A synthesis strategy yielding skeletally diverse small molecules combinatorially

Article abstract of DOI:10.1021/ja0457415

The efficient synthesis of small molecules having many molecular skeletons is an unsolved problem in diversity-oriented synthesis (DOS). We describe the development and application of a synthesis strategy that uses common reaction conditions to transform

Full text of DOI:10.1021/ja0457415

Published on Web 10/06/2004
A Synthesis Strategy Yielding Skeletally Diverse Small
Molecules Combinatorially
Martin D. Burke,1 Eric M. Berger, and Stuart L. Schreiber*
Contribution from the Broad Institute of HarVard and MIT,
Department of Chemistry and Chemical Biology, Howard Hughes Medical Institute,
HarVard UniVersity, 12 Oxford Street, Cambridge, Massachusetts 02138
Received July 15, 2004; E-mail: stuart_schreiber@harvard.edu
Abstract: The efficient synthesis of small molecules having many molecular skeletons is an unsolved
problem in diversity-oriented synthesis (DOS). We describe the development and application of a synthesis
strategy that uses common reaction conditions to transform a collection of similar substrates into a collection
of products having distinct molecular skeletons. The substrates have different appendages that pre-encode
skeletal information, called σ-elements. This approach is analogous to the natural process of protein folding
in which different primary sequences of amino acids are transformed into macromolecules having distinct
three-dimensional structures under common folding conditions. Like σ-elements, the amino acid sequences
pre-encode structural information. An advantage of using folding processes to generate skeletal diversity
in DOS is that skeletal information can be pre-encoded into substrates in a combinatorial fashion, similar
to the way protein structural information is pre-encoded combinatorially in polypeptide sequences, thus
making it possible to generate skeletal diversity in an efficient manner. This efficiency was realized in the
context of a fully encoded, split-pool synthesis of ∼1260 compounds potentially representing all possible
combinations of building block, stereochemical, and skeletal diversity elements.
Introduction
One aim of diversity-oriented synthesis (DOS) is to populate
chemical space in defined ways with small molecules having
diverse and complex structures.2 In contrast to a one synthesis-
one skeleton approach where many building blocks are appended
to a common core structure, DOS aims to generate both
stereochemical and skeletal diversity to create a diverse display
of chemical information in three-dimensional space. It also aims
to create products having attachment sites that facilitate the
appending of substituents during a “follow-up” phase of
research. Achieving these aims should facilitate the search for
specific small-molecule modulators of each of the functions
performed by the structurally diverse macromolecules that
comprise living systems.2,3
Figure 1. The Achmatowicz reaction.
skeletal diversity combinatorially. We recently reported the
development and application of a synthesis strategy for achiev-
ing this aim.5 In this approach, a collection of substrates with
different appendages that pre-encode skeletal information
(named σ-elements) are transformed into a collection of products
having distinct molecular skeletons using common reaction
conditions. An important feature of this approach is that multiple
σ-elements can be used to pre-encode skeletal diversity com-
binatorially and thereby provide efficient access to diverse
displays of chemical information in three-dimensional space.
Here, we describe in detail the development of this σ-element-
based strategy for generating skeletal diversity combinatorially
and its successful application in the context of a fully encoded
split-pool synthesis.
In both combinatorial chemistry and DOS, building block
diversity can be achieved in a highly efficient manner by
harnessing the power of combinatorics (a multiplicative increase
in the number of products with an additive increase in the
number of building blocks). This efficiency can be exploited
by using split-pool synthesis.
The Achmatowicz Reaction
Generating skeletal diversity with this same efficiency has
thus far been a formidable challenge;2,4 it requires generating
Our studies began with efforts to use the Achmatowicz
reaction6 in the context of DOS. As shown in Figure 1, this
complexity-generating reaction transforms a relatively simple
substrate 1 into a more complex product 3 having a [3.2.1]
(1) Adapted from the Ph.D. Thesis of Burke, M. D., Harvard University, 2003.
(2) (a) Schreiber, S. L. Science 2000, 287, 1964-1969. (b) Schreiber, S. L.
Chem. Eng. News 2003, 81, 51-61. (c) Burke, M. D.; Schreiber, S. L.
Angew. Chem., Int. Ed. 2004, 43, 46-58 and references therein.
(3) Schreiber, S. L. Bioorg. Med. Chem. 1998, 6, 1127-1152.
(4) (a) Weber, L. Curr. Opin. Chem. Biol. 2000, 4, 295-302 and references
therein. (b) Ding, S.; Gray, N. S.; Wu, X.; Ding, Q.; Schultz, P. G. J. Am.
Chem. Soc. 2002, 124, 1594-1596.
(5) Burke, M. D.; Berger, E. M.; Schreiber, S. L. Science 2003, 302, 613-
618 and references therein.
(6) Achmatowicz, O., Jr.; Bukowski, P.; Szechner, B.; Zwierzchowska, Z.;
Zamojski, A. Tetrahedron 1971, 27, 1973-1996.
9
10.1021/ja0457415 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 14095-14104
14095

A R T I C L E S
Burke et al.
Figure 2. Achmatowicz reaction and related transformations in the context
of TOS. References: (A) Kobayashi, Y.7 (B) Zhou, W.-S.8 (C) O’Doherty,
G. A.9 (D) Ogasawara, K.10 (E) Martin, S. F.11 (F) Deshong, P.12 (G)
O’Doherty, G. A.13 (H) Wender, P. A.14
Figure 4. Synthetic plan for a skeletal diversity-generating process.
Planning DOS Pathways
In TOS, synthesis pathways are devised most effectively using
retrosynthetic analysis, a planning strategy in which a target
structure is systematically deconstructed by formally performing
chemical reactions in the reverse-synthetic direction.17 In this
way, syntheses in TOS are planned in the direction of products
f reactants. Because DOS aims to generate a collection of
compounds with maximized diversity (rather than a single target
structure), retrosynthetic methods cannot be applied effectively
in this context. Forward-synthetic analysis aims to facilitate the
planning of efficient synthesis pathways that maximize molec-
ular complexity and diversity.2
Synthesis pathways in DOS are planned in the forward-
synthetic direction, that is, from reactants f products. The basic
subunit of forward-synthetic planning is the process, defined
as the transformation of a collection of substrates into a
collection of products. Furthermore, transformations of relatively
similar substrates into collections of more diverse products are
referred to as diversity-generating processes, and DOS pathways
are developed by identifying series of such processes having
products ) substrates relationships. In this context, the σ-ele-
ment-based strategy proposed above represents a potentially
general type of skeletal diversity-generating process, which we
have named a folding process in recognition of its conceptual
analogue.
