Selection of DNA-Encoded Dynamic Chemical Libraries for Direct Inhibitor Discovery

Article abstract of DOI:10.1002/anie.202005070

Dynamic combinatorial libraries (DCLs) is a powerful tool for ligand discovery in biomedical research; however, the application of DCLs has been hampered by their low diversity. Recently, the concept of DNA encoding has been employed in DCLs to create DNA-encoded dynamic libraries (DEDLs); however, all current DEDLs are limited to fragment identification, and a challenging process of fragment linking is required after selection. We report an anchor-directed DEDL approach that can identify full ligand structures from large-scale DEDLs. This method is also able to convert unbiased libraries into focused ones targeting specific protein classes. We demonstrated this method by selecting DEDLs against five proteins, and novel inhibitors were identified for all targets. Notably, several selective BD1/BD2 inhibitors were identified from the selections against bromodomain 4 (BRD4), an important anti-cancer drug target. This work may provide a broadly applicable method for inhibitor discovery.

Full text of DOI:10.1002/anie.202005070

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Selection of DNA-encoded Dynamic Chemical Libraries for Direct
Inhibitor Discovery
Authors: Yuqing Deng, Jianzhao Peng, Feng Xiong, Yinan Song, Yu
Zhou, Jianfu Zhang, Fong Sang Lam, Chao Xie, Wenyin
Shen, Yiran Huang, Ling Meng, and Xiaoyu Li
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202005070
Link to VoR: https://doi.org/10.1002/anie.202005070

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
Selection of DNA-encoded Dynamic Chemical Libraries for Direct
Inhibitor Discovery
Yuqing Deng,  [a] Jianzhao Peng,  [a,b] Feng Xiong,[a] Yinan Song,[a] Yu Zhou,[a] Jianfu Zhang,[a] Fong
Sang Lam,[a] Chao Xie,[a] Wenyin Shen,[a] Yiran Huang,[a] Ling Meng,[a] and Xiaoyu Li *[a]
[a]
Y. Deng, J. Peng, Dr. F. Xiong, Y. Song, Dr. Y. Zhou, J. Zhang, F. S. Lam, C. Xie, W. Shen, Y. Huang, L. Meng, and Prof. X. Li
Department of Chemistry and the State Key Laboratory of Synthetic Chemistry, The University of Hong Kong
Laboratory for Synthetic Chemistry and Chemical Biology of Health@InnoHK of Innovation and Technology Commission
Pokfulam Road, Hong Kong SAR, China
xiaoyuli@hku.hk
[b]
J. Peng
Department of Chemistry
Southern University of Science and Technology China
1088 Xueyuan Road, Shenzhen, China
 These authors contributed equally to this work.
Supporting information for this article is given via a link at the end of the document.
Abstract: Dynamic combinatorial library (DCL) is a powerful tool for
ligand discovery in biomedical research; however, the development of
DCL has been hampered by its low diversity. Recently, the concept of
DNA encoding has been employed in DCL to create DNA-encoded
dynamic libraries (DEDLs); however, all current DEDLs are limited to
fragment identification, and a challenging process of fragment linking
is required after selection. We report an anchor-directed DEDL
approach that can identify full ligand structures from large-scale
DEDLs. This method is also able to convert unbiased libraries to
focused ones targeting specific protein classes. We demonstrated this
method by selecting DEDLs against five proteins, and novel inhibitors
have been identified for all targets. Notably, several selective
BD1/BD2 inhibitors were identified from the selections against BRD4
(bromodomain 4), an important anti-cancer drug target. This work may
provide a broadly applicable method for inhibitor discovery.
DNA-encoded chemical library (DEL), originally proposed by
Brenner and Lerner,[6] has recently become an important
technology in drug discovery.[7] Similar to DCL, DEL also employs
mixed compounds in library processing; however, DELs can
contain many billions to even trillions of compounds since each
compound is encoded with a unique DNA tag, and the library
selection can be efficiently decoded with PCR amplification and
DNA sequencing.[7a] Therefore, introducing DNA encoding to DCL
should be a viable strategy to enable the preparation and
selection of large DCLs.[7k] Previously, the Neri group developed
an elegant dual-pharmacophore DEL approach named Encoded
8]
Self-Assembling Chemical (ESAC) library,[7c,
and we
dual-
incorporated
dynamic
DNA
hybridization
with
pharmacophore DELs and developed a DNA-encoded dynamic
library (DEDL) method that can process large dynamic libraries.[9]
Zhang and co-workers also reported two dynamic DEL
methods.[10] However, all current DEDL methods are limited to
fragments or fragment pairs, and a post-selection tethering
process is still required to elaborate the fragments into full ligands.
As seen in many studies, finding a suitable linker connecting the
fragments to achieve cooperative binding is highly challenging,
which is empirical, tedious, and involves a lot of trial and error.[8b,
11] In fact, fragment linking often takes considerably more efforts
than fragment identification itself. It is highly desirable to be able
to directly obtain full ligand structures from library selections.
Here, we report a DEDL method without the need for post-
selection fragment linking. This method is based on anchor-
directed dynamic exchange and an in situ hit isolation strategy.
