity of the tagged molecules could upset the
tag dominance.4 To spread rather than contract the fluorous
HPLC chromatogram, we tentatively suggest that smaller and/
or more polar analogues be given smaller fluorous tags and
larger and/or less polar analogues be given larger tags. More
sophisticated approaches may become possible as we learn more
about structure/retention effects on fluorous HPLC columns, but
a simple, qualitative approach may suffice for many applications.
As in any mixture synthesis technique, efficiency is maximized
by mixing early and demixing late.
The rapid transition of fluorous mixture synthesis techniques
from “proof-of-principle” experiments to practical applications
in several areas11b,18 bodes well for the wider adoption of this
technique. We suggest that it be considered for mainstream
synthesis of natural products and related molecules or for high-
throughput synthesis of drug candidates whenever leveraging
of a multistep synthetic exercise by making more compounds
is beneficial.
Conclusions
Fluorous mixture synthesis is a new solution-phase high-
throughput technique that allows the production of more
compounds from a synthetic exercise without a proportional
increase in effort. The broad scope and favorable reaction
kinetics associated with solution-phase synthesis are united with
the efficiency advantages of mixture synthesis. The use of
fluorous tags and the associated tag-based separation (demixing)
allows intermediate mixtures to be analyzed and characterized
and produces target products as individual, pure compounds.
The features of fluorous mixture synthesis are highly suitable
for synthesizing relatively small (100-1000) but high quality
optimization libraries for structure-activity relationship (SAR)
studies in medicinal chemistry or other chemical discovery
settings.
Experimental Section
Building blocks, tags, and other starting materials used in the
mappicine library synthesis were readily available. All seven aldehydes
(R1CHO) and two propargyl bromides (HCCCH2Br and MeCCCH2B)
were obtained commercially. The other six propargyl bromides (R2-
CCCH2Br), 10 isonitriles 15{1-10}(R3PhCN), perfluoroalkylsilanes
i
(RfCH2CH2 Pr2SiH), and alcohols 16{1-7} were prepared by following
known literature procedures.9,13
The practical utility of fluorous mixture synthesis has been
demonstrated by the synthesis of a mappicine library. A seven-
component mixture undergoes one-pot and split-parallel syn-
theses with two sets of building blocks to reach a size of 560
in four steps of mixture synthesis. The economy of the mixture
approach is readily illustrated by counting synthetic steps: 90
reactions (1 + 1 + 8 + 80) were conducted during the mixture
synthesis in Scheme 7, whereas 630 (7 + 7 + 56 + 560) would
be needed to conduct the same sequence in parallel. The savings
increase with the number of compounds mixed, with the length
of the sequence, and with the number of splits after mixing.
Comparable savings accrue in the separations; only 80 chro-
matographies were used in the final demixing to produce 560
pure samples.
Fluorous HPLC is a reliable separation method not only for
demixing of tagged products, but also for analysis of tagged
intermediates. Purification of tagged intermediates as mixtures
(that is, without demixing), however, requires a nontag-based
separation method. Fluorous tags are relatively nonpolar, and
experience is showing that normal silica gel flash column
chromatography can sometimes be used to purify tagged
mixtures on the basis of the polarity differences of subsets of
tagged molecules. This unexpected ability to go from impure
mixtures to pure mixtures without demixing adds to the
practicality of the technique.
General LC-MS Analysis Conditions. A Fluofix column (4.6 ×
250 mm, 5 µm, Keystone Scientific, Inc.) was used, with a gradient
90% MeOH-H2O to 100% MeOH in 15 min, then maintained 100%
MeOH for 5-20 min. Mass spectrometer detection was done with a
positive APCI ionization source. Similar conditions were applied to
F-HPLC analyses of intermediates in the mixture synthesis.
Modified Procedure for the Preparation of 13. To a solution of
aldehyde 11 (3.69 g, 11.0 mmol) in ethanol (20 mL) at -40 °C was
added NaBH4 (419 mg, 11.0 mmol). The reaction mixture was further
stirred for 1 h at -40 °C and then quenched with water. The crude
product was purified by silica gel column chromatography (10% EtOAc/
hexanes) to give alcohol 12 (2.60 g, 70%) as a colorless oil. To a
solution of alcohol 12 (18.40 g, 54.44 mmol) in 1,2-dichloroethane
(75 mL) was added triethylsilane (63.15 g, 0.54 mol) followed by slow
addition of boron trifluoride etherate (34.5 mL, 0.27 mmol) at room
temperature. The reaction mixture was then heated at 75 °C for 2 h
before quenching with aqueous NaHCO3. After being extracted with
diethyl ether, the organic layer was dried and passed through a silica
plug with hexanes to give the pure product 139 (15.8 g, 90%).
