Novel irreversible covalent BTK inhibitors discovered using DNA-encoded chemistry

Article abstract of DOI:10.1016/j.bmc.2021.116223

Libraries of DNA-Encoded small molecules created using combinatorial chemistry and synthetic oligonucleotides are being applied to drug discovery projects across the pharmaceutical industry. The majority of reported projects describe the discovery of reve

Full text of DOI:10.1016/j.bmc.2021.116223

Journal Pre-proofs
Novel irreversible covalent BTK inhibitors discovered using DNA-encoded
chemistry
John P Guilinger, Archna Archna, Martin Augustin, Andreas Bergmann,
Paolo A Centrella, Matthew A Clark, John W Cuozzo, Maike Däther, Marie-
Aude Guié, Sevan Habeshian, Reiner Kiefersauer, Stephan Krapp, Alfred
Lammens, Lukas Lercher, Julie Liu, Yanbin Liu, Klaus Maskos, Michael
Mrosek, Klaus Pflügler, Markus Siegert, Heather A Thomson, Xia Tian, Ying
Zhang, Debora L Konz Makino, Anthony D Keefe
PII:
S0968-0896(21)00231-5
https://doi.org/10.1016/j.bmc.2021.116223
BMC 116223
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
28 February 2021
11 May 2021
13 May 2021
Please cite this article as: J.P. Guilinger, A. Archna, M. Augustin, A. Bergmann, P.A. Centrella, M.A. Clark, J.W.
Cuozzo, M. Däther, M-A. Guié, S. Habeshian, R. Kiefersauer, S. Krapp, A. Lammens, L. Lercher, J. Liu, Y. Liu,
K. Maskos, M. Mrosek, K. Pflügler, M. Siegert, H.A. Thomson, X. Tian, Y. Zhang, D.L. Konz Makino, A.D.
Keefe, Novel irreversible covalent BTK inhibitors discovered using DNA-encoded chemistry, Bioorganic &
Medicinal Chemistry (2021), doi: https://doi.org/10.1016/j.bmc.2021.116223
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2021 Published by Elsevier Ltd.

Novel irreversible covalent BTK inhibitors discovered Leave this area blank for abstract info.
using DNA-encoded chemistry
JP Guilinger, A Archna, M Augustin, A Bergmann, PA Centrella, MA Clark, JW Cuozzo, M Däther, MA Guié,
S Habeshian, R Kiefersauer, S Krapp, A Lammens, L Lercher, J Liu, Y Liu, K Maskos, M Mrosek, K Pflügler,
M Siegert, HA Thomson, X Tian, Y Zhang, DL Konz Makino and AD Keefe
X-Chem Inc., 100 Beaver Street, Waltham, Massachusetts, 02453, USA
H
N
N
O
O
N
O
300 x 300 x 300 = 27M
O
H
N
H
N

Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com
Novel irreversible covalent BTK inhibitors discovered using DNA-encoded chemistry
John P Guilinger*1, Archna Archna2, Martin Augustin2, Andreas Bergmann2, Paolo A Centrella1, Matthew
A Clark1, John W Cuozzo3, Maike Däther2, Marie-Aude Guié1, Sevan Habeshian4, Reiner Kiefersauer2,
Stephan Krapp2, Alfred Lammens2, Lukas Lercher2, Julie Liu5, Yanbin Liu6, Klaus Maskos2, Michael
Mrosek7, Klaus Pflügler2, Markus Siegert2, Heather A Thomson8, Xia Tian9,Ying Zhang1, Debora L Konz
Makino*2 and Anthony D Keefe*1
1 X-Chem Inc., 100 Beaver Street, Waltham, Massachusetts 02453, USA
2 Proteros biostructures GmbH, Bunsenstr. 7a, 82152 Planegg-Martinsried, Germany
3 Formerly X-Chem Inc., now ZebiAI Inc., 100 Beaver Street, Waltham, Massachusetts 02453, USA
4 Formerly X-Chem Inc., now Laboratory of Therapeutic Proteins and Peptides École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
5 Formerly X-Chem Inc., now Accent Therapeutics Inc., 65 Hayden Avenue, Lexington, Massachusetts 02421, USA
6 Formerly X-Chem Inc., now Cyteir Therapeutics, 128 Spring St, Lexington, MA 02421, USA
7 Formerly Proteros biostructures GmbH, now Bruker AXS GmbH, Karlsruhe, Germany
8 Formerly X-Chem Inc., now deceased.
9 Formerly X-Chem Inc., now Arrakis Pharmaceuticals, 830 Winter Street, Waltham, Massachusetts 02451, USA
ARTICLE INFO
ABSTRACT
Article history:
Received
Received in revised form
Accepted
Libraries of DNA-Encoded small molecules created using combinatorial chemistry and synthetic
oligonucleotides are being applied to drug discovery projects across the pharmaceutical industry.
The majority of reported projects describe the discovery of reversible, i.e. non-covalent, target
modulators.
We synthesized multiple DNA-encoded chemical libraries terminated in
Available online
electrophiles and then used them to discover covalent irreversible inhibitors and report the
successful discovery of acrylamide- and epoxide-terminated Bruton’s Tyrosine Kinase (BTK)
inhibitors. We also demonstrate their selectivity, potency and covalent cysteine engagement using
a range of techniques including X-ray crystallography, thermal transition shift assay, reporter
displacement assay and intact protein complex mass spectrometry. The epoxide BTK inhibitors
described here are the first ever reported to utilize this electrophile for this target.
Keywords
Bruton’s Tyrosine Kinase
DNA-Encoded Chemical Libraries (DECL)
Covalent Irreversible Inhibitor
Drug Discovery
Tryptoline
1. Introduction
enrichment experiments (Zhu et al 2019). There has recently been
a resurgence of interest in the discovery of covalent drugs and for
Libraries of DNA-Encoded small molecules created using
combinatorial chemistry processes and synthetic oligonucleotides
are being applied to drug discovery projects across the
pharmaceutical industry. Recent reviews of this field include
Favalli et al 2018 and Goodnow et al 2017. The overwhelming
majority of reported projects describe the discovery of reversible,
i.e. non-covalent, target modulators with only a very small number
of reports relating to the discovery of covalent irreversible
modulators. This latter category has thus far been limited to
reported experiments with numerically small libraries - Zambaldo
et al 2016 (10,000 encoded compounds), Chan et al 2017 (136
encoded compounds simultaneously selected against 236 encoded
proteins) and Zimmermann et al 2017 (148,135 proximally
displayed building block pairs). One other report is further limited
to only reporting successful DNA-encoded positive control
reviews of recent developments in the covalent inhibitor discovery
field in general see for example Lagoutte et al 2017, Ghosh et al
2019 and Zhang et al 2019. We were motivated to determine if
DNA-encoded chemistry could be used to discover covalent
irreversible compounds from numerically large DNA-encoded
chemical libraries, i.e. within the numerical range that are
routinely reported for DNA-encoded chemical libraries used for
the purposes of non-covalent drug discovery. To support the
achievement of this goal we synthesized DNA-encoded chemical
libraries terminated in electrophiles with numerical sizes in the
tens to greater than one hundred million compounds and then
developed methods to successfully use them in affinity-mediated
selection processes adapted to discover irreversible target
engagers. We chose the target Bruton’s Tyrosine Kinase (BTK)
for this project as there is clear precedent for the discovery and