Notably, existing DELs could be used in this method to create
large dynamic libraries. Moreover, DEL is generally used as a
one-for-all discovery platform without target bias. Recently,
several studies have exploited the “focused” DELs to target
specific protein classes;[12] however, they all require the
preparation of a new library for each target. Here, we show that
this method has a “plug-and-play” feature that can convert the
general-purpose libraries to focused ones targeting specific
proteins by switching the anchor. To demonstrate this method, we
selected DEDLs against five proteins, and a series of potent and
selective inhibitors were identified and validated for these targets.
Introduction
Screening large-scale chemical libraries against biological
targets to identify novel ligands is a central strategy in drug
discovery and many other research fields. In traditional high
throughput screening, compounds are spatially encoded and
screened against the target individually. In contrast, dynamic
combinatorial library (DCL) employs a mixture of building blocks
(BBs) that undergo dynamic exchange through reversible
chemical reactions, allowing the synthesis and screening of all
library compounds in one pot.[1] With DCL, the target acts as the
template to promote the formation of high-affinity ligands at the
expense of non-binders, and “hit” compounds can be identified by
comparing the equilibria with and without the target. In the past
decade, DCL has shown wide utilities in numerous
applications.[1d-h, 2] However, with a few exceptions,[3] most DCLs
only contain a few hundred compounds or less, mainly due to the
lack of techniques to resolve large number of compounds in a
1h, 4]
single mixture.[1a,
Indeed, low library diversity has been
considered as the major limitation of DCL.[1a, 1h, 4-5]
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
Results and Discussion
Fig. 2. a) Structure of DNA-conjugated amines. b)-c) In-gel fluorescence
analysis of the PN-1/SN-1 reactions with FAM-NHS. d)-e) In-gel fluorescence
analysis of the reactions of various PN/SNs with FAM-NHS. PN/SN: 3 μM;
buffer: MES (pH 6.0), MOPS (pH 7.0), HEPES (pH 9.8). In b): r.t., 1 h; in c): pH
= 8.0; in d): r.t., 500-fold FAM-NHS, 1 h, pH 8.0. Gel: denaturing PAGE. FN: a
15-nt fluorescent DNA as the internal standard. Other conditions are as marked.
See Fig. S1 for ethidium bromide stained gels.
Fig. 1. a) The proposed library selection scheme. An anchor and a DNA-
encoded “BB library“ form the DEDL. After target addition and imine reduction,
bio-NHS and streptavidin beads are used to capture the primary amines. Hit
compounds in the flow-through are isolated and decoded by PCR and DNA
sequencing. b) Species captured by biotin-NHS.
Our strategy is shown in Fig. 1. An “anchor” with an aldehyde
group is incubated with a DNA-encoded “building block (BB)
library”, which could be a regular DEL with a primary amine. The
DNA-encoded BBs compete for the anchor through reversible
imine formation, thereby forming a DEDL. Adding the target shifts
the equilibrium and promotes the formation of high-affinity binders,
and the dynamic exchange can be stopped by NaBH3CN-
mediated imine reduction. To decode the “anchor-BB conjugates”
that bind the target, we devised an in situ hit isolation strategy:
after the reduction, a biotin sulfo-N-hydroxy succinimidyl ester
(bio-NHS; Fig. 1a) is added to capture the primary amine on the
BBs, while the “anchor-BB conjugates” have a less reactive
secondary amine due to the steric hindrance and are less
efficiently captured.[13] As a result, the selected ligands could be
separated from the biotinylated species (non-binders, binders not
conjugated with the anchor, and anchor-independent binders; Fig.
1b) with streptavidin beads and then subjected to hit
deconvolution. Importantly, no-anchor and non-target control
selections will be performed to control for anchor-independent
binding and other variables in the selection process.[9a] This
design requires a known binder as the anchor, but it does not
need to be a high-affinity ligand; instead, the anchor could simply
be a small fragment with weak affinity or a structural moiety
derived from the protein’s natural substrate. Therefore, this
method may be particularly suitable for the “affinity maturation” for
the proteins with known but unoptimized binders.[8b, 14] Moreover,
DEL is mostly a binding assay; here, using an anchor may
facilitate inhibitor discovery, since the anchor could guide the
library compounds to be sampled at the site with functional
relevance (e.g. the catalytic pocket of enzymes). Finally, this
method does not need protein immobilization and is compatible
with unmodified, in-solution proteins.
amines may also be captured. We reason that using a strong
capture condition to overcome the structural variation of BBs may
be beneficial. Although secondary amines would be captured, it is
in a lesser degree and should retain sufficient materials for hit
decoding. To test this, we prepared two DNAs with a simple
primary and secondary amine, respectively (PN-1/SN-1, 41-nt/25-
nt; Fig. 2a); they were mixed at equal ratio and reacted with
excess of an activated fluorescein ester (FAM-NHS). Strikingly,
little capture of the secondary amine was observed (Fig. 2b), while
PN-1 could be efficiently labeled at pH >7. We decided to choose
pH 8.0, as it was reported to be optimal for reductive amination
under physiological conditions.[15] The reaction time and
temperature were varied and again little capture of SN-1 was
observed (Fig. 2c). Next, several different primary and secondary
amines were tested and the selective capture was also observed
(PN-2, 3, 4/SN-1, 2, 3; Fig. 2d). Since the pKa’s of primary and
secondary amines are above 10, the selectivity was presumably
due to the steric bulk of the secondary amines, rather than the
difference in protonation. In fact, SN-2, which has relatively low
steric hindrance, gave more capture products (red arrow; Fig. 2d).