General Procedure for Tagging Alcohols 16{1-7} with Perfluo-
roalkylsilanes (RfCH2CH2(iPr)2SiH). Preparation of 5{3}. To C6F13-
CH2CH2(iPr)2SiH (9.88 g, 21.4 mmol) was added trifluoromethane-
sulfonic acid (2.05 mL, 16.4 mmol) at 0 °C. The reaction mixture was
then stirred at room temperature for 15 h. A solution of 16{3} (4.16 g,
16.4 mmol) and 2,6-lutidine (3.8 mL, 32.8 mmol) in dry CH2Cl2 (40
mL) was added. After being stirred at room temperature for 2 h, the
reaction mixture was quenched with aqueous NH4Cl and extracted with
CH2Cl2 and ether. The combined organic layers were dried, and
chromatography on silica gel with EtOAc/hexanes (5/95) gave 5{3}
(9.97 g, 85%) as a colorless clear oil. 1H NMR (300 MHz, CDCl3): δ
0.25 (s, 9H), 0.77 (dd, 2H), 0.88 (t, 3H), 0.97-1.08 (m, 14H), 1.67
(m, 2H), 1.72-1.98 (m, 2H), 2.10 (s, 3H), 4.91 (s, 1H), 7.21 (s, 1H).
The bounds of fluorous tagging for mixture synthesis are not
yet clear. The synthesis of multiple isomers by fluorous mixture
approaches promises to be reliable because the tags should easily
override the differences of isomers. The mappicine library
described herein shows that fluorous mixture synthesis will also
be useful for synthesizing related, nonisomeric analogues in a
logical series. As the analogues being mixed become more
(18) For a mixture synthesis of four truncated discodermolide analogues, see:
Curran, D. P.; Furukawa, T. Org. Lett. 2002, 4, 2233.
9
J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002 10449

A R T I C L E S
Zhang et al.
General Procedure for TMS-Iodo Exchange. Preparation of
6{1-7}. A mixture of seven tagged alcohols 5{1-7} (1.5 mmol each,
total 10.5 mmol, 8.2 g) in CH2Cl2/CCl4 (3:1, 80 mL) was sonicated at
15 °C. To this mixture was added a solution of ICl (5.1 g, 31.5 mmol)
in CH2Cl2/CCl4 (3:1, 55 mL) over 30 min via an addition funnel. The
reaction was followed by F-HPLC analysis to completion. The mixture
was washed with aqueous Na2S2O3, the aqueous layer was extracted
with ether, and the organic layer was concentrated to give 6{1-7} (8.9
g) as a clear yellow oil. This crude product I was used for the next
reaction without further purification.
General Procedure for Demethylation. Preparation of 7{1-7}.
To a solution of 6{1-7} (8.7 g of crude) in CHCl3 (150 mL) was added
BBr3 (7.9 g, 31.5 mmol) at room temperature. The reaction mixture
was then refluxed for 2.5 h. The reaction was followed by F-HPLC
analysis to completion. The cooled reaction mixture was slowly poured
into aqueous NaHCO3. The organic layer was separated, the aqueous
layer was extracted with ether, and the combined organic layers were
washed with brine and concentrated. The crude product was purified
by chromatography on silica gel with hexanes followed by hexanes/
AcOEt (9:1) to give 7{1-7} (6.50 g, 76% from 5{1-7}) as a clear brown
oil. The purity was checked by F-HPLC.
