Relationship of stereochemical and skeletal diversity of small molecules to cellular measurement space

Article abstract of DOI:10.1021/ja048170p

Systematic and quantitative measurements of the roles of stereochemistry and skeleton-dependent conformational restriction were made using multidimensional screening. We first used diversity-oriented synthesis to synthesize the same number (122) of [10.4.

Full text of DOI:10.1021/ja048170p

Published on Web 10/20/2004
Relationship of Stereochemical and Skeletal Diversity of
Small Molecules to Cellular Measurement Space
Young-kwon Kim,†,‡,§ Midori A. Arai,†,‡ Takayoshi Arai,†,‡ Julia O. Lamenzo,§
Elton F. Dean, III,§ Nick Patterson,§ Paul A. Clemons,*,‡,§ and
Stuart L. Schreiber*,†,‡,§
Contribution from the Howard Hughes Medical Institute, Department of Chemistry and
Chemical Biology, and The Eli and Edythe Broad Institute, Program in Chemical Biology,
HarVard UniVersity, 12 Oxford Street, Cambridge, Massachusetts 02138
Received March 30, 2004; E-mail: pclemons@hms.harvard.edu; stuart_schreiber@harvard.edu
Abstract: Systematic and quantitative measurements of the roles of stereochemistry and skeleton-
dependent conformational restriction were made using multidimensional screening. We first used diversity-
oriented synthesis to synthesize the same number (122) of [10.4.0] bicyclic products (B) and their
corresponding monocyclic precursors (M). We measured the ability of these compounds to modulate a
broad swath of biology using 40 parallel cell-based assays. We analyzed the results using statistical methods
that revealed illuminating relationships between stereochemistry, ring number, and assay outcomes.
Conformational restriction by ring-closing metathesis increased the specificity of responses among active
compounds and was the dominant factor in global activity patterns. Hierarchical clustering also revealed
that stereochemistry was a second dominant factor; whereas the stereochemistry of macrocyclic appendages
was a determinant for bicyclic compounds, the stereochemistry of the carbohydrates was a deter-
minant for the monocyclic compounds of global activity patterns. These studies illustrate a quantitative
method for measuring stereochemical and skeletal diversity of small molecules and their cellular
consequences.
shapes4 in order to assess the influence of stereochemistry and
macrocycle-based conformational constraints on biological
Introduction
The roles of stereochemistry and skeleton-based conforma-
tional constraints on the performance of small molecules in
biological assays can be intuited, but systematic and quantitative
measurements are lacking. Here we describe the use of diversity-
oriented synthesis (DOS),1 multidimensional screening,2 and
statistical and clustering analyses to achieve this goal. The
results, which reveal striking relationships, provide an explicit
illustration of how reagent selection in DOS can be predicted
to effect subsequent biological outcomes involving the resultant
products. This approach may facilitate an understanding of
relationships between chemical descriptor space and biological
measurement space (Figure 1).3
activities. We measured the latter using multidimensional
screening,5 (Figure 1) in which both precursor monocycles (M)
and their bicyclic products (B) were included in a three-
dimensional matrix of biological measurements whose axes
represent small molecules, a panel of four cell-based assay
measurements, and a panel of 18 cell states imposed by exposing
screening cell cultures to a collection of 17 known small-
molecule modulators of biological response (i.e., chemical-
genetic modifier screens6) or not (Figure 2b). To allow test
compounds maximal opportunity to perturb cell-based assays,
cellular assay measurements were chosen that report on broad
biological processes, rather than on highly targeted pathway-
or protein-specific events (Figure 2c). In contrast to conven-
tional, one-dimensional small-molecule screens, in which a
compound collection is typically exposed to a single set of assay
conditions, this approach results in vectors of measured values
Our plan involved the synthesis of a collection of carbohydrate-
derived small molecules differing in stereochemistry and having
two appendages that are either joined at their termini to form a
12-membered ring or not (Figure 2a). We chose a 12-membered
ring-forming reaction yielding macrolides having hexagonal
(4) Lee, D.; Sello, J. K.; Schreiber, S. L. J. Am. Chem. Soc. 1999, 121, 10648-
10649.