SN-3 represented the compound in the actual DEDL; it has a
crowded secondary amine structure and also did not show any
capture (Fig. 2d). Although BB diversity is expected to affect the
capture efficiency, these results indicated there is a large
selectivity window. Importantly, a non-target selection will always
be conducted to control for this variation.
Next, we tested the selection with a model protein CA-II
(carbonic anhydrase II). Since arylsulfonamides are well-known
CA-II binders, a 4-formylbenzenesulfonamide was used as the
anchor (A-1; Fig. 3a). First, two DNA-conjugated primary amines
(PN-1 and PN-5; Fig. S2) were prepared. PN-5 has a phenyl-
glycine motif known to enhance the binding affinity of
sulfonamide.[16] The two DNAs were mixed with A-1 to form the
imines and incubated with CA-II; a 10:1:1 library/anchor/target
ratio was used to create a competitive environment for PN-1/PN-
5. After the imine reduction, bio-NHS capture, and streptavidin
A key to this method is the selective capture of the primary
amines, which has two potential issues: a) the structural diversity
of BBs may result in variations in capture efficiency; the less
efficiently captured BBs would be “enriched” in the flow-through
and become false positives; b) albeit less reactive, the secondary
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
Fig. 3. a) Selection of a model DEDL against CA-II. Library, 1 µM; CA-II, 0.1
µM; A-1, 0.1 µM; incubation: 4 ºC, 2 h in 50 mM MOPS (pH 8.0); NaBH3CN
reduction: 38.5 mM, 1 h; bio-NHS: 100 eq., 1 h, 37 ºC. b) Scatter plot of the
selection results. Red: hits; orange: negatives; open circle: compounds also
enriched in control selections (Fig. S3). x-axis: post-selection sequencing
counts; y-axis: enrichment fold = (post-selection%)/(pre-selection%). c) CA-II
inhibition assay. Error bar represent 3 repeats. d) Compound structures. See
the SI for experimental details.
pulldown, the surviving PN-1/PN-5 in the flow-through were
quantified using qPCR. The results showed a ~40-fold increase
of PN-5, indicating that the “A-1 + PN-5” conjugate was enriched
by CA-II (Fig. S2). Next, we prepared a 440-member dipeptide
library (Fig. 3a), which formed a DEDL with A-1 and selected
against CA-II. The DNA tags of the selected compounds in the
flow-through were amplified and decoded with the next-
generation sequencing (NGS). The sequencing data were
processed to calculate the post-selection sequence count (SC)
and the enrichment fold (EF) for each compound as previously
described.[9] The selection results were plotted in the form of
scatter plots (Fig. 3b) and the compounds with high EF and SC
were considered as potential hits. No-anchor and non-target (with
bovine serum albumin; BSA) control selections were also
conducted (Fig. S3). By comparing the scatter plots, the
compounds also enriched in control selections were identified and
excluded (open circles; Fig. 3b). Four specifically enriched
compounds (1-4) were chosen as “hits”, resynthesized off-DNA,
and assayed for their inhibition activity against CA-II. In general,
all hits exhibited enhanced activity compared with A-1 (Fig. 3c).
Interestingly, 3 hits have the same phenylglycine motif as PN-5,
and the other one has a valine residue, also known to increase
the CA-II-binding affinity.[16] To verify the selection specificity, two
“negatives” with low EF were tested and both exhibited very low
activity (N1 and N2; Fig. 3d). Moreover, to verify that CA-II in fact
shifted the equilibrium to form the imine, using 2 and 4 as the
example, we incubated CA-II with A-1 and the BBs of 2 and 4,
respectively.[17] After the equilibrium was locked by reduction, the
product formation was analyzed with LC-MS. As shown in Fig. S4,
CA-II significantly promoted the product formation, while nearly no
product was observed without CA-II. Furthermore, assuming the
quantity of the amine products reflected the imine formed in the
equilibrium, the binding affinity of the imines was estimated based
on the LC peak integration, which were further compared with the
affinity (Kd) of 2 and 4, determined with fluorescence polarization
Fig. 4. a) The structure of the anchors and BB-1 library. b) The anchors’ activity
against BD1/BD2. c) Scatter plots of the selection results. Red: hits enriched by
both BD1 and BD2; orange: hits enriched by BD1 or BD2; open circles:
compounds also enriched in control selections (Fig. S7). d)-f) BD1/BD2 assay
results of the hits, anchors, non-specifics, and the BB-only controls. n.d.: not
detectable. IC50 fitting curves are shown in Fig. S8. g) Structures of
representative hit compounds. See the SI for experimental details and the full
list of compound structures (Fig. S9).