development of covalent irreversible inhibitors with therapeutic
utility for this target, e.g. ibrutinib (Burger et al 2015) for which
the mode-of-action is known to be via the formation of a covalent
bond with the active site cysteine. BTK is primarily expressed in
B-cells and is activated as a result of B-cell receptor signaling
pathways that in turn cause the recruitment of BTK to the cell
membrane and thereby result in the phosphorylation of tyrosine
551 (Spaargaren et al 2015). Ibrutinib is a potent and selective
covalent BTK inhibitor that binds to cysteine 481 in the active site
of BTK and is an effective monotherapy for CLL (Burger et al
2015). BTK is also a target for inflammatory diseases, for example
rheumatoid arthritis (Di Paolo et al 2011). We first developed
experimental protocols that were able to enrich on-DNA ibrutinib
(irreversible covalent inhibitor) but not on-DNA dasatinib
(reversible inhibitor) using an affinity-mediated selection process
with BTK. We then applied these optimized protocols to DNA-
encoded chemical libraries terminated with acrylamide and then
subsequently to a range of encoded electrophiles including other
α, β-unsaturated carboxamides, epoxides and haloalkanes. Here
we report the successful discovery of novel acrylamide-terminated
BTK inhibitors and the demonstration of their selectivity, potency
and cysteine engagement using a range of techniques including x-
ray crystallography. Subsequent library builds and additional
affinity-mediated selection experiments resulted in the discovery
of more potent epoxide-terminated BTK inhibitors that are the first
epoxide-mediated covalent irreversible inhibitors of BTK to be
reported.
thermal transition shift of +13.2°C observed for wild-type BTK
and of +13.4°C for the C481S mutant (Figure 2 and Table 3). The
similarity of the biophysical parameter values obtained using these
two orthogonal methods on wild-type and C481S BTK strongly
suggested that the presence of of a reactive thiol group in the
compound binding pocket does not affect the binding of
Compound 1 to BTK. Intact protein Mass-Spectrometry analyses
of Compound 1 pre-incubated with either wild-type or C481S
BTK clearly indicate the non-covalent nature of this interaction
(Figures 3 and 4, Supplementary Tables S2 and S3). Compound 1
engages BTK with a non-covalent interaction, and this was further
confirmed by X-ray crystallography of Compound 1 in complex
with BTK (Cuozzo et al 2017). Selectivity was assessed by
screening against a panel of 135 different kinases at DiscoverX
using the KINOMEscanTM profiling assay and this indicated that
this compound inhibited multiple other kinases after a one-hour
incubation at 10 µM (Figure 5). This was perhaps unsurprising
when one considers that this concentration is several hundred-fold
above the observed IC50 for BTK. Nonetheless, this did provide a
comparator for subsequent selectivity assessments with other
compounds reported in this paper. Notably, Compound 1 includes
a methyl amide stub that in the context of the encoded library
formed the first few atoms of the linker between the encoded
building block combination and the DNA tag. This linker stub was
subsequently shown to be required for activity and the crystal
structure shows this carbonyl hydrogen-bound to the backbone
carbonyl nitrogen of Phe413 in the glycine-rich loop of BTK
(Figures 6 and 7). This interaction appears to enable the tryptoline
to sit in a deep hydrophobic pocket while the quinoxaline
simultaneously interacts with the hinge region. The selection
output data revealed that while a range of structurally related
hinge-binding elements were tolerated (Table 4) the central furan
was the only building block co-enriched in this position with the
corresponding thiophene being the one exception with weak but
statistically significant co-enrichment also. The tryptoline was
uniquely co-enriched, but this building block is a natural product
with no closely related other building blocks included in this
library (ECFP6 Tanimoto distance >0.8), including its enantiomer.
Less potent non-tryptoline-containing active compounds were also
discovered from this library in this selection experiment (Cuozzo
et al 2017).
2. Results
2.1. Initial discovery of potent non-covalent BTK inhibitors
Our covalent discovery efforts were predated by the discovery
of potent non-covalent inhibitors of BTK and we have already
reported the discovery of 1, a cell-active non-covalent inhibitor of
BTK (Cuozzo et al 2017). This compound was discovered within
library 1 (Table 1). Library 1 was synthesized by the acylation of
a primary amine-terminated PEG linker displayed upon a
headpiece oligonucleotide with 300 Fmoc amino acids followed
by acylation with 157 formyl acids (carboxylates/aldehydes)
followed by reductive amination with 2,341 amines (Table 1).
Each chemical reaction was conducted in
a segregated
compartment and with concomitant ligation of an encoding
oligonucleotide prior to pooling and splitting for each successive
chemical step. Prior to library synthesis, each building block
admitted into the library had been individually determined to have
a yield of at least 50% in an on-DNA model reaction. Reaction
yields were calculated by examination of the UV and TIC traces
of the LCMS chromatograms. Double-substitutions and hydrolysis
products were identified and excluded. Additionally any products
in which the DNA was damaged by gain or loss of mass, such as
depurination, phosphate hydrolysis, or covalent modification,
were also excluded (Clark et al 2009). Each of the 110,261,100
unique compounds within this library is covalently attached to a
uniquely identifying oligonucleotide tag combination that encodes
its chemical history and can therefore be used to deduce its
identity. Compound 1 (Table 2) when tested in a TR-FRET assay
exhibited an IC50 value of 0.55 nM and it was also determined to
be active in cell culture and whole human blood assays (Cuozzo et
al 2017). During this study Compound 1 was determined to have
an IC50 value of 3.8 nM using a Reporter Displacement Assay
(Neumann et al 2011) with wild-type BTK, and a value of 3.0 nM
for C481S BTK (Figure 1 and Table 3). Residence times were
determined to be 9 and 22 minutes respectively (Table 3). Target
engagement was confirmed by the observation of a large positive
2.2. Subsequent discovery of structurally related non-covalent
BTK inhibitors in a distinct DNA-encoded chemical library
Emboldened by our initial successes, BTK was chosen to be
part of our further technology development efforts. The next
DNA-encoded chemical library utilized for this project was a
capped dipeptide library for which the first two synthesis steps
were the same as the first synthesis step described for library 1.
These were followed by an acylation step with 2,500 carboxylates
to generate library 2 with a total of 225,000,000 unique
compounds (Table 1). This library was also subjected to affinity-
mediated selection with BTK, amplification and sequencing as
described previously (Cuozzo et al 2017). Similarly to what had
been observed with Library 1, a range of enriched features were
observed in the processed selection output data. One cluster of
enriched structurally related compounds was clearly closely
related to compound 1 in that it was defined by the same tryptoline
building block at the first chemical cycle. Co-enriched structurally
related building blocks at the second chemical cycle included L-
phenylalanine and substituted L-phenylalanines and other closely
related amino acids (Table 4). The third and final cycle was
defined by 5-carbamoyl-1H-pyrrole-3-carboxylic acid, a relatively
unusual building block without other closely structurally related