General Procedure for N-Propargylation. To a mixture of seven
pyridones 7{1-7} (0.30 mmol each, total 2.1 mmol) in DME (6 mL)
and DMF (2 mL) was added NaH (60% in mineral oil, 0.095 g, 2.40
mmol) at 0 °C followed by LiBr (0.37 g, 4.2 mmol) after 10 min. The
reaction mixture was warmed to room temperature. A propargyl
bromide 14{1-8} (3.1 mmol) was added. The mixture was heated at
75 °C for 7 h. The cooled reaction mixture was extracted with ether
and washed with brine. The concentrated organic layer was purified
by chromatography on silica gel with hexanes followed by hexanes/
AcOEt (8:2) to give 8{1-7,1-8} in an average yield of 70%. The purity
was checked by F-HPLC.
follows: Fluophase-RP column (20 × 250 mm, 5 µm, Keystone
Scientific, Inc.), gradient 88% MeOH-12% H2
O to 100% MeOH in
28 min, then to 100% THF in 7 min with a flow rate of 12 mL/min. A
mixture sample dissolved in a minimum amount of THF was injected.
The demixed fractions were collected by following the UV detector
signal (manually) or by an automatic fraction collector. The desired
fractions were concentrated to give 560 individually pure tagged
mappicines 4{1-7,1-8,1-10} (3-6 mg of each).
General Procedure for Detagging. Preparation of Individually
Pure Mappicines 3{1-7,1-8,1-10}. Detagging of 560 4{1-7,1-8,1-10}
was accomplished in parallel. To each vial containing a tagged
mappicine (3-6 mg) in THF (0.2 mL) was added 10 drops of HF-
pyridine at room temperature, and then it was heated at 60 °C. The
reactions were followed by TLC for completion. Reactions of shorter-
tagged (C4F9 to C6F13) mappicine analogues took 1 h to complete,
whereas longer reaction times (up to 10 h) for long-tagged (C7F15 to
C10F21) mappicine analogues were necessary for the reaction to go to
completion. Upon completion of the reaction, the mixture was diluted
with EtOAc, washed with aqueous NaHCO3, and the organic layers
were air-dried. Upon solvent evaporation, the residue from each vial
was subjected to solid-phase extraction on reverse-phase silica gel (0.5
g) packed into syringe cartridges of 2.5 mL volume. The residue was
dissolved in a minimum amount of 80:20 MeOH:H2O (several drops
of THF were sometimes added to aid in dissolving the residue) and
loaded onto the prewet (80:20 MeOH:H2O) SPE cartridge which was
set on a 12 × 2 SPE manifold. The first fraction (5-8 mL) eluted
with 80:20 MeOH:H2O was collected, transferred into a vial, and air-
dried giving the mappicine analogue 3 in an average amount of 1-2
mg. The purity was assessed by HPLC analysis (Novapak C18 column,
MeOH-H2O gradient) of 20% of randomly selected library samples.
Additional structure characterizations including MS, LC-MS, 1H NMR,
and LC NMR analyses were also carried out.
General Procedure for Radical Annulation. To a mixture of seven
N-propargyl pyridones 8{1-7,1-8} (0.024 mmol each, total 0.17 mmol)
was added hexamethylditin (15 µL, 0.069 mmol) and a solution of aryl
isonitrile (1.0 M in benzene, 0.5 mL, 0.5 mmol). The mixture was
purged with nitrogen for 5 min and sealed in a vial. The mixture was
irradiated with a sunlamp for 5 h and then loaded onto a SPE cartridge
packed with 2.5 g of silica. The cartridge was eluted with 10% EtOAc/
hexanes (20 mL) followed by 15% MeOH/EtOAc (10 mL). The MeOH/
EtOAc fraction was evaporated to dryness to give a mixture of seven
tagged mappicines 4{1-7,1-8,1-10}. All 80 mixtures were analyzed by
LC-MS before the next step of demixing.
Acknowledgment. We thank the National Institutes of Health
for the SBIR funding (1 R43 GM62717), Dr. Fu-Mei Lin
(University of Pittsburgh) for the help of LC-NMR and LC-
MS analyses, and Mr. Brian Bucher’s effort on making the
starting material. We dedicate this paper to the memory of Prof.
Jean-Marie Surzur.
Supporting Information Available: Analytical data and
spectra of representative intermediates and final products (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
General Procedure for Demixing of Tagged Mappicines 4{1-7,1-
8,1-10}. Demixing of 80 tagged mappicines mixtures was carried out
on a Waters HPLC system (manually) or a Gilson serial HPLC with
an automatic fraction collector. The separation conditions were as
JA026947D
9
10450 J. AM. CHEM. SOC. VOL. 124, NO. 35, 2002