(Fig. S5). The results showed that the two species had similar CA-
II-binding affinity.
Next, we moved on to large libraries. A 2-BB DEL (BB-1;
67,600 dipeptides) was prepared using the split-and-mix DEL
encoding method (Fig. S6).[18] First, BB-1 was selected against
the BD1/BD2 domains of BRD4 (bromodomain 4), an important
epigenetic regulator that functions through binding to the
acetylated lysine on histones.[19] Despite the high similarity, BD1
and BD2 have distinct cellular functions due to their interactions
with different lysine-acetylated histones or transcriptional
proteins;[20] thus, it is highly desirable to identify selective
BD1/BD2 inhibitors. We prepared three BD1/BD2 anchors (A-2,
A-3, A-4; Fig. 4a). A-2 is derived from acetylated lysine, the
natural BRD4-binding structure; A-3 has an isoxazole ring, which
is a “privileged” BRD4-binding structure;[21] A-4 is based on the
well-known, highly potent BRD4 inhibitor (+)-JQ-1.[22] The IC50’s
of the anchors were determined using a TR-FRET assay, and
they range from high µM to nM (Fig. 4b). Thus, these anchors
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
should be able to assess the method across a wide range of
binding affinities.
The anchors were mixed with BB-1 to form 3 dynamic libraries
and selected against BD1 and BD2, respectively. Again, non-
target and no-anchor controls were performed for all selections.
The post-selection sequencing data were processed as same as
in Fig. 3 and scatter plots are shown in Fig. 4c. In general, the
selections exhibited relatively high number of surviving DNA,
probably because the compounds were enriched in flow-through,
where the background DNAs might not be completely removed by
streptavidin beads and were also PCR-amplified. Nevertheless,
the specifically enriched compounds could still be identified by
comparing with the control selections (Fig. S7). Interestingly, for
each anchor, one compound was enriched by both BD1 and BD2
(2-1, 3-1, and 4-1; Fig. 4c), strongly suggesting that they were
specific binders. Next, the selected hits were re-synthesized and
assayed against BD1/BD2. As shown in Fig. 4d-4f, most
compounds showed enhanced activities compared with the
anchor. For the weakest anchor A-2, the greatest increase was
observed with 2-1 for BD2 (from 178.7 µM to 12.0 µM). For A-3,
more significant increase was observed with 3-1; the IC50 was
reduced from mid-µM’s to 1.55 µM (BD1) and 1.46 µM (BD2). For
anchor A-6, which was already a nM inhibitor, a ~9-fold increase
was still obtained (4-1 for BD1; from 173.2 nM to 19.8 nM). To
verify the selection specificity, we tested several compounds also
enriched in control selections (2-n1, 3-n1, 4-n1) and a negative
compound with low EF (2-n2); the results showed that they had
no activity or lower activity than the anchor (Fig. 4d-4f). In addition,
all the “BB-only” compounds without the anchor motif were
completely inactive. The selection data actually reflected the
combined effect of three factors: target binding, imine
formation/reduction, and biotin-based hit separation. Considering
the structural diversity of BBs, all three are expected to affect the
selection results. The non-target (with BSA) and no-anchor
selections (Fig. S7) were used to control for non-specific binding
and the variations in biotin capping/hit separation. Thus, to control
for the effect from the differing imine reactivities, we conducted a
no-target selection with the BB-1 library and the three anchors,
respectively (Fig. S10a); a simple biotin capping control was also
performed to delineate the primary amine reactivity (Fig. S10b).
The results showed that the identified BD-1 ligands in Fig. 4c were
not significantly enriched in these controls, and they all had similar
EF values and sequence counts, indicating that their enrichments
in the actual BB-1 selection were not significantly altered by the
difference in imine reactivities (Fig. S11).
Fig. 5. a) Five anchors (A-2 to A-6) were used to form 5 dynamic libraries with
BB-2. b) BD1/BD2 inhibition assay of A-5 and A-6. c) Scatter plots of the BD1
selection results. Orange: hits; open circles: non-specific compounds also
enriched in control selections (Fig. S12). c) BD1 assay results of the hits,
anchors, non-specifics, and BB-only controls. n.d.: not detectable. Several
compounds were assayed against BD2 and the IC50’s are shown below the
column graph. Fitting curves are shown in Fig. S13. d) Structures of the
representative compounds. See Fig. S14 for the full list.
than the ones from BB-1. With BD1, the potency of the hit
compounds was similar to the ones from BB-1 (Fig. 5d).
Surprisingly, the selection with the high-affinity A-4 did not identify
any specific binders. With BD2, only one specific binder was
identified (A-3; Fig. S15). We reasoned this might be due to the
larger size of the BB-2 compound; the tripeptide may dominate
target-binding and have attenuated the anchor’s directing effect
to the acetyl-lysine-binding site on BD1/BD2. Nevertheless,
although the IC50’s of the BD1 inhibitors were in µM range, they
exhibited good selectivity against BD2. Compounds 3-5 gave
~20-fold selectivity for BD1 and the others were inactive against
BD2 (Fig. 5d), while the anchors alone showed very low selectivity.