Figure 3. The σ-element-based approach for generating skeletal diversity.
Substrates having different appendages that pre-encode skeletal information,
named σ-elements, are transformed into products having distinct molecular
skeletons using a common set of reaction conditions.
bicyclic ketal skeleton with potential for diversification via
diastereoselective functionalization of the resultant enone. It also
offers the potential for skeletal diversification, as hinted by the
intermediacy of a monocycle on route to the bridged ketal.
Stereochemical diversification was envisioned as arising from,
among others, the application of stereocontrolled reactions
leading to the diol-containing side chain.
Since the pioneering studies by Achmatowicz in 1971, skeletal
transformations based on furan oxidation have found widespread
use in the context of target-oriented synthesis (TOS). Some
selected examples are shown in Figure 2.
Key to this approach is the transformation of substrates having
different σ-elements into products having different skeletons
using common reaction conditions (split-pool synthesis renders
ad-hoc optimization for each substrate impractical). This type
of forward-synthetic planning can begin with the identification
of a core structure or functional group with the capacity for
transformation into a more reactive intermediate using, ideally,
a set of mild and selective reaction conditions. As shown in
Figure 4A, the furan ring is one example of this type of latent
intermediate: it can be transformed into a more reactive cis-
enedione using mild oxidative reaction conditions. The planning
of a folding process proceeds by conceptually appending
These reactions, involving transformations of relatively simple
and similar substrates into more complex and diverse products
using similar reaction conditions, are reminiscent of the process
of protein folding. Similar to the way the three-dimensional
structures of proteins are pre-encoded by distinct primary amino
acid sequences,15 the skeletal information for each small
molecule product is pre-encoded in the corresponding substrate,
not the applied reaction conditions. This conceptual analogy
led to a strategy for generating skeletal diversity in DOS, that
is, pre-encoding skeletal information into relatively simple and
similar substrates, and transforming them under single reaction
conditions into more complex and diverse products having
different molecular skeletons (see Figure 3).16
(12) Deshong, P.; Ramesh, S.; Perez, J. J. J. Org. Chem. 1983, 48, 2117-2118.
(13) Balachari, D.; O’Doherty, G. A. Org. Lett. 2000, 2, 4033-4036.
(14) Wender, P. A.; Rice, K. D.; Schnute, M. E. J. Am. Chem. Soc. 1997, 119,
7897-7898.
(7) Kobayashi, Y.; Nakano, M.; Kumar, G. B.; Kishihara, K. J. Org. Chem.
1998, 63, 7505-7515.
(15) Anfinsen, C. B. Science 1973, 181, 223-230.
(8) Yang, C.-F.; Xu, Y.-M.; Liao, L.-X.; Zhou, W.-S. Tetrahedron Lett. 1998,
39, 9227-9228.
(16) Burke, M. D.; Schreiber, S. L. HarVard-MIT DiVision of Health Sciences
and Technology Forum, The Speech Chain, Book of Abstracts; Cambridge,
MA, March, 2000; p 5.
(9) Harris, J. M.; O’Doherty, G. A. Tetrahedron Lett. 2000, 41, 183-187.
(10) Takeuchi, M.; Taniguchi, T.; Ogasawara, K. Synthesis 1999, 2, 341-354.
(11) Martin, S. F.; Gluchowski, C.; Campbell, C. L.; Chapman, R. C. J. Org.
Chem. 1984, 49, 2512-2513.
(17) (a) Corey, E. J.; Cheng, X.-M. The Logic of Chemical Synthesis; John Wiley
and Sons: New York, 1995. (b) Corey, E. J. Angew. Chem., Int. Ed. Engl.
(Nobel Lecture) 1991, 30, 455-465.
9
14096 J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004

Synthesis of Skeletally Diverse Small Molecules
A R T I C L E S
Scheme 1 a
Scheme 2 a
a Reagents and conditions: 5-hexen-1-ol, 2,6-lutidine, CH2Cl2, room
temperature, 12 h; 9-BBN, THF, room temperature, 5 h; 5-bromofuraldehyde
(29), PdCl2dppf, NaOH, THF:H2O 5:1, 65 °C, 20 h; loading ) 0.679
mequiv/g, purity >85% (LCMS).
a Reagents and conditions:22 (a) allyldiethylphosphonoacetate, LiOH,
THF, room temperature, 25 h; (b) Pd(PPh3)4, thiosalicylic acid, THF, room
temperature, 24 h; (c) isobutylchloroformate, 4-methylmorpholine, i-Pr2NEt,
THF, 0 °C, 2 h; LiBH4, i-Pr2NEt, THF, 4 °C, 24 h, 76% (three steps),
purity 68%; (d) phenyl isocyanate, pyridine, CH2Cl2, room temperature,
24 h; (e) OsO4, (DHQD)2PHAL, 4-methylmorpholine N-oxide, TEAAT,
acetone:H2O 10:1, 4 °C, 48 h, 65% (two steps), purity >90%; (f) (S)-(+)-
4-benzyl-3-propionyl-2-oxazolidinone, n-Bu2BOTf, Et3N, CH2Cl2, 72 h, -78
to 0 °C; 30% aqueous H2O2, pH7 buffer, MeOH, 4 °C, 12 h, >95%, purity
>90%; (g) Ac2O, i-Pr2NEt, DMAP, CH2Cl2, room temperature, 28 h, 88%,
purity >90%.
substituents (σ-elements) with complementary chemical reactiv-
ity to the reactive intermediate. Common conditions can then
be used to liberate the latent intermediate and to allow the
trapping of this reactive intermediate with the appended
σ-elements, leading to the formation of distinct molecular
skeletons.
This logic was used to develop the synthetic plan shown in
Figure 4B. First, the common furaldehyde substrate 20 is
transformed into three furan products 21-23 having three
different σ-elements - two-carbon side chains containing two,
one, or zero nucleophilic hydroxyl groups. These three com-
pounds 21-23 collectively serve as substrates for a skeletal
diversity-generating process. A common set of oxidative and
acidic reaction conditions is used to transform substrates 21-
23 into three products having distinct molecular skeletons, a
[3.2.1] bicyclic ketal 24 (via oxidative ring-expansion and acid-
promoted bicycloketalization), a cyclic hemiketal 25 (via
oxidative ring expansion and acid-promoted equilibration of the
anomeric center), and a trans-enedione 26 (via oxidative furan
cleavage and acid-promoted cis f trans isomerization).