† Howard Hughes Medical Institute.
(5) See Supporting Information (Section V) for a description of multidimen-
sional screening and data analysis methods. A detailed description of
multidimensional screening (including conceptual basis, detailed assay
development and error-modeling methods, and role in hypothesis-genera-
tion) will be presented elsewhere: Lamenzo, J. O.; Dean, E. F.; Gorenstein,
J.; Kim, Y.-K.; Wagner, B. K.; Schreiber, S. L.; Clemons, P. A. Manuscript
in preparation.
‡ Department of Chemistry and Chemical Biology.
§ The Eli and Edythe Broad Institute.
(1) (a) Schreiber, S. L. Science 2000, 287, 1964-1969. (b) Burke, M. D.;
Schreiber, S. L. Angew. Chem., Int. Ed. 2004, 43, 46-58.
(2) Haggarty, S. J.; Koeller, K. M.; Wong, J. C.; Butcher, R. A.; Schreiber, S.
L. Chem. Biol. 2003, 10, 383-396.
(3) (a) Schreiber, S. L. Chem. Eng. News 2003, 81, 51-61. (b) Strausberg, R.
L.; Schreiber, S. L. Science 2003, 300, 294-295.
(6) Koeller, K. M.; Haggarty, S. J.; Perkins, B. D.; Leykin, I.; Wong, J. C.;
Kao, M.-C. J.; Schreiber, S. L. Chem. Biol. 2003, 10, 397-410.
9
14740
J. AM. CHEM. SOC. 2004, 126, 14740-14745
10.1021/ja048170p CCC: $27.50 © 2004 American Chemical Society

Diversity Relationships to Biological Measurements
A R T I C L E S
Figure 1. Glossary of terms relating experimental design and data analysis.
Figure 2. Experimental design to test roles of stereochemistry and ring closure on biological activities by multidimensional screening. (a) Structural variability
present in test library members due to appendages, carbohydrates, and ring closure. (b) Variables in multidimensional screening experiment, illustrating the
generation of a data matrix amenable to multidimensional data analysis. (c) Cell-based assay measurements chosen to provide broad biological interrogation.
for each compound and is formally analogous to other types of
profiling experiments, such as transcriptional or metabolic
profiling.
method allowed the application of clustering analysis, specif-
ically hierarchical clustering, to uncover relationships between
stereochemical and skeletal elements in determining biological
performance (Figure 1).
To analyze the results of these experiments, we took two
approaches. First, we grouped compounds from each of the M
and B compound classes according to the number of assays in
which each compound scored as positive. The resulting fre-
quency distributions were analyzed using a novel statistical test
that allowed experimental results to be compared to the results
of random permutation testing. This method allowed an explicit
statistical statement about the likelihood of observing our results
by chance. Second, we categorized compounds according to
stereochemical descriptors and computed average chemical-
genetic fingerprints for each category that encode their perfor-
mance in a multidimensional biological measurement space. This
Results and Discussion
Efficiency of Macrocyclic Ring-Closing Metathesis on
Model Substrates. We first determined reactivity patterns of
carbohydrate-based substrates when exposed to ring-forming
reactions. For ring-closing metathesis (RCM) reactions, R-sub-
stituents on the unsaturated ester appendages, independent of
their relative stereochemistry, have only minor effects on rates
and yields (Figure 3; 1b, 1c); however, allylic and vinylic
substitutions dramatically decrease the rate of ring closure (data
not shown). The carbohydrate-based six-membered ring
9
J. AM. CHEM. SOC. VOL. 126, NO. 45, 2004 14741

A R T I C L E S
Kim et al.
Figure 3. Role of stereochemistry in ring closure. Trans-stereochemistry of unsaturated diesters results in a ring-closing reaction yielding a “planar” bicycle;
cis- (versus trans-) stereochemistry provides faster cyclization and results in a “bent” bicycle.