building blocks in the library (ECFP6 Tanimoto distance > 0.75).
The one exception to this statement being 5-(pyrrolidine-1-
carbonyl)-1H-pyrrole-3-carboxylic acid which was not co-
enriched and contains no terminal amide hydrogen atoms and
would not therefore be able to provide these for hydrogen-bonding
interactions to the hinge region. The synthesis of the most highly
co-enriched building block combination yielded Compound 2
(Table 2) which was subsequently shown in the Reporter
Displacement Assay to have an IC50 value of 23 nM and 56 nM on
wild-type and C481S BTK, respectively (Table 3). Similarly to
Compound 1, the removal of the methyl amide linker stub resulted
in an inactive compound which is consistent with the hypothesis
of a similar binding mode for both compounds.
nature of the covalent bond between the target and the library
member to permit increased wash stringency by altering the wash
conditions to make them denaturing. We tried a range of
conditions while ensuring that the association between the target
protein and the affinity matrix was maintained. Most protein
affinity tags interact with immobilized capture reagents that are
themselves proteins and as such these interactions are prone to
becoming destabilized under denaturing conditions. Affinity tags
that are reliant upon the maintenance of protein conformation
include biotin/streptavidin, Strep-Tag or SBP-Tag/streptavidin,
epitope/antibody e.g. FLAG-, myc- or C-tags, Protein A or
G/antibody, glutathione/GST, amylose/MBP etc. One affinity tag
that is not susceptible to denaturation-mediated disassociation is
the polyhistidine tag captured via an immobilized transition metal
ion such as Ni2+. We chose to use hexahistidine-tagged BTK,
immobilized Ni2+ matrices and on-DNA ibrutinib and dasatinib to
discover conditions that would maximize enrichment of ibrutinib
(irreversible) over dasatinib (reversible). The chemical structures
of the on-DNA dasatinib and ibrutinib are shown in
Supplementary Figure S1. Any protocol used for identifying
irreversible covalent binders from a DNA-encoded library also
needs to be able to support the recovery of sequence information
from the encoding DNA as it emerges from the affinity-mediated
selection, i.e. covalently bound to the target protein. Sufficiently
denaturing washing conditions would of course also destabilize
non-covalent target-library member interactions and should
therefore provide a selection output that does not contain these,
hence the additional utilization of on-DNA dasatinib. As a
consequence of this methodological approach, separate selection
experiments with distinct protocols would be needed to separately
discover reversible and irreversible inhibitors and reversible and
irreversible libraries therefore should not be combined into a single
library super-mix.
2.3. Adaptation of DNA-encoded chemical library synthesis to
support the discovery of irreversible covalent inhibitors
Next we undertook our first attempt at the synthesis of a library
that could be used to discover covalent irreversible inhibitors. For
this library we used a tripeptide scaffold with a terminal primary
or secondary amine that was available for further subsequent
functionalization. This library was synthesized in a similar fashion
to Library 2, as described above, except that the first cycle of
chemistry comprised acylation with 305 amino acids and the
second and third with 293 amino acids for a total numerical
diversity of 26,183,954 tripeptides. In order to transform this
library into one capable of forming irreversible covalent bonds
with nucleophiles such as thiols we chose to install acrylamide by
acylation at the terminal amine with acrylic acid N-
hydroxysuccinimide ester. Because acrylamide contains an α,β-
unsaturated carboxamide group it can function as an electrophile
i.e. as a Michael acceptor. This library is listed as Library 3 in
Table 1. Because only a single electrophile was installed there was
no need to encode its identity using a building block identifying
tag. The acrylamide electrophile is found in several designed
covalent drugs including afatinib (Dungo et al 2013), ibrutinib
(Burger et al 2015), rociletinib (Sequist et al 2015) and others
(Jackson et al 2017) and in particular, acrylamide reacts with thiols
such as cysteines (Jackson et al 2017).
In order to validate and optimize selection conditions to identify
irreversible covalent inhibitors, the control compounds ibrutinib
and dasatinib were conjugated to specific sequences of DNA with
DNA tag architecture similar to that used for the DNA-encoded
chemical libraries. Ibrutinib (Burger et al 2015) was conjugated
to DNA as a positive control to detect irreversible covalent binding
to BTK. Dasatinib (Lombardo et al 2004) was conjugated to DNA
as a positive control to detect reversible non-covalent binding to
BTK. A no-compound on-DNA negative control was also
generated to detect non-specific interaction of the oligonucleotide
tag architecture with BTK. We then used these on-DNA
compounds to optimize affinity-mediated selection protocols for
the discovery of covalent irreversible binders while discriminating
against reversible binders. 50 nM of each on-DNA control was
incubated as a mixture with no or 1 uM BTK in protein storage
buffer supplemented with 10 mM MgCl2 and no or 1 mg/mL
sheared salmon sperm DNA as a general blocking agent for 1 hour.
This mixture was then diluted 2-fold into a native buffer, a urea-
based denaturing buffer or a guanidine-based denaturing buffer.
Subsequently the BTK was captured on beaded agarose or
magnetic bead IMAC matrices, washed in the same buffer used for
dilution and then eluted using 400 mM imidazole in the same
buffer used for dilution and washing. The eluate was then diluted
and, along with a dilution series of the input of each on-DNA
control, was subjected to qPCR analysis to determine the fraction
of the input of each on-DNA control recovered in the elution for
each of the different selection conditions. The first model
selection (Figure 8a) used a guanidine-based denaturing buffer and
a urea-based denaturing buffer and no sheared salmon sperm DNA
as a general blocking agent during incubation of BTK with the on-
DNA control mixture. This experiment was used to determine the
optimal composition of the denaturing wash buffer. As expected,
2.4. Design of an affinity-mediated selection process for the
discovery of irreversible target engagers within numerically
large DNA-encoded chemical libraries
Most reported methodologies for the enrichment of rare target-
engaging compounds existing within numerically large
combinatorial libraries of building block combinations utilize
multiple cycles of affinity-mediated selection to amplify the
enrichment of binders with respect to non-binders and thereby
render them observable and discoverable. We knew from our prior
experience with the affinity-mediated selection of non-covalent
libraries that a single cycle of selection was not sufficient to
support discovery when working with library super-mixes
containing upwards of tens of billions of compounds. If the
washing efficiency of a single round of selection is 99.9% then the
maximum enrichment possible is one thousand-fold. In reality this
extent is significantly reduced by the dissociation of genuine
binders during the wash phase, the likelihood that only a fraction
of library members display fully elaborated compounds and the
reality that only a fraction of the active molecules that are capable
of engaging the target will do so. Covalent irreversible selection
methods are reliant upon taking advantage of a single target-library
member encounter and any methods introduced for enrichment
amplification must operate within this very significant limitation.
The strategy that we took advantage of was to use the irreversible

on-DNA Ibrutinib was recovered with a substantially higher
efficiency compared to the no-compound on-DNA control when
BTK was present, and on-DNA ibrutinib was recovered with a
similar efficiency compared to no compound on-DNA when no
BTK was present indicating selective enrichment of the known
irreversible covalent binder Ibrutinib (Figures 8a and 8b).
Equivalent amounts of ibrutinib on-DNA were recovered in the
elution using model selections with guanidine-based denaturing
buffer vs. a urea-based denaturing buffer (Figure 8a). Similarly,
equivalent amounts of no-compound on-DNA were recovered in
the elution using model selections with guanidine-based
denaturing buffer vs. urea-based denaturing buffer. Interestingly,
dasatinib on-DNA amounts in the elution were substantially less
using model selections with guanidine-based denaturing buffer
when compared to the urea-based denaturing buffer, suggesting
that the guanidine-based denaturing buffer was better able to
denature BTK and/or otherwise wash away non-covalent binders
such as dasatinib. The second model selection experiment (Figure
8b) used a native buffer, a guanidine-based denaturing buffer and
also included sheared salmon sperm DNA and DNA-encoded
chemical libraries during the incubation of BTK with the on-DNA
control mixture to better mimic actual selection conditions.
Equivalent amounts of on-DNA ibrutinib were recovered in the
elution using model selections with magnetic bead IMAC vs.
agarose IMAC (Figure 8b). However, both on-DNA dasatinib and
no-compound on-DNA amounts were substantially less enriched
in the elution using model selections with magnetic bead IMAC
compared to the corresponding amounts with beaded agarose
IMAC suggesting that magnetic bead IMAC enables better
washing efficiencies of on-DNA dasatinib and no-compound on-
DNA in the presence of BTK (Figure 8b). Taken together, these
results indicate that on-DNA ibrutinib can be effectively enriched
in model selections with optimized conditions yielding fold
enrichments of >1000 for a single selection round when comparing
the amount of on-DNA ibrutinib to the amount of no-compound
on-DNA in model selections with BTK and magnetic bead IMAC
under the optimized conditions (Figure 8b). We acknowledge that
only approximately 3% of the input positive control molecules
were recovered in this experiment (Figure 8b). However, the
>1,000-fold relative enrichment observed for these when
compared to the negative control will result in active molecules
within library 4 being present at a 1-in-157 thousand frequency
after selection, compared to a 1-in-157 million frequency prior to
selection, assuming they behave similarly to this positive control.
When 100 pmoles of library 4 is used for a selection experiment
an average of 380,000 copies of each unique library member will
be present in the input, and if 50% are passaged through to PCR
we expect 5,000 copies to be present in the output. A 1-in-157
thousand frequency within the output translates to around six
counts per million reads, and since we aim for 1-2 million reads
per library per selection per time-point this signal strength is
observable using our standard workflow. Numerically larger
libraries will be usable with inputs of >100 pmoles, greater
sequencing depth, or both.
of the protein and associated library members from the matrix and
efficient amplification of the accompanying encoding
oligonucleotides. Common perceptions of covalent inhibitors
include the idea they are often not specific and so we chose to
design indicators of specificity directly into our affinity-mediated
selection campaign design. Firstly we chose a suite of seventeen
diverse and independent protein off-targets, all of which contained
cysteines and nine of which contained a cysteine in their active
sites. Additional information about these seventeen off-targets is
shown in Supplementary Table S1. These were subjected to
affinity-mediated selection in parallel and building block
combinations that enriched against BTK were only prioritized for
hit re-synthesis if they failed to demonstrate enrichment against
any of these seventeen off-targets. As a second independent
approach to collecting compound promiscuity data we determined
which building block combinations had ever been observed to
enrich above a statistical significance threshold for any of the
targets screened at X-Chem at any point prior to this screen.
Again, synthesis prioritization was directed to compounds defined
by building block combinations that had never before been seen to
enrich against any target screened at X-Chem. When this
experiment was conducted in 2015 the number of targets screened
to-date was already several hundred. Additional parallel selections
were designed to provide information on whether the enrichment
signal increased over time, as would be expected for an irreversible
interaction, and also whether the enrichment signal was reduced
by pre-incubating the target with a saturating concentration of an
orthosteric BTK inhibitor, as would be expected for a library
compound binding to the active site. A representation of the
number of compounds observed to enrich against BTK is shown
in Figure 9. This figure shows that an abundance of compounds
were observed to enrich in this affinity-mediated selection
experiment. Individual clusters are colored according to the extent
to which the building block combinations they represent have been
observed to have enriched against targets previously screened at
X-Chem with red representing the most promiscuous and green the
least promiscuous. Our attention therefore was focused upon the
clusters colored green. The most highly enriched cluster in this
plot, circled in black, is defined by the same, i.e. previously-
observed, tryptoline building block at the second cycle of
chemistry, piperidine-3-acetic acid at the third cycle of chemistry
displaying the acrylamide electrophile via acylation of its amine.
Further selection output analysis revealed other structurally related
building blocks at the third cycle of chemistry including other
constrained saturated cyclic amino acids (Table 4). Two of these
are represented by the green clusters circled in blue in Figure 9.
All of these clusters of related compounds showed tolerance of a
broad range of structurally unrelated building blocks at the first
cycle of chemistry, which lead us to the conclusion that this region
of the molecule may be dispensable. Synthesis of a combination
of the most highly enriched building blocks at the second and third
cycles of chemistry while truncating the first cycle of chemistry to
a methyl amide resulted in Compound 3 (Table 2). The large
number of parallel distinct selection conditions utilized in this
experiment enabled us to make a number of predictions with
regard to the selectivity characteristics of this compound prior to
testing it in any assays. Figure 10 shows a profile of the relative
enrichment of the building block combination defining Compound
3 across all of the parallel selections that were conducted. Aliquots
of the BTK incubation solution were frozen at different time-
points prior to their capture upon the affinity matrix and the profile
indicates an increasing enrichment signal with increasing
incubation time, as is consistent with the establishment of an
irreversible interaction with the target, i.e. Compound 3 is
predicted to exhibit a covalent mode of action. Parallel selections
conducted with 18-hour incubations with and without the presence
2.5. Subsequent discovery of structurally related irreversible
covalent BTK inhibitors in a DNA-encoded chemical library
terminated with acrylamide
Having successfully demonstrated the enrichment of on-DNA
ibrutinib using an optimized affinity-mediated selection protocol
we then used this protocol in experiments with Library 3 (Table 1)
and BTK. This protocol included incubation in native solution,
introduction of guanidine as a denaturant, capture on IMAC
magnetic beads, washing under denaturing conditions, liberation