Scheme 3 a
a Reagents and conditions: N-bromosuccinimide (NBS), NaHCO3,
NaOAc, THF:H2O 4:1, room temperature, 1 h; PPTS, CH2Cl2, 40-45 °C,
20 h, 35, 33%, purity 64%,23 36, 35%, purity 86%, 37, 81%, purity >90%.
Transforming σ-Elements into Distinct Skeletons
To test the feasibility of this plan, we performed solid-phase
syntheses of three substrates having two-carbon side chains
bearing two, one, or zero nucleophilic hydroxyl groups. A two-
step, solid-phase synthesis of model compound 5-(6-hydroxy-
hexyl)-2-furaldehyde (30) was developed as shown in Scheme
1. Commercially available 5-hexen-1-ol was loaded onto 500-
560 µm polystyrene-based “macrobeads” using a previously
published protocol.18 Hydroboration of the terminal olefin with
9-BBN followed by PdCl2(dppf)-mediated Suzuki coupling with
commercially available 5-bromo-2-furaldehyde (29) led to the
desired macrobead-bound furaldehyde 30 with high loading
(0.679 mequiv/g) and purity (>85% by LCMS).
reaction21 with aldehyde 30 provided the R-hydroxy furan 33
with >90% purity and >20:1 dr. Acylation of aldol adduct 33
generated the R-acetoxyalkyl furan 34 (>90% purity), which
bears no nucleophilic hydroxyl groups.
As shown in Scheme 3, when 32-34, each having a common
furan skeleton, were exposed to the same optimized set of
“folding conditions” (NBS, NaHCO3, and NaOAc in THF:H2O
4:1 at room temperature for 1 h followed by PPTS in CH2Cl2
at 40 °C for 20 h), they were transformed into products having
three different skeletons.
As shown in Scheme 2, this common furaldehyde starting
material was then transformed into three furan derivatives 32,
33, and 34 having two, one, and zero appended nucleophilic
hydroxyl groups, respectively. Horner-Wadsworth-Emmons
olefination of aldehyde 30 with allyldiethylphosphonoacetate
followed by deallylation, reduction, carbamate formation, and
Sharpless asymmetric dihydroxylation19 generated macrobead-
bound diol 32 in >90% purity (LCMS analysis of crude product
after cleavage from macrobeads) and 66% ee.20 Evans’ aldol
Furan derivatives 32 and 34 were transformed into the
anticipated [3.2.1] bicyclic ketal 35 (64% purity23) and trans-
enedione 37 (>90% purity), respectively. However, aldol adduct
33, having a single nucleophilic hydroxyl group appended to
the same furan core, underwent an initial oxidative ring
expansion to the expected cyclic hemiketal followed by an
unanticipated, acid-mediated dehydration reaction to yield the
alkylidene-pyran-3-one 36 (86% purity) as a single geometric
isomer.24 (For an animation of this skeletal diversity-generating
(18) (a) Tallarico, J. A.; Depew, K. M.; Pelish, H. E.; Westwood, N. J.; Lindsley,
C. W.; Shair, M. D.; Schreiber, S. L.; Foley, M. A. J. Comb. Chem. 2001,
3, 312-318. (b) M. Narovlyansky and J. Tallarico are gratefully acknowl-
edged for generously providing all of the macrobeads used in these studies.
(19) (a) Kolb, H. C.; Van Nieuwenhze, M. S.; Sharpless, K. B. Chem. ReV.
1994, 94, 2483-2547. (b) For a study of the Sharpless dihydroxylation on
solid phase, see: Han, H.; Janda, K. Angew. Chem., Int. Ed. Engl. 1997,
36, 1731-1733.
(21) Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc. 1981, 103, 2127-
2129.
(22) Yields were determined by mass of recovered products after HF-mediated
cleavage from ∼75 mg of macrobeads and purification by silica gel
chromatography. Purities were determined by LCMS analysis (UV detection
at λ214) of crude products following cleavage from macrobeads.
(23) An impurity in the LCMS chromatogram that was 1H NMR-silent and had
an isotope pattern in its mass spectrum consistent with an osmium-
containing substance was not included in the calculation of the purity of
product 35.
(20) In a solution-phase model study, the asymmetric dihydroxylation of a related
substrate proceeded with >90% ee. It is unclear why the enantioselectivity
was diminished for the solid-phase reaction.
9
J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004 14097

A R T I C L E S
Burke et al.
Scheme 5
Figure 5. Generating skeletons of small molecules combinatorially.
Scheme 4
Scheme 6
Scheme 7
folding process, see http://slsiris.harvard.edu/home/animation.ht-
ml.) Unanticipated reaction outcomes, particularly those involv-
ing the formation of new skeletons, can be highly valuable in
DOS.
Generating Skeletal Diversity Combinatorially
Scheme 8
Scheme 9
Expanding the above approach to generate skeletal diversity
combinatorially requires the identification of at least two sets
of σ-elements that function in combination to pre-encode
different molecular skeletons (see Figure 5). This approach has
the potential to achieve a multiplicative increase in skeletal
outcomes with an additive increase in σ-elements (just as a
multiplicative increase in building block combinations is
achieved with the one synthesis-one skeleton approach), and
thereby to generate skeletal diversity efficiently.
During the course of our studies, it was determined that
substituents at the 4-position of furan can also pre-encode
different skeletal outcomes. Moreover, we found that substit-
uents appended at the 4-position of furan and those on the
R-carbon can function in combination to pre-encode a combi-
natorial matrix of diverse skeletons.