Figure 4. Synthesis of test library of monocyclic and bicyclic compounds. (a) 5 mM DMSO stock solutions of the illustrated 122 bicyclic products (B) and
their 122 monocyclic precursors (M, structures not shown) were prepared as indicated, affording (b) products, each as nearly a single entity;9b three-
dimensional renderings are based on X-ray structure determination for analogous compounds. (c) Numbered building-block fragments corresponding to
superatoms R and R1 and encoded designations of products found in the completed library.
increases the overall efficiency of the ring closure regardless
of its ring elements and substitution patterns. Internal competi-
tive ring closure of mannose-derived tris-unsaturated ester 2a
revealed that cis-fused, conformationally constrained bicyclic
product 2b (79%) forms more rapidly than trans-fused product
2c (8%). Our reactivity studies also indicate that the ring closures
are irreversible under the conditions used.7
Library Synthesis. We next developed a DOS pathway
yielding 244 compounds, 122 having two acyclic, unsaturated
ester appendages, and 122 having an appended, 12-membered
ring resulting from joining the two terminal olefins by RCM
(Figure 4). In principle, all possible combinations of reagent
selections could have yielded 432 products. However, we
synthesized and screened only the 122 pairs of compounds
shown here, primarily due to the efficiency of the chemical
transformations. Six differentially bis-protected carbohydrate-
derived diols were synthesized in several steps from com-
mercially available, inexpensive starting materials8 and attached
selectively to macrobeads (Figure 4; I-VI, ca. 90 nmol/bead).9
Each set of macrobead-bound substrates was divided into four
(7) (a) Resubmission of each pure product to the reaction conditions did not
yield the other product. (b) The galactose-based triene showed the same
preference for cis-fused product formation.
(8) Diols were synthesized in two to five steps from commercially available
starting materials (<$5/g), proceeding in overall yields of 40-75%.
9
14742 J. AM. CHEM. SOC. VOL. 126, NO. 45, 2004

Diversity Relationships to Biological Measurements
A R T I C L E S
Figure 5. Global analysis of outcomes in biological measurement space. Codes for assay readouts (A-D) and known small-molecule modulators (0-17)
are the same as those in Figure 2. (a) Frequency of activity of bicyclic (light bars) or monocyclic (dark bars) compound classes in chemical-genetic assays.
Italicized numbers represent the number of compounds in each group. Smooth curves represent theoretical fits of biological measurement data for bicyclic
(brown curve) or monocyclic (green curve) classes. (b) Calculated molecular properties for bicyclic and monocyclic classes with entries expressed as the
mean ( SD over all 122 compounds in each class.13 (c) Forty-dimensional average chemical-genetic fingerprints for bicyclic (brown trace) or monocyclic
(green trace) classes, resulting from the multidimensional screen illustrated in Figure 1. All measurements were made in duplicate, and values represent
averages over all members (122 compounds; 244 measurements) for each of the two classes. Gray bars indicate assays for which class-average Z-scores
differ by >0.5 σ. (d) Heat-map representation of average chemical-genetic fingerprints for monocyclic and bicyclic classes, depicting both high-signal (red
blocks) and low-signal (blue blocks) statistical outliers. (e) Hierarchical clustering (using a linear correlation function) of 12 conformationally distinct categories
on the basis of biological measurements, showing both high-signal (red blocks) and low-signal (blue blocks) statistical outliers.15 Superatoms R and R1 are
defined essentially as in Figure 4, with R1 including the reagents {2-6,8,9}. Each row of blocks represents average scores (24 measurements) for 12
appendage-matched compounds with connectivity and stereochemistry defined as indicated on the dendrogram (B/M; dieq/diax; Glc ) glucose (I), Gal )
galactose (II), Man ) mannose (III); see Figure 4).