of a saturating concentration of staurosporine, an orthosteric
inhibitor of BTK, show a starkly differential response with highly
statistically significant enrichment observed for the latter and no
enrichment observed for the former. This is consistent with
Compound 3 binding BTK in its active site, i.e. that selections with
each of the seventeen off-targets indicates that none of these
resulted in a statistically significant signal (i.e. none exhibited a
value of >10 for the negative log10 of the asymptotic significance
value). This observation and the lack of enrichment observed for
targets historically screened at X-Chem lead us to predict that
compound 3 would be a highly selective inhibitor of BTK.
Compound 3 was subsequently determined to have an apparent
IC50 value of 4.8 µM after 1 hour of incubation using the Reporter
Displacement Assay with wild-type BTK, and was inactive against
C481S BTK (Figure 1 and Table 3) as would be expected for a
compound that elicits its activity by covalent modification of
cysteine 481. Selectivity was then assessed by screening against a
panel of 135 different kinases at DiscoverX using the
KINOMEscanTM profiling assay and this indicated that Compound
3 inhibited few kinases after a one-hour incubation at 10µM
(Figure 5). Covalent target engagement was confirmed by the
high-resolution 1.44Å X-ray crystal structure of compound 3 in
complex with wild-type BTK (Figures 6 and 7). And by a positive
thermal transition shift of +6.5°C for wild-type BTK and of
<0.2°C for the C481S mutant (Figure 2 and Table 3), as would be
expected for a covalent modifier of cysteine 481. Target
engagement was also confirmed by incubation of Compound 3
with BTK followed by intact-protein mass spectroscopy indicating
that the protein had become an adduct with Compound 3 of the
expected mass (Figure 3 and Table 3) and that no such adduct was
observed after incubation with C481S BTK (Figure 4 and Table
3).
In addition to the tryptoline-containing series of compounds,
additional structurally distinct series were also observed to have
been enriched in the selection experiment. These include one
exemplified by Compound 4. Similar to observations made for the
set of compounds exemplified by Compound 3, the selection
output data for Compound 4 also indicated that this compound
exhibited time-dependent enrichment (data not shown), would be
selective and would bind to BTK in a manner that inhibits its
activity. This new cluster of compounds was defined by 3 and 4-
cyano L-phenylalanine at the second cycle of chemistry and
substituted L-proline derivatives at the third cycle of chemistry
(Table 4). Similar to the way Compound 3 was represented in the
selection output data, the first cycle of chemistry exhibited a high
tolerance for a wide range of structurally unrelated building blocks
and was ultimately determined not to be necessary for activity. We
do not have crystallography data for the complex of compound 4
with BTK, but it is likely that the cyano phenylalanine side-chain
occupies the same hydrophobic pocket as the tryptoline in
compounds 1 and 3, and that the acrylamide reacts with the same
active site cysteine. Compound 4 was determined to have an
apparent IC50 value of 36 µM after 1 hour of incubation using the
Reporter Displacement Assay with wild-type BTK, and was
observed to be inactive against C481S BTK (Table 3). Target
engagement was confirmed by incubation of Compound 4 with
BTK followed by intact-protein mass spectroscopy indicating that
the protein had become an adduct with Compound 4 of the
expected mass (Figure 3 and Table 3) and that no such adduct was
observed after incubation with C481S BTK (Figure 4 and Table
3). Selectivity was assessed by screening against a panel of 135
different kinases at DiscoverX using the KINOMEscanTM
profiling assay and this indicated that this compound is highly
selective against over other kinases under these assay conditions
(Figure 5).
The overall structure of the complex of wild-type BTK (kinase
domain) with Compound 3 (Figures 6 and 7) was almost identical
to that of Compound 1 (Cuozzo et al 2017) with a Cα RMSD ≈
0.1. Compound 3 occupies the same binding site as was described
for Compound 1. Similar observations have also been made for
other inhibitors, for example for B43 (PDB: 3GEN, (Marcotte et
al. 2010), in which the inhibitor resides in the ATP-binding pocket
of BTK (Caldwell et al. 2019). Comparing the overall structure
with previously reported structures of BTK (PDB 3PIZ and 3PJ2)
(Kuglstatter et al, 2011), helix-C adopts the ‘out’ conformation in
which it moves away from the active site and prevents the
formation of the salt bridge between Lys430 and Glu445 that is
important for activity (Supplementary Figures S2 and S3).
Interestingly, we also observed that the binding of both Compound
1 and Compound 3 displaces the side chain of Lys430 further away
from Glu445. The tryptoline moiety is common to both
Compounds 1 and 3, and it occupies the hydrophobic pocket
defined by Phe413, Ile432, Met437, Ile472 and Leu542 (Figures 6
and 7). The quinoxaline moiety of Compound 1 occupies the
adenine-binding pocket and interacts with the hinge region by H-
bonding between quinoxaline N-32 and the backbone amide
nitrogen of Met477. In contrast to this, in the BTK structure with
Compound 3 there are no hydrogen bonding interactions observed
between the compound and the hinge region. Compound 3 also
interacts with the glycine-rich loop and Asp539 in a DFG-in
conformation via hydrogen bonding with the main-chain amide
nitrogen of Gly411/Phe413 and the side-chain of Asp539,
respectively. Further structural analysis of the BTK-Compound 3
complex confirms a covalent bond between this compound and
cysteine 481 near to the ATP-binding pocket with continuous
electron density between cysteine 481 and the Compound 3
warhead (Figure 7 and Supplementary Figure S2).
2.6. Encoded Electrophile-Terminated Encoded Library
Synthesis
The irreversible covalent inhibitors that we had discovered thus
far all used acrylamide as a cysteine-reactive electrophile. This
class of warhead (α, β-unsaturated carboxamide) has precedent for
this target, not least with ibrutinib (Burger et al 2015). We next
set out to determine if we could construct a DNA-encoded library
terminated in a range of encoded electrophiles. There is increasing
interest in alternatives to α, β-unsaturated carboxamides as
electrophiles in covalent drugs (Gehringer et al 2018, Ray and
Murkin 2019) including epoxides, haloalkanes, sulfonyl fluorides,
activated alkynes and several other classes. The strategy we
adopted for the synthesis of library 4 (Table 1) utilized the same
starting point that we used for library 3, a 26,183,954-member
tripeptide library with a free terminal amine available for
acylation. We chose six electrophilic building blocks for
installation by acylation or carbamate formation at this terminal
amine including five epoxides and one haloalkane. In order to
encode the identity of each of these electrophiles we divided the
nascent library into six aliquots and to each aliquot ligated a
different fourth-chemistry-cycle tag prior to installing the same
library-identifying tag, degenerate region and primer-binding
sequence (Tailpiece) to each. Because we anticipated that specific
electrophiles would have different generic reactivity, we assessed
the relative reactivity of each electrophile by combining library
aliquots containing each of the electrophiles at an equimolar ratio
and then performed affinity-mediated selection using BTK and
one of the off-target proteins. For each electrophile, the fractional
counts in the processed and selection output were calculated and
used to inform the final mixing ratios with electrophiles included