In model studies originally directed at expanding building
block diversity, it was determined that a regioselective Suzuki
reaction with commercially available 4,5-dibromo-2-furaldehyde
(38)25 could be used to generate macrobead-bound 4-bromo-5-
alkyl-2-furaldehyde 39 (see Scheme 4).26 Competitive â-hydride
elimination, observed when using Pd(PPh3)4 as catalyst in this
reaction, was minimized by using PdCl2dppf.27 No 4,5-bis-alkyl
functionalization (via on-bead site-site interactions)28 was
observed. It was also determined that the 4-position of 39 could
subsequently be functionalized26,29 at higher temperatures using
Pd(PPh3)4 as catalyst and a variety of arylboronic acids as
coupling agents to yield bis-functionalized furaldehydes 40.30
As shown in Scheme 5, macrobead-bound-4-bromo-5-alkyl-
2-furaldehyde 39 and 4-m-methylphenyl-5-alkyl-2-furaldehyde
41 were determined to be excellent substrates for the Evans
aldol reaction with or without subsequent acetylation (see
Scheme 11 below for optimized syntheses of these compounds)
to generate model substrates 42-45,31 each having a distinct
combination of substituents at the 2- and 4-positions of the furan
ring. Treating these four substrates with the oxidative and acidic
conditions described above led to four different skeletal
outcomes.
(24) The mechanism for this dehydration presumably involves acid-promoted
E1-type elimination via an oxocarbenium ion intermediate. The indicated
cis-alkene in 36 (stereochemistry assigned with 1H NMR nOe studies) likely
represents the thermodynamic product (minimization of sterics).
(25) This reagent was purchased from Moscow MedChemlabs (Moscow, Russia)
or synthesized in one step via treatment of 2-furaldehyde with molecular
bromine in the presence of aluminum trichloride: Chiarello, J.; Joullie´,
M. M. Tetrahedron 1988, 44, 41-48.
(26) Similar regioselective Sonogashira and Stille couplings have been re-
ported: Bach, T.; Kru¨ger, L. Eur. J. Org. Chem. 1999, 2045-2057.
(27) (a) Miyaura, N.; Ishiyama, T.; Sasaki, H.; Ishikawa, M.; Satoh, M.; Suzuki,
A. J. Am. Chem. Soc. 1989, 111, 314-321. (b) Hayashi, T.; Konishi, M.;
Kobori, Y.; Kumada, M.; Higuchi, T.; Hirotsu, K. J. Am. Chem. Soc. 1984,
106, 158-163.
(28) Blackwell, H. E.; Clemons, P. A.; Schreiber, S. L. Org. Lett. 2001, 3, 1185-
1188.
(29) (a) Chang, C. K.; Bag, N. J. Org. Chem. 1995, 60, 7030-7032. (b) Zhou,
X.; Tse, M. K.; Wan, T. S. M.; Chan, K. S. J. Org. Chem. 1996, 61, 3590-
3593.
(30) Efforts to couple a variety of B-alkyl reagents at the 4-position were not
fruitful. Efficient coupling was achieved under similar conditions using as
coupling agents a variety of trans-alkenylboronic acids. Less efficient
coupling was achieved with tetramethyltin [using PdCl2(Po-Tol3)2 in DMA
at 115 °C, two cycles].
9
14098 J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004

Synthesis of Skeletally Diverse Small Molecules
A R T I C L E S
Scheme 10
4-hydrido-2-R-acetoxyalkyl furan 34, each bearing a single
deactivating substituent, were resistant to oxidation at 0 °C
(Scheme 7B and C, respectively), even though both substrates
were oxidized quantitatively in 1 h at room temperature. By
combining these two deactivating substituents into the same
substrate (43), a distinct skeletal outcome was achieved upon
exposure to the same reagents at room temperature, that is,
preservation of the initial furan skeleton via complete resistance
to oxidation.
Scheme 11 a
The ability of electron-withdrawing substituents to deactivate
nearby olefins toward electrophilic reagents is well-known.33
In early studies of furan oxidation, Clauson-Kaas and co-workers
noted that furan derivatives having highly electron-withdrawing
substitutents (e.g., furoic acid and R-acetylfuran) were resistant
to oxidation with molecular bromine.34 It seems reasonable that
the deactivating effect of the bromine substituent at the
4-position is predominantly due to its electron-withdrawing
nature (Pauling electronegativity35 for -Br is 2.8 as opposed
to 2.1 for -H), although steric effects may also contribute.
The deactivating effect of the 2-R-acetoxyalkyl group is likely
the result of two factors: (1) the electron-withdrawing nature
of the acetoxy group, and (2) the removal of the hydrogen-
bond donating hydroxyl group on the R-carbon, which likely
facilitates furan oxidation via intramolecular directing of the
NBS reagent. This proposed two-fold effect is consistent with
the classic studies by Henbest and co-workers on the epoxidation
of cyclohexene and its derivatives using the oxidant perbenzoic
acid.36
As shown in Scheme 8, treatment of 4-aryl-2-R-hydroxyalkyl
furan 44 with NBS resulted in oxidative ring expansion to yield
the expected aryl-substituted cyclic hemiketal 48. However,
when exposed to acidic conditions (e.g., PPTS), complete
conversion into a new product was observed, identified as the
R-keto furan 49.37
The basis for this reactivity is not known. A similar type of
reaction, involving the Br2/H2SO4-mediated transformation of
an R-hydroxy-â-butoxycarbonyl furan into an R-keto-â-butoxy-
carbonyl derivative, was reported in early studies by Achma-
towicz,6 who proposed the intermediacy of an enediol tautomer.
The oxidative ring expansion and acid-promoted rearrange-
ment observed for 4-m-methylphenyl-2-R-hydroxyalkyl furan
44 proved to be general for a series of related 4-aryl-2-R-
hydroxyalkyl furans that were synthesized and tested. It was
observed that this outcome is more preferred with substrates
having electron-donating aryl substituents (e.g., p-methoxyphe-
nyl) versus electron-withdrawing aryl substituents (e.g., 2,4-
dichlorophenyl).
a X ) (S)-4-benzyl-2-oxazolidinone, Ar ) m-methylphenyl. Reagents
and conditions: (a) 9-BBN, THF, room temperature, 5 h; 4,5-dibromo-
furaldehyde (38), PdCl2dppf, NaOH, THF:H2O 5:1, 65 °C, 18 h, 0.188
mequiv/g; (b) 9-BBN, THF, room temperature, 5 h; 4-m-MePh-5-bromo-
furaldehyde (53), PdCl2dppf, NaOH, THF:H2O 5:1, 65 °C, 22 h, 0.545
mequiv/g; (c) (S)-(+)-4-benzyl-3-propionyl-2-oxazolidinone, n-Bu2BOTf,
Et3N, CH2Cl2, 72 h, -78 to 0 °C; 30% aqueous H2O2, pH 7 buffer, MeOH,
4 °C, 12 h, 42, >95%, purity >90%, 44, 95%, purity >90%; (d) Ac2O,
i-Pr2NEt, DMAP, CH2Cl2, room temperature, 28 h, 43, 90%, purity >90%,
45, 84%, purity >90%; (e) NBS, NaHCO3, NaOAc, THF:H2O 4:1, room
temperature, 1 h; PPTS, CH2Cl2, 40-45 °C, 20 h, 46, 82%, purity 90%,
43, 88%, purity >90%, 49, 74%, purity 72%, 50, 72%, purity 66%.