In addition, the compounds also enriched in the control selections
or with low EF showed little activity. These results indicated that
the method may be used to improve the target selectivity of the
ligands, but large compound size appeared to be detrimental for
the anchor-directed target-binding.
Collectively, these selections have demonstrated the
capability of this method in ligand optimization across a wide
range of binding affinity. For example, the potency of the weak
anchor A-2 was improved from high µM to low µM and the IC50 of
A-3 was reduced to single-digit µM, making them much more
suitable for further optimization.
Next, we prepared a 3-BB tripeptide DEL (BB-2; 17.576-million
tripeptides), considering that it may identify more potent binders.
We also prepared two more anchors (A-5, A-6; Fig. 5a), which
were similar to A-3 but with different heterocycles.[21] Five DEDLs
were formed using BB-2 with the anchors and selected against
BD1 and BD2, respectively. The results are shown in Fig. 5c and
the specific binders were identified by comparing with the control
selections (Fig. S12). However, albeit with greater diversity, the
inhibitors identified from BB-2 were not distinctively more potent
Usually, DELs are designed as general-purpose libraries with
little consideration of the ligand space of the specific target. With
this method, the same BB library could be directed to bias
different proteins by changing the anchor. To test this, we
conducted selections against two more proteins: acetyl-
cholinesterase (AChE) and X-linked inhibitor of apoptosis protein
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
(XIAP). AChE is
a
key hydrolyzing enzyme for the
target binding, imine formation/reduction, and biotin-based hit
separation. Considering the vast BB diversity in a DEL, BB-
dependent variations are expected for all three factors. Our data
suggested that non-target (using BSA or other unrelated proteins)
and no-anchor control selections were essential to exclude the
potential false positives arising from non-specific binding and the
variation in biotin capture, and the no-target selection was
important to control for the differing imine reactivities. Finally, this
method is limited to the optimization of known ligands. Future
studies will focus on the method development to realize
completely de novo ligand discovery.
neurotransmitter acetylcholine, and A-7 (Fig. S16a) was prepared
based on tacrine, a ligand that binds to the catalytic pocket of
AChE.[23] The selection using BB-2/A-7 against AChE identified
two compounds, and they exhibited a 3- and 18-fold of
improvement of activity over A-7 (7-1 and 7-2; Fig. S16). XIAP is
an inhibitor of cell apoptosis and has been pursued as an anti-
cancer drug target. The tetra-peptides with a sequence of N-
methyl-Ala-t-Leu-Pro-Xaa are potent XIAP antagonists.[24] Based
on this sequence, we prepared an XIAP anchor by truncating the
key proline and the 4th amino acid and replacing them with a
benzaldehyde (A-8; Fig. S17a); as expected, A-8 showed very
weak affinity (>600 µM; Fig. S17).[25] The selection of BB-2/A-8
against XIAP identified two compounds with significantly
improved affinity (8-1 and 8-2; 32.9 µM and 53.2 µM; Fig. S17). In
addition, their corresponding BBs without A-8 and a non-specific
compound 8-n1, showed no or very weak affinity, which verified
the selection specificity and that A-8 was required for binding (Fig.
S17d). Although the identified compounds in these selections
were weak, they all exhibited considerably improved potency
towards the target, indicating that the same BB library could be
used for different proteins by changing the anchor.
Acknowledgements
This work was supported by grants from the Research Grants Council
of the Hong Kong SAR (AoE/P-705/16, 17321916, 17302817,
17301118, 17111319), Laboratory for Synthetic Chemistry and
Chemical Biology of Health@InnoHK of ITC, HKSAR, grants from the
National Natural Science Foundation of China (21572014, 21877093,
91953119). We thank Prof. Yizhou Li at Chongqing University for the
mass analysis support and the the Centre for PanorOmic Sciences
(CPOS) Genomics Core at HKU for NGS support.
Keywords: DNA-encoded chemical library • dynamic combinatorial
library • high-throughput screening • drug discovery • DNA
Conclusion
[1] a) S. Ladame, Org. Biomol. Chem. 2008, 6, 219-226; b) M. F.
Schmidt, J. Rademann, Trends Biotechnol. 2009, 27, 512-521;
c) F. B. Cougnon, J. K. Sanders, Acc. Chem. Res. 2012, 45,
2211-2221; d) J. Li, P. Nowak, S. Otto, J. Am. Chem. Soc. 2013,
135, 9222-9239; e) A. Herrmann, Chem. Soc. Rev. 2014, 43,
1899-1933; f) M. Mondal, A. K. Hirsch, Chem. Soc. Rev. 2015,
44, 2455-2488; g) J. M. Lehn, Angew. Chem. Int. Ed. 2015, 54,
3276-3289; h) P. Frei, R. Hevey, B. Ernst, Chem. Eur. J. 2019,
25, 60-73.
[2] M. Mondal, A. K. Hirsch, Chem. Soc. Rev. 2015, 44, 2455-2488.