in inverse proportion to their prevalence in the selection output.
Based on these data the final library was generated using an
equimolar ratio of each of the five epoxides with one tenth the
amount for the haloalkane (i.e. a molar ratio of 1 haloalkane to 51
total library). Subsequent to the generation of this library we were
able to show that highly similar selection output signatures could
be generated by repeating the same selection experiments after
storage of libraries at -80°C for multiple years. This demonstrated
sufficient stability of the electrophiles in this physical and
chemical context to allow this and similar other libraries to be used
on an ongoing basis.
3). No such adduct was observed for the C481S BTK for either
compound (Figure 4 and Table 3). Target engagement was further
confirmed by the observation of positive thermal shift values of
+4.3°C and +6.5°C with wild-type BTK for Compounds 6 and 7
respectively (Figure 2 and Table 3).
3.1 Experimental Methods
3.11 Optimization of Affinity-Mediated Selection Conditions
for Covalent Irreversible Binders
BTK (0 or 1 µM) (Signal Chem B10-10H-BULK) in 1x protein
storage buffer Sodium Phosphate, pH 7 (50 mM), NaCl (150 mM),
PMSF (0.1 mM), DTT (0.25 mM), glycerol (25%), supplemented
with MgCl2 (10 mM) and supplemented with or without sheared
salmon sperm DNA (sssDNA) (1mg/ml, Invitrogen) was
incubated with a mixture of 50 nM of ibrutinib on-DNA, 50 nM
dasatinib on-DNA and 50 nM no compound on-DNA control in a
total volume of 50 uL for 1 hour at room temperature. Then to
each mixture was added 50 uL of appropriate buffer which is either
1x non-denaturing buffer comprised of Sodium Phosphate pH 7.4
(50 mM), NaCl (200 mM), Imidazole (5 mM), sssDNA (1mg/ml,
Invitrogen), TCEP (1 mM), Tween 20 (0.02%), 1x urea-based
denaturing buffer comprised of 1x non-denaturing buffer with urea
(6 M), or 1x guanidinium -based denaturing buffer comprised of
1x non-denaturing buffer with Guanidinium chloride (6 M).
immobilized on agarose-based or magnetic resin IMAC, then
washed 5 times with 200 uL of appropriate buffer then incubation
of the aspirated beads with 50 uL 0.8x appropriate buffer with an
additional 400 mM imidazole, pH 7.5 For experiments with
immobilization on agarose resin, pre-washed PhyTips containing
5 uL of HisSelect resin (Phynexus ) were used as previously
described in Cuozzo et al 2017. For experiments with
immobilization on magnetic resin, 50 ul aliquot of Pierce HisPur™
Ni-NTA Magnetic Beads (ThermoFisher #88831) was aspirated
and washed in 10 mM Tris (pH8.0) followed by washing in
Denaturing Buffer followed by resuspension in 30 ul of
Denaturing Buffer was used. Elution, or a titration series of the
input on-DNA control mix in 50 uL,, then subjected to qPCR
analysis in 50 ul qPCR volume using Power SYBR Green Master
Mix (ThermoFisher 4368577) with 5 uL of dilution of elution or
standard curve of input as template and 0.5 uM primer pair specific
for each on-DNA control compound sequence.
2.7. Subsequent discovery of structurally-related irreversible
covalent epoxide-containing BTK inhibitors in a DNA-encoded
chemical library
We then used library 4 in an affinity-mediated selection
experiment conducted with BTK in the same manner as that used
for library 3, i.e. with a protocol optimized to enrich covalent
irreversible binders only. The analysis of the output from this
selection experiment again showed a broad range of enriched
compounds. These included clusters of structurally closely related
compounds. One such compound cluster contained compound 5
(Table 2) which, in common with compounds 1, 2 and 3, contained
the tryptoline building block. Compound 5 was co-enriched with
a set of closely structurally related compounds, all of which
contained the tryptoline building block at the second cycle of
chemistry. At the third cycle of chemistry these compounds
contained
a
range of halogenated β-homophenylalanine
derivatives acetylated with glycidic acid, a substituted epoxide
(Table 4). Again, the selection output data suggested that the first
cycle of chemistry was dispensable. The synthesis of the most
highly enriched building block combinations at cycles two, three
and four with the first cycle being truncated to a methyl amide
provided Compound 5 (Table 2). Compound 5 exhibited an
apparent IC50 value of 0.10 µM (1 hour incubation) when tested in
the Reporter Displacement Assay against wild-type BTK (Table
3) and was inactive against C481S BTK (Table 3). The
KINOMEscanTM profiling assay also indicated that Compound 5
is highly selective (Figure 5).
A separate set of co-enriched building blocks defined
compounds with clear structural relatedness to Compound 4. For
these compounds, the first cycle of chemistry again appeared to be
dispensable; the second cycle of chemistry included the 4-cyano
L-α-phenylalanine building block seen in compound 4, but also L-
β-phenylalanine derivatives (Table 4). In the third cycle of
chemistry α-amino acids with hydrophobic side-chains were co-
enriched, which in turn were acetylated by glycidic acid or its
methyl analog, with the latter being more highly enriched (Table
4). The synthesis of the most highly enriched building block
combinations for each of the alternatives at chemistry cycle three
with the first cycle being truncated to a methyl amide provided
compounds 6 and 7.
3.12 Affinity-Mediated Selection for Irreversible Target
Binders
A
mixture of independently synthesized electrophile-
terminated DNA-Encoded Chemical Libraries and 1 uM of BTK
(Signal Chem B10-10H-BULK) or 1 uM off-target protein were
combined in native solution and incubated with aliquots
withdrawn and frozen at 2 and 18 hours. The Incubation Buffer
was designed to model the cytoplasmic environment and was
comprised of HEPES (20 mM), potassium acetate (134 mM),
sodium acetate (8 mM), sodium chloride (4 mM), magnesium
acetate (0.8 mM), sheared salmon sperm DNA (1mg/ml,
Invitrogen), TCEP 0.5 mM and Tween 20 (0.02%) at pH 7.2 with
17 uM total library concentration and 1 µM BTK. Each 30 µl
aliquot, once withdrawn, was immediately combined with 30 µl of
a Denaturing Buffer comprised of guanidinium hydrochloride (6
M) di/mono sodium phosphate (pH 7.4, 50 mM), sodium chloride
(200 mM), imidazole (2 mM), sheared salmon sperm DNA
(1mg/ml, Invitrogen), TCEP 0.5 mM and Tween 20 (0.02%) and
then frozen for storage. Once all samples had been generated and
frozen each was heated to 95°C for two minutes and then cooled
to room temperature and then incubated with pre-washed Ni-NTA
magnetic beads. Each 50 ul aliquot of Pierce HisPur™ Ni-NTA
Compounds 6 and 7 were determined to have apparent IC50 values
of 26 and 40 nM respectively using the reporter displacement TR-
FRET assay and one-hour incubations (Figure 1 and Table 3).
Kinact and Ki values were also determined for both compounds with
values of 0.0082 and 0.0094 per second and 1.3 and 1.9 µM
respectively (Table 3).
Both were inactive in the Reporter
Displacement Assay against C481S BTK (Figure 1 and Table 3).
Target engagement was confirmed for both compounds by
incubation with BTK followed by intact-protein mass
spectroscopy indicating that the protein had become an adduct
with each compound with the expected mass (Figure 3 and Table