parts and treated under conditions leading to carbamates or
benzoates (Figure 4; R). Treatment with Pd(PPh3)4 and thiosali-
cylic acid in THF afforded 3,4- or 4,5-diol products quantita-
tively. Differentially substituted ω-pentenoic acids were coupled
to the diols under standard conditions, affording bis-pentenoate
products in high purity and yield (Figure 4; R1). A portion of
each of these diesters was subjected to the action of the second-
generation Grubbs’ catalyst,10 resulting in bicyclic products in
high yield. This library synthesis strategy yields equal numbers
of mono- and bicyclic products having twelve distinct stereo-
chemical patterns (Figure 4) resulting from the combination of
two (3,4- and 4,5-protected) substitution patterns, three carbo-
hydrate templates, and two R-substitution relationships. This
strategy is an illustration of assay measurements guiding the
synthesis design; since the objective is to assess the role of
stereochemical and skeletal diversity on biological outcomes,
we turned to the use of DOS to prepare sets of compounds
matched in numbers to their siblings in every respect.
Multidimensional Screening. 244 compounds were used to
create a matrix of 19 520 () 244 × 40 × 2) biological
measurements involving 40 parallel cell-based assays. Cell-based
assay readouts were chosen to assess broadly the consequences
of small-molecule perturbation of mammalian cells and include
measurements that report on DNA synthesis,11a cellular esterase
(9) (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) See Supporting Information (Section IV) for a detailed
description parallel library synthesis.
(10) Scholl, M.; Trnka, T. M.; Morgan, J. P.; Grubbs, R. H. Tetrahedron Lett.
1999, 40, 2247-2250.
9
J. AM. CHEM. SOC. VOL. 126, NO. 45, 2004 14743

A R T I C L E S
Kim et al.
activity,11b mitochondrial membrane potential,11c and intracel-
lular reducing equivalents.11d These four readouts were expanded
to 40 cell-based assays, each performed in duplicate, by exposing
cells simultaneously to library members and to small molecules
of known biological activity (i.e., chemical-genetic modifier
screens6). A matrix of data was created using 40 (of 72 possible)
combinations of the four readouts with 18 different cellular
conditions (see Figure 2b). In duplicate experiments, each
compound was independently assigned a signed (i.e., “+” or
“-”) Z-score corresponding to the number of standard deviations
it fell from the mean of a well-defined mock-treatment distribu-
tion. This distribution was determined to be consistent with
experimental noise observed under cell-based assay conditions.5
The resulting collection of continuous-valued Z-scores represents
the primary dataset used for further analysis. For analyses
dependent upon discrete (i.e., binned) outcome states, Z-score
data were further subjected to a threshold that resulted in each
measurement being scored as a high- or low-signal outlier, or
as a nonoutlier, from the mock-treatment distribution, based on
the probability that the measurement could be explained by assay
noise (Pnoise < 0.0006).5 To reduce further the possibility of
false-positive scoring, we required that any compound called a
positive be scored as a positive in both replicates for a given
assay, resulting in an overall average “hit” rate for the binned
dataset of less than 1%.
oral bioavailability.14 In that study, limitations on oral bioavail-
ability resulted from the interaction of small molecules with
complex biological processes involving numerous metabolic
enzymes, a situation comparable to the use of broad cell-based
assay measurements, as described herein.
In a preliminary attempt to determine which cell-based assays
might best distinguish between the M and B compound classes,
we represented 40-dimensional average chemical-genetic fin-
gerprints5 both as activity “signatures” (Figure 5c) and as a
mathematically equivalent heat map (Figure 5d). In these
visualizations, activities are represented as Z-scores averaged
over all members (and replicates) within the M and B compound
classes. These data represent two different visual approaches
to defining assays that discriminate between the two classes.