[3] a) D. A. Erlanson, J. W. Lam, C. Wiesmann, T. N. Luong, R. L.
Simmons, W. L. DeLano, I. C. Choong, M. T. Burdett, W. M.
Flanagan, D. Lee, E. M. Gordon, T. O'Brien, Nat. Biotechnol.
2003, 21, 308-314; b) B. R. McNaughton, B. L. Miller, Org. Lett.
2006, 8, 1803-1806; c) R. F. Ludlow, S. Otto, J. Am. Chem. Soc.
2008, 130, 12218-12219; d) B. R. McNaughton, P. C. Gareiss,
B. L. Miller, J. Am. Chem. Soc. 2007, 129, 11306-11307; e) P.
C. Gareiss, K. Sobczak, B. R. McNaughton, P. B. Palde, C. A.
Thornton, B. L. Miller, J. Am. Chem. Soc. 2008, 130, 16254-
16261.
In summary, we have developed a method to create and select
large DNA-encoded dynamic libraries for ligand optimization.
Compared with other approaches, this method avoids the post-
selection fragment linking and full ligands could be directly
obtained from the selection. Although a primary amine is required,
it could be feasibly incorporated in library synthesis using regular
amino acids, which are the most commonly used BBs in DELs. In
addition, this method does not require special library design; in
principle, existing DELs prepared with different methods or
encoded with double-stranded DNA tags could also be used, as
long as the library contains a primary amine.[8a, 8b, 18a, 26] One
caveat of this method is that the actual binders in the selection
are imines, while the hit compounds are isolated and
characterized as amines; thus, although the imine
formation/reduction chemistry has been widely used in numerous
DCL studies,[27] an assumption is taken that the amines largely
represent the properties of their imine counterparts. In addition, in
principle, some of the other well-developed reversible reactions in
DCL may be implemented to diversify the ligand structure.[1h]
The data have shown that this method could be used for ligand
optimization in a wide range of affinities; however, the results from
the BB-2 library indicated that the large compound size may offset
the anchor’s directing effect. Thus, using BB libraries with more
compact structures[28] may be more favorable; although this would
limit the library size, a recent study showed that focused DELs
with 2 sets of BBs were sufficient to identify high-quality
inhibitors.[12a] Moreover, in previous reports on focused DELs,[12]
the target-biasing elements were permanently built-in, so that the
library was less suitable for other types of targets. Here, this
method converts the unbiased DELs to focused ones by simply
changing the anchor in the selection.
[4] D. A. Erlanson, S. W. Fesik, R. E. Hubbard, W. Jahnke, H. Jhoti,
Nat. Rev. Drug Discov. 2016, 15, 605-619.
[5] P. Frei, L. J. Pang, M. Silbermann, D. Eris, T. Muhlethaler, O.
Schwardt, B. Ernst, Chem. Eur. J. 2017, 23, 11570-11577.
[6] a) S. Brenner, R. A. Lerner, Proc. Nat. Acad. Sci. USA 1992, 89,
5381-5383; b) J. Nielsen, Brenner, S., Janda, K.D., J. Am.
Chem. Soc. 1993, 115, 9812-9813.
[7] a) R. A. Goodnow, A handbook for DNA-encoded chemistry:
theory and applications for exploring chemical space and drug
discovery, John Wiley & Sons, Inc., Hoboken, New Jersey,
2014; b) T. Kodadek, N. G. Paciaroni, M. Balzarini, P. Dickson,
Chem. Commun. 2019, 55, 13330-13341; c) D. Neri, R. A.
Lerner, Annu. Rev. Biochem. 2018, 87, 479-502; d) V. Kunig, M.
Potowski, A. Gohla, A. Brunschweiger, Biol. Chem. 2018, 399,
691-710; e) B. Shi, Y. Zhou, Y. Huang, J. Zhang, X. Li, Bioorg.
Med. Chem. Lett. 2017, 27, 361-369; f) L. H. Yuen, R. M.
Franzini, ChemBioChem 2017, 18, 829-836; g) R. M. Franzini,
C. Randolph, J. Med. Chem. 2016; h) K. D. Janda, Chem. Eng.
News 2017, 95, 4-4; i) Y. Huang, O. Savych, Y. Moroz, Y. Y.
Chen, R. A. Goodnow, Aldrichim Acta 2019, 52, 75-87; j) M.
On another aspect, this method does not require protein
immobilization, and all the selections were performed in solution.
We envision this method may be suitable for selections in more
complex biological environment. However, this in-solution
selection reflected the combination of three implicating factors:
Song,
G.
T.
Hwang,
J.
Med.
Chem.
2020,
DOI:10.1021/acs.jmedchem.1029b01782; k) F. V. Reddavide,
M. Thompson, L. Mannocci, Y. X. Zhang, Aldrichim Acta 2019,
52, 63-74.
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
[8]
a) S. Melkko, J. Scheuermann, C. E. Dumelin, D. Neri, Nat.
Biotechnol. 2004, 22, 568-574; b) M. Wichert, N. Krall, W.
Decurtins, R. M. Franzini, F. Pretto, P. Schneider, D. Neri, J.