Magnetic Beads (ThermoFisher #88831) was aspirated and
washed in 10 mM Tris (pH8.0) followed by washing in Denaturing
Buffer followed by resuspension in 30 ul of Denaturing
Buffer. Incubation samples were then captured upon the magnetic
beads by incubation for 30 minutes with shaking at room
temperature, non-captured mixture components were then
removed with five 200 ul washes with Denaturing Buffer. Elution
of captured components was then achieved by incubation of the
aspirated beads with the Elution Buffer (0.8x Denaturing Buffer
with an additional 400 mM imidazole, pH 7.5). Additional
magnetic aspirations were performed to ensure removal of all
magnetic beads from the elution sample. Denaturants were
removed from the elution sample by desalting column purification
with Zeba Spin Desalting Columns, 40 kDa MWCO, 0.5mL
(ThermoFisher #87766) according to the manufacturer’s
instructions. Additional selections were performed in parallel
with no target, off-targets, BTK plus 0.1 mM
staurosporine. Encoding oligonucleotides present in the processed
eluates for each selection condition were then amplified using
Platinum PCR SupermixTM (Invitrogen) with denaturation at 94°C,
annealing at 55°C and extension at 72°C for 24 cycles using 5’-
and 3’- primer oligonucleotides (each at 0.5 µM) that each
incorporate sequences complementary to the tailpiece or
headpiece along with Illumina READ1 or READ2 sequences
required to support clustering and subsequent single-read 100
base-pair sequencing on an Illumina HiSeq 2500. Sequencing was
also performed for PCR-amplified samples of the naïve
(unselected) library. Sequencing depth for irreversible selection
outputs is targeted at 1-2 million single-end reads for each
library/selection-condition/time-point combination. Sequence
data were converted back into encoded chemical information
computationally and demographic and statistical information were
calculated for all individual building block combinations.
resolution of 60,000 FWHM (full width at half maximum). The
end-plate offset was set to 500 V, the capillary voltage to 4500 V,
the nebulizer gas to 2.0 bar, the dry gas to 10.0 l/min, and the dry
temperature to 220°C. Furthermore, a Funnel 1 RF of 400 Vpp, a
multipole RF of 400 Vpp, a quadrupole ion energy of 7.0 eV, a
low mass of 322 m/z, a collision energy of 10.0 eV, a collision RF
of 1200.0 Vpp, a transfer time of 110.0 µs, and a pre pulse storage
of 10.0 µs was applied. The first two minutes and the last seven
minutes of each LC run were diverted to the waste. This time was
used for external mass calibration using the ESI-L low
concentration tuning mix (Agilent, Waldbronn, Germany) infused
by a syringe pump (New Era Pump Systems, Inc., Farmingdale,
NY, USA). In addition, internal mass calibration was performed
using the compound at m/z 922.009789 originating from the ESI-
L low concentration tuning mix. The whole LC-MS system was
controlled by Bruker Compass Hystar 5.1 with DCMS link plugin
and otofControl 5.2 (both Bruker). For data evaluation and
processing, DataAnalysis 5.1 and BioPharma Compass 2021 was
used (both Bruker). Samples for LC-MS analysis were diluted to
a protein concentration of 0.1 mg/ml in 5% acetonitrile (v/v) and
0.1% FA (v/v) prior to the injection of 5 µl of this solution. Prior
to analysis compounds were incubated with BTK for one hour.
3.15 Reporter Displacement Assay
Binding kinetics determinations were performed with the
Proteros reporter displacement assay as previously described
[Neumann et al. 2011, Schneider et al. 2013]. In the assay, the
time- and competitor concentration-dependent displacement of a
Cy5 labelled ATP-competitive kinase inhibitor probe is measured.
Probe and enzyme are pre-incubated, before addition of unlabeled
competitor. After brief mixing, data are recorded for 1 hour. Data
were normalized to high (enzyme with probe, no competitor) and
low (enzyme without probe, with competitor) controls for every
time-point. For covalent irreversible compounds, the fluorescence
decay over time was fit to a mono-exponential decay, excluding
the earliest time-point. Obtained kobs values were replotted against
concentration and fit using kobs = kinact*[I] / (KI + [I]).
3.13 Synthesis of compounds 1-7
Compound 1 was synthesized according to (Cuozzo et al 2017),
additional chemical synthesis information can be found in the
Supplementary Information.
3.16 Thermal Shift Assay
3.14 Intact Protein Complex Liquid Chromatography - Mass
Spectrometry
To determine the protein melting temperatures, 5 µM BTK
(389-659) wikd-type or C481S mutant was incubated with 50 µM
of the respective compound for 120 min in 120 µL buffer (20 mM
HEPES pH 8.0, 0.5 mM TCEP, 150 mM NaCl and
0.01%Tween20, 0.5% DMSO) at room temperature.
Subsequently, samples were loaded into nanoDSF capillaries in
technical triplicates (PR-C006 NanoTemper Technologies) and
loaded into the NanoTemper Prometheus NT.48. Fluorescence at
330 and 350 nm was monitored during a temperature ramp from
20.0 – 95.0°C at 2.5°C/min. The 350 nm / 330 nm ratio was
analyzed and melting points were determined using the
PR.ThermControl (NanoTemper Technologies, v.2.1.2) software.
For liquid chromatography mass spectrometry (LC-MS)
analysis of ligands, proteins, and protein-ligand complexes, a
Vanquish Flex LC-system (Thermo Fisher Scientific, Bremen,
Germany) was used in combination with a maXis plus MS-system
(Bruker, Bremen, Germany). The LC-system was equipped with
a binary pump, a 25 µl sample loop, a column oven and a variable
wavelength detector (VWD, all Thermo Fisher Scientific). For
analyte separation, a BIOshell™ A400 Protein C4 column
(100 mm x 2.1 mm, 3.4 µm, 400 Å, Merck, Darmstadt. Germany)
as stationary protected by a BIOshell™ A400 Protein C4 guard
cartridge (5 mm x 2.1 mm, 3.4 µm, 400 Å, Merck) and the
following gradient of solvent A (0.1% formic acid, FA, v/v) and
solvent B (acetonitrile with 0.1% FA v/v) at a flowrate of
300 µl/min was applied: t [min]/B [%]: 0/15; 2/15; 10/65; 10.5/98;
14.5/98; 15/15; 19/15. The solvent was actively preheated to
80 °C and the column oven was kept at 80 °C. Before entering the
VWD, the solvent was cooled to 40 °C using a post-column cooler
(Thermo Fisher Scientific). In the VWD, analytes were detected
using two wavelengths of 214 nm and 280 nm and a data
collection rate of 4 Hz. The MS-system was coupled on-line to the
LC via the ESI interface. The maXis plus operated in the positive
ion mode with activated focus mode for a mass spectrometric
3.17 Kinase Panel
Compounds were tested for selectivity against a panel of 135
different kinases at DiscoverX using the KINOMEscanTM
profiling assay. This is a reporter displacement assay with
detection of DNA-conjugate reporter molecules conducted by
qPCR. Incubations with test molecules were conducted at 10 uM
for one hour at room temperature in PBS.

3.18 Complex Crystallization and Structure-Determination
interactions might be expected to equilibrate and demonstrate
time-independent enrichment signal at least at the incubation times
used in our covalent selections (2-15 hours). Covalent inhibitors
generally exhibit increasing enrichment at longer time points, see
for example Figure 10. The success of our approach is exemplified
by the characteristics of the compounds that we discovered.
Modes of action were confirmed using a series of biophysical and
crystallographic methods including assessments of potency and
selectivity which also proved their ability to engage the target with
a covalent bond. Notably, none of the covalent irreversible
compounds discovered in this study required the first cycle of
chemistry to be present to exhibit activity and in all cases it was
evident from the large number of structurally unrelated building
blocks co-enriched in this register that this was the case. This is
in great contrast to our experience of working with the X-Chem
DNA-Encoded Chemistry platform in general. Reviewing all of
the four hundred and fifty-six non-covalent compounds that we
have licensed to our external partners to-date (between December
2012 and December 2020) revealed that greater than 94% of these
contained building blocks co-enriched at all of the cycles of
encoded chemistry. For licensed covalent irreversible compounds
the corresponding number is 92%. It is unclear to us why this
project was so unusual in that it provided two different chemical
series, both of which amenable to truncation when compared to
their original library format. This may be because of one or more
of the following reasons. The electrophile-terminated libraries
utilized for this project contain four cycles of chemistry whereas
those we apply to drug discovery projects generally contain only
two or three cycles of chemistry. The chemistries used in the first
and second cycles for the electrophile-terminated libraries 3 and 4
were identical meaning that individual molecules for which the
installation of the building block at the first cycle did not occur
would still be able to accept building blocks at all downstream
chemistry cycles. All building blocks were validated for each of
the chemistry steps within each library synthesis scheme with only
those achieving yields of >50% being admitted to the library.
Accordingly, a significant fraction of the resultant library is fully
elaborated molecules, but it is also inevitable that truncates will
also exist within the encoded mixture. It is also possible that this
is a target-specific observation. Since these proof-of-concept
experiments were conducted in 2015 (Clark et al 2016) we have
offered affinity-mediated irreversible selection as a means to
discover covalent inhibitors of therapeutic targets to our partners
and observed that 92% of all licensed compounds thus far were
comprised of all encoded cycles of chemistry.
N-terminally GST-tagged BTK (aa389-659) was expressed and
purified as previously described (Cuozzo et al 2017). For
crystallization purposes the GST tag was cleaved off and the
protein was subjected to crystallization using a grid screen. This
grid screen was designed by systematically varying the PEG
concentration and pH around previously published Polyethylene
glycol based conditions (PEG 5kDa MME, ammonium sulfate and
MES pH 6.0-7.0). Crystals of BTK were obtained in a few days.
In order to obtain the co-crystal structure with Compound 3, these
crystals were soaked into the final concentration of 10 mM
compound 3 (dissolved in 100% DMSO) in the crystallization
buffer. Soaked crystals were cryo-protected and flash-cooled in
liquid nitrogen. The X-ray crystal diffraction dataset was collected
at the SLS (Swiss Light Source, Villigen, Switzerland).
Afterwards, these diffraction data were indexed and integrated
using XDS (Kabsch et al 2010) and scaling was performed using
XSCALE (Kabsch et al 2010). The structure was solved by a
molecular replacement (MR) method using a previously published
BTK structure as the MR model (Marcotte et al 2010). The
structure was refined by iterative model building using COOT
(Emsley & Cowtan, 2004) and refinement using the CCP4
software package. Data collection and refinement statistics are
described in Table 5.
4.1 Discussion
We have demonstrated the ability to synthesize numerically large
electrophile-terminated
oligonucleotide-encoded
chemical
libraries and then to apply affinity-mediated selection protocols to
them to enrich irreversible covalent target-engagers to a sufficient
extent to permit their identification within the selection output.
For example, the most enriched building block combinations
containing compound 3 encoded at the second and third cycles of
chemistry with additional defined building blocks at the first cycle
of chemistry were observed with count of seven in the selection
output. Because there were 551,923 qualifying reads derived from
this selection these individual fully elaborated compounds each are
0.00127% of the selection output. Because the numerical size of
library 3 is 26,183,954 the DNA tag combination encoding
Compound 3 was 0.00000382% within the selection input and so
a
single cycle of affinity-mediated selection effected an
enrichment of (at least) 330-fold which with sufficient sequencing
depth enabled the identification of active compounds. The
corresponding number for compound 4 discovered within Library
3 was 420-fold (9 counts). For the discoveries of Compounds 5, 6
and 7 in Library 4 it might be expected that because of the larger
numerical size of this library (157,103,724 encoded compounds)
observed enrichment extents would be smaller. The observed
fold-enrichment values for these compounds were 53, 440 and
520-fold respectively. The amount of library used for these
experiments was 100 pmoles with 50% of the output used as input
for PCR amplification.
5.1 Conclusion
We have developed and now report a method to adapt DNA-
encoded chemistry to the discovery of covalent irreversible
inhibitors to targets of therapeutic interest. And we have
demonstrated the utility of this method to discover multiple novel
series of covalent irreversible inhibitors of Bruton’s Tyrosine
Kinase (BTK). The design of the selection campaign was adapted
to inform upon the mode-of-action and selectivity of the enriched
compounds and the compounds thus discovered were
characterized in sufficient detail to demonstrate the predictive
nature of these data. The compounds were fully characterized by
Reporter Displacement Assay, Thermal Stability Assay, and intact
protein mass-spectrometry of reversible and irreversible
compounds with wild-type and C481S (unreactive) BTK proteins,
and also using high-resolution X-ray crystal structures. These
different series employ multiple covalent warheads to support
bond formation including α, β-unsaturated carboxamides and
epoxides. As far as we are aware, these epoxides are the first ever-
The advances that made these discoveries possible were achieved
by undertaking a number of protocol optimization experiments in
which on-DNA ibrutinib (irreversible) and dasatinib (reversible)
were used as positive and negative controls respectively and
enrichment was quantified under a range of different conditions.
Most notably increased wash stringency using buffer with
denaturant increased the enrichments of the irreversible covalent
binder over reversible binders compared to “standard” wash
stringency using buffer with no denaturant (Figure 8). Ultimately,
during selections, incubation time was used as a variable to help
identify irreversible covalent binding events since non-covalent