As shown in Scheme 6, model substrate 42 having a bromine
atom at the 4-position of furan and a hydroxyl group on the
R-carbon of the 2-substituent underwent quantitative, NBS-
mediated oxidative ring expansion to yield the anticipated cyclic
hemiketal 46 as a >9:1 mixture of epimers.32 However, upon
exposure to the same acidic conditions (PPTS in CH2Cl2 at 40
°C) that effected dehydration with the hemiketal derived from
the oxidation of model substrate 33 (having a hydrogen atom
at the 4-position of furan, see 33 f 36, Scheme 3), the bromine-
substituted cyclic hemiketal 46 remained unchanged. It seems
plausible that formation of an oxocarbenium ion intermediate,
presumably required for dehydration to occur, is disfavored both
electronically and sterically by the electronegative and large (1,2-
type strain) bromine substituent.
In contrast, the 4-bromo-2-R-acetoxyalkyl furan 43 proved
completely resistant to NBS-mediated oxidation and was
likewise unchanged upon subsequent treatment with PPTS,
resulting in preservation of the original R-alkoxyalkyl furan
skeleton. As shown in Scheme 7, the deactivating effect of both
the 4-bromo and the 2-R-acetoxyalkyl substituents toward furan
oxidation was observed independently when the reaction was
performed at 0 °C (∼1 h reaction time). As a point of reference,
the 4-hydrido-2-R-hydroxyalkyl furan 33 was readily oxidized
both at room temperature and at 0 °C by NBS (Scheme 7A).
Hovever, both the 4-bromo-2-R-hydroxyalkyl furan 42 and the
As shown in Scheme 9, when treated with the same oxidative
and acidic reagents, 4-aryl-2-R-acetoxyalkyl furan 45 was
(33) (a) Swern, D. J. Am. Chem. Soc. 1947, 69, 1692-1698. (b) Swern, D.
Chem. ReV. 1949, 45, 1-68.
(34) Clauson-Kaas, N.; Limborg, F.; Fakstorp, J. Acta Chem. Scand. 1948, 2,
109-115.
(35) (a) Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell University
Press: Ithaca, NY, 1960. (b) Carey, F. A.; Sundberg, R. J. AdVanced
Organic Chemistry, Part A: Structure and Mechanisms, 3rd ed.; Plenum
Press: New York, 1990; p 15.
(31) The HWE olefination, reduction, functionalization, and dihydroxylation
sequence was also explored with these and related substrates; however,
the length of the synthesis pathway precluded the generation of the desired
macrobead-bound diols with acceptable macrobead integrity and compound
purity.
(32) The stereochemistry at the anomeric carbon of cyclic ketal 46 was tentatively
assigned as R on the basis of two-dimensional NOESY and one-dimensional
nOe experiments.
(36) Henbest, H. B.; Wilson, R. A. L. J. Chem. Soc. 1957, 1958-1965.
(37) (a) Epimerization of the potentially labile methyl-bearing stereogenic center
in 49 was not observed. This observation is consistent with the lack of
epimerization observed with the structurally similar Evans’ extended
polypropionate (â-ketoimide) reagents. (b) Evans, D. A.; Clark, J. S.;
Metternich, R.; Novack, V. J.; Sheppard, G. S. J. Am. Chem. Soc. 1990,
112, 866-868. (c) Evans, D. A.; Ng, H. P.; Clark, J. S.; Rieger, D. L.
Tetrahedron 1992, 48, 2127-2142.
9
J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004 14099

A R T I C L E S
Burke et al.
Figure 6. A combinatorial matrix of σ-elements pre-encodes a complete
matrix of distinct skeletal outcomes.
transformed into product 50 having a cis-enedione skeleton,38
in contrast to the trans-enedione skeleton formed from the
(otherwise identical) 4-hydrido derivative 34 (see 34 f 37,
Scheme 3).39 The selective formation of products having a cis-
enedione skeleton proved to be general for a series of 4-aryl-
2-R-acetoxyalkyl furans similar to 45 that were synthesized and
tested. Sterics and increased electron delocalization likely
contribute to making the cis isomers thermodynamically more
favorable for these compounds.
These cis-enedione products, for example, 50 proved to be
somewhat sensitive to acid40 (more than the related trans-
enedione derivatives, e.g., 37), and it was observed that this
acid-sensitivity correlated positively with the electron-donating
capacity of the aryl unit. Therefore, while strongly electron-
donating aryl substituents were most effective in promoting
rearrangements to R-keto furans (see above), electron-withdraw-
ing aryl substituents were best for producing stable cis-enedione
products. Studies using various aryl-substituted furan derivatives
in both transformations demonstrated that a good balance in
reactivity was achieved with furans substituted at the 4-position
with the m-methylphenyl substituent, making it possible to carry
out both transformations using a common set of oxidative (NBS)
and acidic (PPTS) reaction conditions.
To summarize these results, it was determined that substitu-
tions at the 4-position of furan (-H, -Br, or -aryl) function
as a second σ-element by pre-encoding distinct skeletal out-
comes upon exposure to a common set of oxidative and acidic
reaction conditions. Moreover, as summarized in Figure 6, it
was determined that this second σ-element functions in com-
bination with substituents on the R-carbon (-OH or -OAc) of
the 2-substituent to pre-encode a combinatorial matrix of distinct
skeletal outcomes (3 × 2 ) 6; i.e., a multiplicative increase in
skeletal outcomes with an additive increase in σ-elements).
Figure 7. (A) A complete, combinatorial matrix of distinct skeletal
outcomes achieved using a common set of reaction conditions. (B) A skeletal
diversity metric for quantitating the diversity in the display of chemical
information in three-dimensional space.
commercially available 4-bromo-2-furaldehyde (51) and m-
methylphenylboronic acid followed by direct, regioselective
bromination of the product 52 using molecular bromine in
DMF.41 Additionally, reaction conditions were optimized for
the solid-phase synthesis of model substrates 42-45 and their
transformation into skeletally distinct products 46, 43, 49, and
50 using a common set of oxidative and acidic reaction
conditions (see Scheme 11).