Scheuermann, Nat. Chem. 2015, 7, 241-249; c) J.
Scheuermann, D. Neri, Curr. Opin. Chem. Biol. 2015, 26, 99-
103.
[22] P. Filippakopoulos, J. Qi, S. Picaud, Y. Shen, W. B. Smith, O.
Fedorov, E. M. Morse, T. Keates, T. T. Hickman, I. Felletar, M.
Philpott, S. Munro, M. R. McKeown, Y. Wang, A. L. Christie, N.
West, M. J. Cameron, B. Schwartz, T. D. Heightman, N. La
Thangue, C. A. French, O. Wiest, A. L. Kung, S. Knapp, J. E.
Bradner, Nature 2010, 468, 1067-1073.
[9]
a) Y. Zhou, C. Li, J. Peng, L. Xie, L. Meng, Q. Li, J. Zhang, X.
D. Li, X. Li, X. Huang, X. Li, J. Am. Chem. Soc. 2018; b) G. Li,
W. Zheng, Z. Chen, Y. Zhou, Y. Liu, J. Yang, Y. Huang, X. Li,
Chem. Sci. 2015, 6, 7097-7104.
[23] W. G. Lewis, L. G. Green, F. Grynszpan, Z. Radic, P. R. Carlier,
P. Taylor, M. G. Finn, K. B. Sharpless, Angew. Chem. Int. Ed.
2002, 41, 1053-1057.
[24] T. K. Oost, C. Sun, R. C. Armstrong, A. S. Al-Assaad, S. F. Betz,
T. L. Deckwerth, H. Ding, S. W. Elmore, R. P. Meadows, E. T.
Olejniczak, A. Oleksijew, T. Oltersdorf, S. H. Rosenberg, A. R.
Shoemaker, K. J. Tomaselli, H. Zou, S. W. Fesik, J. Med. Chem.
2004, 47, 4417-4426.
[10] a) F. V. Reddavide, W. Lin, S. Lehnert, Y. Zhang, Angew. Chem.
Int. Ed. 2015, 54, 7924-7928; b) F. V. Reddavide, M. Cui, W.
Lin, N. Fu, S. Heiden, H. Andrade, M. Thompson, Y. Zhang,
Chem. Commun. 2019, 55, 3753-3756.
[11] a) M. Bigatti, A. Dal Corso, S. Vanetti, S. Cazzamalli, U. Rieder,
J. Scheuermann, D. Neri, F. Sladojevich, ChemMedChem 2017,
12, 1748-1752; b) S. Chung, J. B. Parker, M. Bianchet, L. M.
Amzel, J. T. Stivers, Nat. Chem. Biol. 2009, 5, 407-413; c) A. W.
Hung, H. L. Silvestre, S. Wen, A. Ciulli, T. L. Blundell, C. Abell,
Angew. Chem. Int. Ed. 2009, 48, 8452-8456; d) J.
Scheuermann, C. E. Dumelin, S. Melkko, Y. X. Zhang, L.
Mannocci, M. Jaggi, J. Sobek, D. Neri, Bioconjug. Chem. 2008,
19, 778-785; e) G. Zimmermann, U. Rieder, D. Bajic, S. Vanetti,
A. Chaikuad, S. Knapp, J. Scheuermann, M. Mattarella, D. Neri,
Chem. Eur. J. 2017, 23, 8152-8155; f) F. Abendroth, A. Bujotzek,
M. Shan, R. Haag, M. Weber, O. Seitz, Angew. Chem. Int. Ed.
2011, 50, 8592-8596; g) J. P. Daguer, C. Zambaldo, M. Ciobanu,
P. Morieux, S. Barluenga, N. Winssinger, Chem. Sci. 2015, 6,
739-744.
[25] a) Z. Nikolovska-Coleska, R. Wang, X. Fang, H. Pan, Y. Tomita,
P. Li, P. P. Roller, K. Krajewski, N. G. Saito, J. A. Stuckey, S.
Wang, Anal. Biochem. 2004, 332, 261-273; b) F. Cossu, F.
Malvezzi, G. Canevari, E. Mastrangelo, D. Lecis, D. Delia, P.
Seneci, C. Scolastico, M. Bolognesi, M. Milani, Protein Sci.
2010, 19, 2418-2429.
[26] a) M. Potowski, V. B. K. Kunig, F. Loscha, A. Brunschweiger,
MedChemComm 2019, 10, 1082-1093; b) Y. Li, P. Zhao, M.
Zhang, X. Zhao, X. Li, J. Am. Chem. Soc. 2013, 135, 17727-
17730; c) A. Litovchick, M. A. Clark, A. D. Keefe, Artificial DNA,
PNA & XNA 2014, 5, e27896; d) M. H. Hansen, P. Blakskjaer,
L. K. Petersen, T. H. Hansen, J. W. Hojfeldt, K. V. Gothelf, N. J.
V. Hansen, J. Am. Chem. Soc. 2009, 131, 1322-1327; e) D. L.
Usanov, Chan, A. I., Maianti, J. P. & David R. Liu, Nat. Chem.