reported inhibitors of BTK to utilize this electrophile. We have
been operating this platform on behalf of our partners since 2015
and have licensed multiple series of covalent irreversible
compounds to a range of targets, including those more challenging
than BTK. This is the only screening method currently in
existence that combines irreversible target engagement assessment
with the opportunity to simultaneously screen hundreds of millions
of compounds and so we believe that this platform has broad utility
for projects for which a covalent mode of action and novel
chemical equity are both considered desirable.
Figure 1

Figure 1 Reporter Displacement Assay data for Compounds 1, 3 and 7 against wild-type and C481S BTK indicating the activity of Compound 1 against both and
of Compounds 3 and 7 against wild-type BTK only. This observation is consistent with the hypothesis that the mode of action of Compounds 3 and 7 is through
irreversible covalent engagement of Cysteine 481, whereas for Compound 1 the interaction is non-covalent and reversible in nature and not dependent upon
engagement of Cysteine 481. Time-dependent normalized signals at different compound concentrations (in µM) are shown on the right (see legend). Experiments
were performed in technical triplicates.
Figure 2
Figure 2 Thermal Shift Assay data derived from the intrinsic tryptophan fluorescence of BTK measured using nanoDSF with a NanoTemper Prometheus NT.48
instrument. Large positive thermal transition shifts are observed for Compounds 1 (orange), 3 (green) and 7 (red) against wild-type BTK (apo: blue) whereas only
for Compound 1 for C481S BTK. This observation is consistent with the hypothesis that the mode of action of Compounds 3 and 7 is through irreversible covalent
engagement of Cysteine 481, whereas for Compound 1 the interaction is non-covalent and irreversible in nature and not dependent upon engagement of Cysteine
481. Raw 350 / 330 nm ratio data are shown in the top panel. The first derivative of the ratio is shown in the bottom panel. Melting temperatures, as calculated
from the maximum in the first derivative, are shown as dashed lines in the respective colors. Data are averages from three technical replicates
Figure 3
A
G
wt
wt BTK
B
H
wt BTK
wt BTK + Compound 1
+ Compound

Figure 3 A-F: UV chromatograms of wild-type BTK after incubation with compound: A: BTK only; B: BTK and Compound 1; C: BTK and Compound 3; D:
BTK and Compound 4; E: BTK and Compound 6; F: BTK and Compound 7. BTK eluted with a retention time of 7.9 min - 8.0 min. Chromatographic peaks
highlighted with a hash correspond to the elution of the respective unreacted compounds. G-L: deconvoluted mass spectra of wild-type BTK; G: BTK only; H:
BTK and Compound 1 indicating no covalent engagement, only the mass of the unreacted BTK was detected; I: BTK and Compound 3 indicating a covalent
adduct; J: BTK and Compound 4 indicating a covalent adduct; K: BTK and Compound 6 indicating a covalent adduct; L: BTK and Compound 7 indicating a
covalent adduct. Masses marked with an asterisk indicate unreacted BTK; in the inserts a zoom into the corresponding region of interest is shown; the blue line
indicates the measured spectrum and the red line indicates the simulated spectrum, respectively
Figure 4
A
G
C481S BTK
C481S BTK
B
H
C481S BTK
C481S BTK + Compound 1

Figure 4 A-F: UV chromatograms of C481S BTK after incubation with compound: A: BTK only; B: BTK and Compound 1; C: BTK and Compound 3; D:
BTK and Compound 4; E: BTK and Compound 6; F: BTK and Compound 7. BTK eluted with a retention time of 7.8 min – 7.9 min. Chromatographic peaks
highlighted with a hash correspond to the elution of the respective unreacted compounds. G-L: deconvoluted mass spectra of C481S BTK; G: BTK only; H: BTK
and Compound 1; I: BTK and Compound 3; J: BTK and Compound 4; K: BTK and Compound 6; L: BTK and Compound 7. None of these compounds are able
to form a covalent adduct with BTK481S. In the inserts a zoom into the corresponding region of interest is shown; the blue line indicates the measured spectrum
and the red line indicates the simulated spectrum, respectively.
Figure 5

Figure 5. Compounds were screened for selectivity against a panel of 135 different kinases at DiscoverX using the KINOMEscanTM
profiling assay. This is a reporter displacement assay with detection of displaced DNA-conjugate reporter molecules conducted by
qPCR. Incubations with test molecules were conducted at 10 uM for one hour at room temperature in PBS. Scale indicates extent of
remaining activity under assay conditions with 0% corresponding to full inhibition.
Figure 6
Figure 6. The 1.44 Å resolution BTK structure in complex with covalent irreversible inhibitor Compound 3. (A) Close-up view of the binding of Compound 3.
Red arrow indicates the covalent bond between Compound 3 and Cys481. (B) Close-up view of overlay of active site of BTK bound to Compound 3 (yellow) and
Compound 1 (black). Protein structure is shown as cartoon, inhibitor and selected residue side chains are depicted using stick model. Water molecules that bridge
the interaction between the protein and the inhibitor are shown as red spheres. Magenta and black dashed lines represent hydrogen bonds between BTK kinase
domain with Compounds 1 and 3, respectively.
Figure 7

Figure 7 Overall binding pocket of Compound 3 with BTK kinase domain. Protein residues and compound 3 are shown in cyan and yellow sticks respectively.
Water molecules are represented as red spheres and hydrogen bonds as dashed line. The red arrow indicates the covalent bond between Compound 3 and Cys481.
Figure 8a
1
Ibrutinib on-DNA
Dasatinib on-DNA
No compound on-DNA
0.1
0.01
0.001
0.0001
Figure 8b