Realizing the combinatorial matrix of molecular skeletons
shown in Figure 6 in the context of a split-pool synthesis requires
effecting each of the transformations under the same reaction
conditions, for example, in one pot. This is because the process
of splitting and pooling obscures the identity of the σ-elements
(and combinations thereof) appended to the furan substrates
attached to each bead. To explore the potential of achieving
this aim, we performed the following experiment in triplicate.
As shown in Figure 7A, a common reaction vessel was charged
with a collection of six macrobead-bound substrates 33, 34, and
42-45, collectively representing a complete, combinatorial
matrix of σ-elements, that is, -H, -Br, or -Ar at the 4-position
of furan combined with -OH or -OAc on the R-carbon. We
then exposed this pooled collection of substrates to the same
oxidative and acidic reaction conditions (NBS, NaHCO3,
NaOAc, THF:H2O 4:1, room temperature, 1 h; PPTS, CH2Cl2,
40-45 °C, 20 h), washed and dried the product beads, separated
them into six individual Eppendorf tubes, and performed single-
bead compound cleavage followed by LCMS analysis of each
of the cleaved products. The results were consistent with the
formation of 6/6 of the anticipated molecular skeletons as the
major products in all three experiments.
Studies Directed Toward Split-Pool Synthesis
To provide some quantification for the skeletal diversity
generated in the collective transformation shown in Figure 7A,
a skeletal diversity metric was developed on the basis of the
distance, angle, and dihedral angle between common atoms in
computationally derived three-dimensional structures. The miss-
ing bonds in both the substrates and the products in Figure 7A
represent potential attachment sites to which building blocks
could be appended. The six substrates, having a common
2-substituted R-alkoxyalkyl furan skeleton, resemble the types
A more concise route (with respect to the number of reactions
performed on solid phase) to the 4-m-methylphenyl-5-alkyl-2-
furaldehyde 41 was developed involving direct coupling of 4-m-
MePh-5-bromo-2-furaldehyde (53) (see Scheme 10). This
reagent was prepared in two steps: Suzuki coupling between
(38) The olefin geometry was determined via nOe studies.
(39) Similar distinct stereochemical outcomes have been observed in the
oxidation of an alternative series of furan derivatives: Sayama, S.; Inamura,
Y. Heterocycles 1996, 43, 1371-1374.
(40) Upon prolonged treatment with acidic reagents, slow degradation of 50 to
unidentified byproducts was observed (this degradation was more rapid
with CSA than with PPTS).
(41) For a related bromination, see: Sessler, J. L.; Hoehner, M. C.; Gebauer,
A.; Andrievsky, A.; Lynch, V. J. Org. Chem. 1997, 62, 9251-9260.
9
14100 J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004

Synthesis of Skeletally Diverse Small Molecules
A R T I C L E S
Scheme 12
Figure 8. Revised synthetic plan. Efficiently generating overlapping,
combinatorial matrices of both molecular skeletons and appended building
blocks using split-pool synthesis. The diamond-filled arrows represent
appending processes carried out in split-pool format.
approach, consecutive appending processes are performed in
split-pool format to yield a common core structure with both a
combinatorial matrix of building blocks and a combinatorial
matrix of σ-elements. (The diamond-filled arrow is used in DOS
to represent an appending process carried out using the split-
pool technique.) These pooled substrates are then exposed to a
common set of reaction conditions resulting in a matrix of
distinct molecular skeletons, each derivatized with the same
matrix of building blocks. Specifically, as shown in Figure 8B,
we planned a consecutive series of four appending processes
involving macrobead loading, Suzuki reaction, Evans’ aldol
coupling, and ( acetylation to introduce BB1, σ1, BB2, and σ2,
respectively. The resulting products 58, representing the com-
plete matrix of building blocks and σ-elements appended to a
common R-alkoxy furan core, then serve as substrates for a final
NBS- and PPTS-mediated skeletal diversity-generating folding
process.
Realizing this proposed plan required that all of the following
questions be answered in the affirmative: (1) Can combinatorial
building block diversity be achieved by using diverse coupling
partners in both the Suzuki and the Evans’ aldol coupling
reactions? (2) Can the combinatorial matrix of σ-elements
discovered in the previous section prove to be general and
effectively pre-encode the same matrix of distinct skeletal
outcomes when a diverse collection of building blocks is
appended to the same common core? (3) Is this chemistry
compatible with an effective encoding strategy that enables the
identification of the compounds attached to each macrobead at
the end of the synthesis? We therefore set out to answer each
of these questions experimentally.
of compounds typically derived from the one synthesis-one
skeleton approach. Alternatively, the six products represent six
molecular skeletons generated combinatorially using the σ-ele-
ment-based strategy. Comparing and contrasting these two
collections (which are almost constitutionally isomeric) can
provide a metric for the skeletal diversity generated in this one
reaction using a common set of reagents.
We generated structures in silico in which each of the missing
bonds in Figure 7A was “replaced” with a methyl group (or a
methylene group for the “left side” of structure 36) to form a
collection of 12 simplified structures sharing in common the
seven contiguous carbon atoms labeled C1-C7. Using the
Spartan software package (Spartan ’02, Wavefunction, Inc.),
the following two-step calculation was performed on all 12
structures: (1) the equilibrium conformer was determined using
the Spartan equilibrium conformer search with semiempirical
AM1 calculations, and (2) the equilibrium geometry was
determined using the Hartree-Fock method with the 6-31G*
split-valence basis set. The positions of every other carbon in
the common, contiguous seven-carbon atom stretch were then
used to determine the following three parameters (each param-
eter provides unique information regarding the relative positions
of the building block attachment sites, C1 and C7, in three-
dimensional space): (1) the distance (in angstroms) between
C1 and C7; (2) the angle C1-the midpoint between C3 and C5-
C7; and (3) the dihedral angle comprising C1, C3, C5, and C7.
Plotting these parameters for both substrates and products in
a three-dimensional plot using the Spotfire graphing package
produced the results shown in Figure 7B. The six simplified
substrates create a dense cluster (the two lobes correspond to
the acetylated and nonacetylated compounds). In contrast, the
six skeletally distinct products distribute much more broadly,
consistent with a diverse display of chemical information in
three-dimensional space.