2018, 10, 704-714.
[12] a) L. H. Yuen, S. Dana, Y. Liu, S. I. Bloom, A. G. Thorsell, D.
Neri, A. J. Donato, D. Kireev, H. Schuler, R. M. Franzini, J. Am.
Chem. Soc. 2019, 141, 5169-5181; b) S. Barluenga, C.
Zambaldo, H. A. Ioannidou, M. Ciobanu, P. Morieux, J. P.
Daguer, N. Winssinger, Bioorg. Med. Chem. Lett. 2016, 26,
1080-1085; c) L. Mannocci, S. Melkko, F. Buller, I. Molnar, J. P.
Bianke, C. E. Dumelin, J. Scheuermann, D. Neri, Bioconjug.
Chem. 2010, 21, 1836-1841.
[27] Y. Jin, C. Yu, R. J. Denman, W. Zhang, Chem. Soc. Rev. 2013,
42, 6634-6654.
[28] R. M. Franzini, T. Ekblad, N. Zhong, M. Wichert, W. Decurtins,
A. Nauer, M. Zimmermann, F. Samain, J. Scheuermann, P. J.
Brown, J. Hall, S. Graslund, H. Schuler, D. Neri, Angew. Chem.
Int. Ed. 2015, 54, 3927-3931.
[13] S. Dou, J. Virostko, D. L. Greiner, A. C. Powers, G. Liu,
Nucleosides Nucleotides Nucleic Acids 2015, 34, 69-78.
[14] a) S. Melkko, J. Scheuermann, C. E. Dumelin, D. Neri, Nature
biotechnology 2004, 22, 568-574; b) S. Melkko, Y. Zhang, C. E.
Dumelin, J. Scheuermann, D. Neri, Angew. Chem. 2007, 119,
4755-4758; c) J. Scheuermann, C. E. Dumelin, S. Melkko, Y.
Zhang, L. Mannocci, M. Jaggi, J. Sobek, D. Neri, Bioconj. Chem.
2008, 19, 778-785.
[15] a) G. A. Aleku, S. P. France, H. Man, J. Mangas-Sanchez, S. L.
Montgomery, M. Sharma, F. Leipold, S. Hussain, G. Grogan, N.
J. Turner, Nat Chem 2017, 9, 961-969; b) Q. Zhao, I. Gottschalk,
J. Carlsson, L. E. Arvidsson, S. Oscarsson, A. Medin, B. Ersson,
J. C. Janson, Bioconjug. Chem. 1997, 8, 927-934.
[16] A. Jain, G. M. Whitesides, R. S. Alexander, D. W. Christianson,
J. Med. Chem. 1994, 37, 2100-2105.
[17] a) I. Huc, J. M. Lehn, Proc. Nat. Acad. Sci. USA 1997, 94, 2106-
2110; b) G. Nasr, E. Petit, C. T. Supuran, J. Y. Winum, M.
Barboiu, Bioorg. Med. Chem. Lett. 2009, 19, 6014-6017.
[18] a) M. A. Clark, R. A. Acharya, C. C. Arico-Muendel, S. L.
Belyanskaya, D. R. Benjamin, N. R. Carlson, P. A. Centrella, C.
H. Chiu, S. P. Creaser, J. W. Cuozzo, C. P. Davie, Y. Ding, G.
J. Franklin, K. D. Franzen, M. L. Gefter, S. P. Hale, N. J. V.
Hansen, D. I. Israel, J. W. Jiang, M. J. Kavarana, M. S. Kelley,
C. S. Kollmann, F. Li, K. Lind, S. Mataruse, P. F. Medeiros, J.
A. Messer, P. Myers, H. O'Keefe, M. C. Oliff, C. E. Rise, A. L.
Satz, S. R. Skinner, J. L. Svendsen, L. J. Tang, K. van Vloten,
R. W. Wagner, G. Yao, B. G. Zhao, B. A. Morgan, Nat. Chem.
Biol. 2009, 5, 647-654; b) F. Buller, Y. Zhang, J. Scheuermann,
J. Schafer, P. Buhlmann, D. Neri, Chem. Biol. 2009, 16, 1075-
1086.
[19] a) T. Fujisawa, P. Filippakopoulos, Nat. Rev. Mol. Cell Biol.
2017, 18, 246-262; b) A. G. Cochran, A. R. Conery, R. J. Sims,
3rd, Nat. Rev. Drug Discov. 2019, 18, 609-628.
[20] Z. Liu, P. Wang, H. Chen, E. A. Wold, B. Tian, A. R. Brasier, J.
Zhou, J. Med. Chem. 2017, 60, 4533-4558.
[21] P. P. Sharp, J. M. Garnier, D. C. S. Huangab, C. J. Burns,
MedChemComm 2014, 5, 1834–1842.
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202005070
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
An amine-bearing DNA-encoded chemical library (DEL) could be converted to DNA-encoded dynamic library (DEDL) and subjected to
the selection against biological targets for inhibitor discovery. Full ligand structures could be obtained without the need for fragment-
linking.
7
This article is protected by copyright. All rights reserved.