Noting that some assays result in greater variation of data than
others, chemical-genetic fingerprints based on a subset of these
cell-based assays might be used in future related studies. It is
also possible that a subset of these assays might be used as an
empirical surrogate for molecular properties related to confor-
mational flexibility (e.g., oral bioavailability).14
To expand the scope of our global analysis of the impact of
chemical diversity elements on biological outcomes, we prepared
average chemical-genetic fingerprints (see Figure 1) for each
of six stereochemical categories (for each of the M and B
classes) resulting from the combination of three carbohydrate
templates and two R-substitution relationships on the 12-
membered rings (or corresponding acyclic chains). To eliminate
bias in our analysis introduced by different appendage repre-
sentations between the categories, only 3,4-bis-pentenoates (and
their products) were considered in this analysis, resulting in 12
groups of 12 compounds each having a uniform representation
of appendages across all 12 groups. For each of these 12
categories, whose membership was based solely on the chemical
diversity elements of interest, average chemical-genetic finger-
prints were computed as before, and these fingerprints were
hierarchically clustered using linear correlation between fin-
gerprints as a similarity metric (Figure 5e).15 If categories with
a common structural element cluster together based on biological
performance, then there exists a correlation between this
structural element and global biological outcome. Conversely,
if structurally unrelated categories cluster based on performance,
there is no such correlation of biological outcome with stereo-
chemical diversity elements. Our observations suggest the
former scenario for this dataset. Notably, membership in the
M or B compound class has the dominant overall effect on
biological outcome (Figure 5e; similarity < 0.43), extending a
less complex result (based on a binary “hit” versus non-“hit”
metric; see Figure 5a) to a 40-dimensional biological measure-
ment space. Furthermore, the R-substitution relationship on the
12-membered rings determines subclustering with the B class
irrespective of carbohydrate template (Figure 5e, top branch of
dendrogram; similarity < 0.85). Each of these two subclusters
is further divided into trans-fused (Glc,Gal) and cis-fused (Man)
[10.4.0] bicyclic systems. In contrast, subclusters from the M
class are chiefly determined by the identity of the carbohydrate
template itself (Figure 5e, bottom branch of dendrogram;
similarity < 0.87).
Data Analysis. One of our primary observations was that
precursor monocycles (M) generally scored as positive in a
larger number of the 40 assays than did their bicyclic (B)
counterparts (Figure 5a). To test the statistical significance of
this result, we fit compound Gamma-Poisson model distribu-
tions to each dataset and applied a likelihood ratio test for
significance.12 The reported P-value (P < 0.0002; the probability
of observation in a random dataset) was obtained by permutation
testing; matched pairs of compounds were randomly assigned
labels B and M (one of each label per pair) before computing
the likelihood ratio test on each permuted dataset. For only 11
of more than 89 000 permuted datasets was the likelihood ratio
score for randomly permuted data higher than that for the
experimental data, indicating that our experimental results are
extremely unlikely to have occurred by chance. Overall, this
observation suggests that the propensity of these compounds
to score frequently as active in multidimensional screening is
reduced upon imposition of conformational constraints resulting
from the formation of the second ring.
In an attempt to rationalize these results in terms of calculable
properties of small molecules, we calculated certain average
molecular descriptor values we intuitively thought relevant to
conformational restriction, including the number of rotatable
bonds, for each of the M and B compound classes (Figure 5b).13
These tabular data show that the number of rotatable bonds, of
the descriptors computed, are best able to distinguish between
the M and B compound classes, a result reminiscent of a
previous report of the importance of this descriptor in predicting
(11) (a) Stockwell, B. R.; Haggarty, S. J.; Schreiber, S. L. Chem. Biol. 1999, 6,
71-83. (b) Parish, C. R. Immunol. Cell. Biol. 1999, 77, 499-508. (c) Wong,
A.; Cortopassi, G. A. Biophys. Res. Commun. 2002, 298, 750-754. (d)
Perrot, S.; Duterte-Catella, H.; Martin, C.; Rat, P.; Warnet, J. M. Toxicol.
Sci. 2003, 72, 122-129.
(12) See Supporting Information (Section VI) for a detailed description of the
Gamma-Poisson model fit and the likelihood ratio test.
(13) Molecular descriptor values were calculated using Pipeline Pilot (Scitegic,
Inc.; San Diego, CA).
(14) Veber, D. F.; Johnson, S. R.; Cheng, H.-Y.; Smith, B. R.; Ward, K. W.;
Kopple, K. D. J. Med. Chem. 2002, 45, 2615-2623.
(15) Hierarchical clustering and data visualization were performed using Spotfire
Decision Site 7.0 (Spotfire, Inc.; Somerville, MA).