We first explored the building block diversity that might be
generated by using different coupling partners in both the Suzuki
and the Evans’ aldol coupling reactions. For the Suzuki reaction
(BB1), we tested a variety of commercially available compounds,
each containing both a primary hydroxyl group and a terminal
olefin, and found a diverse set of seven that effectively
underwent loading onto macrobeads and subsequent B-alkyl
Suzuki coupling with 5-bromofuraldehyde (29), 4,5-dibromo-
furaldehyde (38), and 4-m-MePh-5-bromofuraldehyde (53)
(Scheme 12).
Revised Synthetic Plan: Overlapping, Combinatorial
Matrices of Molecular Skeletons and Appended
Building Blocks
The discovery of σ-elements that pre-encode a complete
matrix of distinct skeletal outcomes stimulated the revised
synthetic plan shown schematically in Figure 8A. In this
We then prepared candidate aldol coupling reagents (BB2)
by generating various combinations of acyl side chains and
9
J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004 14101

A R T I C L E S
Burke et al.
Figure 9. (A) Parallel synthesis of 36 substrates 61a-jj representing all possible combinations of a matrix of building and a matrix of σ-elements appended
to the same R-alkoxy furan core. (B) Transformation of this collection of 36 relatively simple and similar substrates into a collection of 36 more complex
and diverse products - a set of complete, overlapping matrices of molecular skeletons and appended building blocks.
commercially available chiral oxazolidinones. From these stud-
ies, we identified five distinct commercially available chiral
oxazolidinones and three different acyl side chains (ethyl,
methoxymethyl, and phenylethyl) that in all possible combina-
tions (5 × 3 ) 15) yielded acyl oxazolidinones that underwent
efficient and diastereoselective coupling with macrobead-bound
furaldehyde 30 (Scheme 13). The 15 enantiomeric acyl oxazo-
lidinones were also prepared, allowing us to take advantage of
reagent-based stereocontrol to prepare both sets of enantiomeric
(or diastereomeric when BB1 is chiral) products.
skeletal outcomes described above, the same matrix of σ-ele-
ments would need to pre-encode effectively distinct skeletal out-
comes in the presence of diverse combinations of building blocks.
To explore these questions efficiently, we selected representative
Suzuki and Evans’ aldol coupling partners (Figure 9A). We then
used parallel synthesis to prepare all possible combinations of
these representative building blocks and σ-elements appended
to the same R-alkoxy furan core [(2 × 3) building blocks × (3
× 2) σ-elements ) 36 individually synthesized compounds,
61a-jj, see Figure 9A]. For each of these 36 compounds, we
cleaved ∼5 mg of resin and characterized the crude products
by 1H NMR, LCMS, and HRMS. These data revealed successful
formation of the anticipated compounds for 36/36 cases (HRMS
Although two collections of candidate building blocks (BB1
and BB2) had been identified, it remained to be seen whether
different combinations of these building blocks would be
compatible. To achieve the combinatorial matrix of distinct
9
14102 J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004

Synthesis of Skeletally Diverse Small Molecules
A R T I C L E S
Scheme 13
sions achieved in the solid-phase reactions or the purities of
the cleaved products. Encouraged by these results, we set out
to realize the synthetic plan outlined in Figure 8 in the context
of a five-step, fully encoded split-pool synthesis.
Split-Pool Synthesis
As shown in Figure 10, we ancipated the formation of 1260
distinct compounds representing all possible combinations of
the following variables: [(7 BB1) × (3 σ1) × (15 BB2) × (2
enantiomers) × (2 σ2) ) 1260]. We therefore started our
synthesis with ∼10 400 macrobeads 54 (2 g, ∼5.2 macrobeads/
g), representing >8 copies of the complete, theoretical matrix.44
In the first step, the pooled collection of macrobeads was divided
evenly into seven portions, and each was subjected to a loading
reaction with a unique BB1 (see Scheme 12). Each portion was
tagged with polychlorinated aromatic tags T1A-T3A in a
unique combination42 using rhodium triphenylacetate. The
resulting seven portions were pooled and thoroughly mixed. The
pooled collection of macrobeads 55 was divided into three equal
portions, and to each was coupled one of three different
5-bromofuraldehyde derivatives, each bearing a distinct σ-ele-
ment (σ1) in the 4-position. After the three parallel Suzuki
reactions were completed, to each portion of product macrobeads
was coupled a unique combination of tags 4A and 5A. The
resulting three portions of resin were then combined and
thoroughly mixed (56), and then split evenly into 30 portions.
To each of these 30 portions was then coupled a unique acyl
oxazolidinone (BB2) (see Scheme 13) followed by a unique
combination of tags T6A-T11A. These 30 portions were then
pooled to yield the collection of aldol adducts 92. In the final
split-pool step, this pooled collection of macrobeads 92 was
divided evenly into two portions (1.08 g each).45 One portion
was subjected to an acetylation reaction followed by tagging
with tag T13A, and the other portion was left unreacted to yield
the collection of macrobeads 93, representing theoretically all
possible combinations of BB1, σ1, BB2, and σ2, in both
enantiomeric and diastereomeric (when BB1 is chiral) forms.
Prior to pooling these two portions, 30 individual macrobeads
were removed from each portion for analysis. The synthesized
compound and corresponding chemical tags were orthogonally
cleaved from each individual macrobead and analyzed using
LCMS and GC, respectively. These data revealed that the
structure predicted by the chemical tags matched the LCMS of
the cleaved products in 60/60 cases, and 55/60 (92%) of the
products were determined to be g70% pure.
error < 5 ppm), and, by LCMS analysis, 35/36 (97%) of these
compounds were determined to be g70% pure.
Each of these macrobead-bound substrates 61a-jj was then
exposed to the same set of oxidative (NBS) and acidic (PPTS)
reaction conditions in 36 parallel experiments. A portion of
macrobeads (∼5 mg) from each reaction was then cleaved, and
the crude products were analyzed using 1H NMR, LCMS, and
HRMS. These three forms of characterization were consistent
with the formation of the anticipated functionalized molecular
skeleton in 36/36 cases (HRMS error < 5 ppm) (Figure 9B).
Moreover, 26/36 (72%) of these products were determined to
be g70% pure by LCMS. This series of parallel experiments
demonstrated high compatibility between combinations of
representative building blocks and generality for the skeletal
diversity-generating combinatorial matrix of σ-elements.
We tested the compatibility of this chemistry with an
optimized protocol for the Still chemical encoding methodol-
ogy.42,43 We performed the solid-phase synthesis of compounds
36 and 37 (see Scheme 3) using the five-step route proposed in
Figure 8B, while incorporating after each synthetic step the
minimum number of polychlorinated aromatic chemical tags
required to generate a nonredundant binary code for the
anticipated number of appendages to be used in the planned
split-pool synthesis [step 1, 7 BBs, 3 tags; step 2, 3 σ-elements,
2 tags; step 3, 15 BBs × 2 enantiomers, 6 tags; step 4, 2
σ-elements ((acetylation), 1 tag]. After each tagging step, the
compound and chemical tags were orthogonally cleaved [using
HF-pyridine and ceric ammonium nitrate (CAN), respectively]
from a portion (∼5 mg) of macrobeads and were analyzed
separately (compounds were analyzed by LCMS, and tags were
analyzed by GC as described previously).42 It was determined
that the synthesis plan and the tagging methodology were
compatible, resulting in no significant decreases in the conver-
Finally, 853 mg of macrobeads 93 (∼4410 macrobeads,
multiplicative factor ) 3.5)44 was pooled and collectively
exposed to the same oxidative (NBS) and acidic (PPTS) reaction
conditions described above to yield the final collection of
products 94. Next, 120 of these product macrobeads were
cleaved and analyzed as above. The chemical structures
predicted by the chemical tags were consistent with LCMS
analysis of the cleaved products for 120/120 macrobeads.
(44) Statistical calculations and computer simulations suggest that a multiplicative
factor of 3.1 is required to provide 99% confidence of achieving 95%
coverage of the complete, theoretical combinatorial matrix for a split-pool
synthesis involving four split-pool cycles with 10 pools per cycle: Burgess,
K.; Liaw, A. I.; Wang, N. J. Med. Chem. 1994, 37, 2985-2987.
(45) Throughout the course of the synthesis, the total mass of macrobeads
increased somewhat, reflecting the added mass of the attached compounds.
There was also some loss of intact macrobeads throughout the split-pool
synthesis due to bead-breakage.
(42) (a) Nestler, H. P.; Bartlett, P. A.; Still, W. C. J. Org. Chem. 1994, 59,
4723-4724. (b) Blackwell, H. E.; Pe´rez, L.; Stavenger, R. A.; Tallarico,
J. A.; Eatough, E. C.; Foley, M. A.; Schreiber, S. L. Chem. Biol. 2001, 8,
1167-1182.
(43) Jennifer Raggio, Leticia Castro, and John Tallarico are gratefully acknowl-
edged for providing encoding reagents, and for cleaving and analyzing
chemical tags.
9
J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004 14103

A R T I C L E S
Burke et al.
Figure 10. A five-step, fully encoded split-pool synthesis of a collection of ∼1260 compounds 94 representing a complete, combinatorial (3 × 2 ) 6)
matrix of molecular skeletons, each derivatized with a complete, combinatorial (7 × 15 ) 105) matrix of building blocks in both enantiomeric and diastereomeric
forms (6 × 105 × 2 ) 1260).
Moreover, 84/120 (70%) of these final products were determined
by LCMS to be g70% pure. These results are consistent with
the plan to generate combinatorially a complete set of overlap-
ping matrices of molecular skeletons and appended building
blocks in both enantiomeric and diastereomeric forms.
potentially representing all possible combinations of building
block, stereochemical, and skeletal diversity elements. Such
collections will likely prove to be valuable in efforts to
understand the roles each of these three diversity elements plays
in small-molecule protein interactions. Systematic screening
efforts designed to address this specific issue are now underway.
Summary and Conclusions
We have developed and implemented a synthesis strategy for
generating skeletal diversity in DOS. The approach is conceptu-
ally analogous to the natural process of protein folding. It in-
volves using common reaction conditions to transform relatively
simple and similar substrates having different appendages that
pre-encode skeletal information (σ-elements) into more complex
and diverse products having distinct molecular skeletons.
An attractive feature of this approach is that skeletal diversity
can be pre-encoded into substrates combinatorially, thus making
it possible to generate a complete matrix of molecular skeletons
in an efficient manner. This strategy also permits the formation
of diverse skeletons late in the synthesis pathway, which has
two theoretical advantages. First, maintaining relative structural
similarity in the early stages of a synthesis facilitates the
identification of products ) substrates relationships and the use
of the powerful technique of split-pool synthesis. Second,
forming new skeletons late in a synthesis can facilitate the
generation of functionalized skeletons that might be otherwise
difficult to access in DOS, such as those having building blocks
coupled via carbon-carbon bonds at stereogenic quaternary
carbon centers (e.g., 46) and/or potentially unstable molecular
skeletons (e.g., acid-sensitive enediones 37 and 50).
Acknowledgment. We wish to thank M. Narovlyansky and
J. Tallarico for providing macrobeads, J. Raggio, L. Castro, and
J. Tallarico for providing encoding reagents and for cleaving
and analyzing chemical tags, Q. Liao, P. Wang, and J. Dudek
for high resolution mass spectroscopic analysis, and W. Collins
and S. Huang for assistance with NMR instrumentation.
Additionally, Y.-K. Kim is gratefully acknowledged for helpful
discussions regarding the transformation of 30 f 31. Financial
support for this work was provided by the NIGMS and the NCI.
The Keck Foundation, Merck KGaA, Merck & Co., and the
NCI are gratefully acknowledged for their past support of
Harvard’s ICCB, and the NCI is additionally acknowledged for
support of the Initiative for Chemical Genetics. M.D.B. was a
Howard Hughes Medical Institute predoctoral fellow, E.M.B.
was an HHMI undergraduate fellow, and S.L.S. is an Investiga-
tor at the HHMI.
Supporting Information Available: Experimental procedures
and analytical data for model compounds, building block testing,
parallel synthesis, and fully encoded split-pool synthesis. This
material is available free of charge via the Internet at
http://pubs.acs.org.
As demonstrated in the split-pool synthesis in Figure 10, this
strategy can provide access to a collection of compounds
JA0457415
9
14104 J. AM. CHEM. SOC. VOL. 126, NO. 43, 2004