9
14744 J. AM. CHEM. SOC. VOL. 126, NO. 45, 2004

Diversity Relationships to Biological Measurements
A R T I C L E S
Figure 6. Summary of results interpretation. (a) Table depicting the overall frequencies of activity of monocyclic (M) or bicyclic (B) compounds in chemical-
genetic assays. (b) Dominance of stereochemical features on biological outcomes before and after RCM.
In each of the M and B compound classes, average chemical-
genetic fingerprints for six categories (based on the stereochem-
istry of the 3-,5-positions of the carbohydrate and the R-position
of the appended ester) established the relative importance of
diversity elements to different degrees. Our rationale in using
aVerage chemical-genetic fingerprints across measurements was
to augment truly significant values while diminishing the impact
of both random and systematic error inherent in the collection
of cell-based assay data. Though a smaller number of assays
may have sufficed to discriminate between the M and B
compound classes (Figure 5c,d), a substantial fraction (37.5%)
of the 40 cell-based assays were necessary to rank stereochem-
ical elements with respect to their association with macrocy-
clization (data not shown), a substantially richer biological
measurement space than would be afforded (10%) using the
four cell-based assay readouts alone (i.e., without a chemical-
genetic modifier screening strategy). The addition of further
orthogonal measurements would likely be necessary to provide
significant resolution between additional diversity elements (e.g.,
appendage-based diversity elements) present in this library of
DOS-derived small molecules.
6a). The average activity frequency for active compounds of
the B class is 1.2 assays per compound, while that of the M
class is 2.3 assays per compound. Hierarchical clustering (Figure
5e) was further able to resolve this remarkable difference in
performance between M and B compound classes when groups
of compounds with common stereochemical elements were
taken together. Overall, relationships between stereochemical
elements within the M and B compound classes depends
primarily on M versus B class membership; whereas the
stereochemistry of macrocyclic appendages was the dominant
determinant for the B class, the stereochemistry of the carbo-
hydrate ring itself was the primary determinant for the M class
in global activity patterns (Figure 6b). These studies illustrate
a process for measuring the biological consequences of stereo-
chemical and skeletal diversity in small molecules and small-
molecule assays.
Acknowledgment. We gratefully acknowledge John Tallarico
and Max Narovlyansky for experimental advice and assistance
and Stephen Haggarty and Justin Lamb for helpful comments
on the manuscript. We thank the National Institute of General
Medical Sciences for support of this research and the National
Cancer Institute (NCI), the Keck Foundation, Merck KGaA,
and Merck & Co. for support of the Broad Institute Chemical
Biology Program (formerly ICCB), and the NCI for support of
the Initiative in Chemical Genetics. S.L.S. is an investigator
with the Howard Hughes Medical Institute at Harvard Univer-
sity.
Conclusion
Conformational restriction by macrocyclization has previously
been intuited to enhance binding affinity and metabolic stability
of the resulting products.16 Here, we report an experimental
demonstration of the effects of macrocyclization and stereo-
chemistry (on both the macroring precursors and the macrorings
themselves) using a set of closely related compounds prepared
by DOS and a multidimensional chemical-genetic approach.
Cyclic connectivity of an otherwise identical compound set
significantly influences compound performance in cell-based
assays. For example, 17 out of 19 active compounds from the
B compound class scored as a “hit” in exactly one of 40 assays,
whereas more than 60% of active compounds from the M
compound class scored as a “hit” in two or more assays (Figure
Note Added after ASAP Publication: This article was
published on the Web 10/20/2004 with minor errors in refs 9
and 11. The correct version posted 10/28/2004 and the print
version are correct.
Supporting Information Available: General experimental
procedures, spectral characterization data, assay protocols, assay
data, and data analysis methods. This material is available free
of charge via the Internet at http://pubs.acs.org.
(16) Burger, M. T.; Bartlett, P. A. J. Am. Chem. Soc. 1997, 119, 12697-12698
and references therein.
JA048170P
9
J. AM. CHEM. SOC. VOL. 126, NO. 45, 2004 14745