Figure 8c
Figure 8. Quantitative PCR-derived enrichment data for on-DNA ibrutinib and dasatinib under model selection conditions. 8a shows relative enrichment for
different denaturants and capture matrices in the absence of sheared salmon sperm DNA and a DNA-Encoded Chemical Library super-mix, while 8b shows the
same in the presence of sheared salmon sperm DNA and a DNA-Encoded Chemical Library super-mix. Under these conditions, guanidine and IMAC magnetic
beads provided superior discrimination between ibrutinib (covalent, irreversible) and dasatinib (non-covalent, reversible). 6c indicates the experimental scheme.
Figure 9

Figure 9. Affinity-mediated selection output of Library 3 (Table 1) against 1 µM BTK incubated for 18 hours. The x- and y-axes represent building blocks in the
second and third cycles of chemistry respectively. The vertical axis is a relative enrichment metric for each combination of building blocks B and C (negative log10
of the asymptotic significance value) with a value of >10 considered to be statistically significant.. Instances (compounds) are randomly displaced by 0.5% to
indicate the extent to which building block variation at the first cycle of chemistry is tolerated. The extent to which enriched building block combinations have
been seen to enrich against prior targets is represented by colour with green representing the least promiscuous (building block combinations never previously seen
to have been enriched against any target, i.e. negative log10 of the asymptotic significance value has never previously exceeded 10) and red the most promiscuous.
Clusters colored in green had never been seen to enrich before in other screens against other targets. Compound 3 was found within the enriched cluster circled in
dark blue. Closely related compounds to compound 3 with variation at building block C (see Table 4) are circled in light blue. Compound 4 was found within the
enriched cluster circled in red.
Figure 10

Figure 10. Profile of the enrichment of the building block combination defining Compound 3 across all affinity-mediated selections conducted within the selection
campaign. The vertical axis is a relative enrichment metric for each combination of building blocks B and C (negative log10 of the asymptotic significance value)
with a value of >10 considered to be statistically significant. Each off-target contained at least one cysteine, additional off-target information is shown in
Supplementary Table S1.
Table 1. DNA-Encoded chemical libraries used in this study
Library
First Chemical Step
Second Chemical Step
Third Chemical Step
Reductive Amination with
2,341 Amines
Electrophile
None
Unique Compounds in Library
110,261,100
1
Acylation with
Acylation with
300 Amino Acids
157 Formyl Acids
2
3
4
Acylation with
Acylation with
Acylation with
None
225,000,000
26,183,954
157,103,670
300 Amino Acids
300 Amino Acids
2,500 Carboxylates
Acylation with
Acylation with
Acylation with
Acrylamide
305 Amino Acids
293 Amino Acids
293 Amino Acids
Acylation with
Acylation with
Acylation with
Six, including
305 Amino Acids
293 Amino Acids
293 Amino Acids
range of epoxides
Table 2. Compounds reported in this study

IC50 Value
RDA wt BTK
Library
Compound
1
Structure
H
O
H
N
N
N
0.55 nM (Cuozzo et al 2017);
3.8 nM (this work)
N
1
N
O
O
H
N
H
H
NH
N
N
2
3
3
4
4
4
2
3
4
5
6
7
23 nM
NH2
N
O
O
N
O
H
O
H
N
O
O
4.8 µM (1h)
36 µM (1h)
0.23 µM (1h)
58 nM (1h)
84 nM (1h)
N
N
NH
O
O
O
HN
NH
N
NC
O
O
O
NH
H
N
N
F
O
H
N
O
O
O
*
O
HN
H
N
O
N
H
O
N
O
NH
H
H
N
N
H
O
*
O
O
N
Table 3

BTK Wild-Type
BTK C481S
kinact / Ki
(per
second per
M)
Residence
Time /
min
Residence
Time /
min
Compound
1
IC50 at 1h / µM
kinact / per second
Ki / µM
ΔTm / °C MS/%
IC50 at 1h / µM
ΔTm / °C MS/%
0.00378 ± 0.00036
9 ± 2
NA
NA
NA
13.2 ± 0.1
0
0.00304 ± 0.00036
22 ± 3
13.4 ± 0.1
0
2
3
4
5
6
7
0.0227 ± 0.0023
4.76 ± 0.54
34 ± 4
NA
NA
ND
NA
NA
ND
NA
NA
NA
0.0562 ± 0.0036
>5.92E+01
>5.92E+01
>5.92E+01
>5.92E+01
>5.92E+01
14 ± 1
NA
NA
0
59.2 ± 1.0 6.55 ± 0.15 97.9
0.10 ± 0.06
36.4 ± 4.6
NA
NA
94.4
NA
NA
0
0.232 ± 0.034
0.0581 ± 0.0048
0.0842 ± 0.0062
NA
1.54E-02 ± 0.09E-02 9.27 ± 1.97 1660 ± 365
NA
NA
0
NA
8.22E-03 ± 1.98E-04 1.25 ± 0.11 6580 ± 617 4.51 ± 0.06 92.3
9.41E-03 ± 3.78E-04 1.93 ± 0.32 4880 ± 826 6.36 ± 0.10 95.2
NA
0.23 ± 0.33
-0.15 ± 0.08
NA
NA
0
Table 3. Summary of data determined for Compounds 1-7 using the Reporter Displacement Assay (RDA), Thermal Shift Assay (nanoDSF) and intact protein LC-
MS. IC50 values, residence times, kinact, Ki and kinact / Ki were determined in the Proteros RDA. ΔTm values were obtained by subtracting the Tm of the apo BTK
(wild-type or C481S) from the Tm in presence of 50 µM compound as determined by nanoDSF. The amounts of protein bound to compound (MS/%) were calculated
by comparing the peak heights in the deconvoluted MS spectrum of unreacted and reacted protein.
Table 4. Tolerated building block variability as exhibited by co-enriched structurally related compounds in the selection output
Library
Compound
First Chemical
Cycle Building
Blocks (A)
Second Chemical
Cycle Building
Blocks (B)
Third Chemical
Cycle Building
Blocks (C)
Terminal
Electrophile
1
1
None
NH2
N
N
NH2
N
S
H
N
N
O
O
NH2
Fmoc
N
O
OH
CHO
H
N
N
Linker
O
2
2
None
NHFmoc
O
OH
NHFmoc
O
OH
F
H
N
NHFmoc
O
O
Fmoc
N
O
OH
O
H
N
OH
S
N
Linker
H2N
H
H
3
3
Variable,
dispensable
N
HO
Fmoc
O
N
Fmoc
N
O
O
HO
H
N
Linker
O
HN Fmoc
HO
N
O
Fmoc
HO

3
4
Variable,
dispensable
O
N
OH
OH
HO
Fmoc
NHFmoc
NHFmoc
O
O
O
N
NC
NC
HO
Fmoc
F
O
O
N
HO
HO
Fmoc
4
5
Variable,
dispensable
HO
O
NHFmoc
F
HO
O
H
N
NHFmoc
Fmoc
N
O
H
N
HO
O
Linker
CF3
O
HO
O
4
6 & 7
Variable,
O
NHFmoc
NHFmoc
dispensable
HO
Cl
Cl
NC
O
NHFmoc
HO
O
NHFmoc
HO
O
O
O
O
HO
NC
NHFmoc
HO
O
HO
NHFmoc
HO
O
O2N
O
Table 5
Data Collection

X-ray source
Wavelength [Å]
Detector
Space group
Cell: a; b; c; [Å]
α ; β; γ; [◦]
Resolution [Å]
Unique reflections
Multiplicity
Completeness [%]
Rsym [%]
PXIII/X06DA (SLS1)
1.0000
PILATUS 2M
P 21 21 2
71.62; 103.23; 37.86
90.0; 90.0; 90.0
1.44 (1.69-1.44)
50687 (19233
3.2 (3.1)
98.0 (98.9)
5.6 (44.6)
Rmeas [%]
6.7 (53.7)
<I/σ(I)>
13.54 (2.78)
Refinement
Resolution [Å]
Number of reflections (working /test)
Rcryst [%]
58.84-1.44
45594 / 5090
16.1
Rfree[%]
20.4
Total number of atoms:
Protein
Water
2390
333
30
Ligand
Deviation from ideal geometry:
Bond lengths [Å]
Bond angles [°]
0.009
1.45
Ramachandran plot
Most favoured regions [%]
allowed regions [%]
Disallowed regions [%]
95.1
4.9
0.0
I/σ(I) corresponds to the average of the intensity divided by its average standard deviation.
Rcrsyt/free = ∑hkl│|FobsFcalc|│/ ∑hkl|Fobs|, where Fobs and Fcalc are the observed and calculated structure factors, respectively. Rfree is the same as Rwork,
calculated for the 10% of the data that were randomly omitted from refinement
Table 5 Data collection and refinement statistics for structure determination of the complex of BTK kinase domain with
Compound 3 using X-ray diffraction and crystallography.
8.
9.
Dungo, R.T. and Keating, G.M., 2013. Afatinib: first global
approval. Drugs, 73(13), pp.1503-1515.
Emsley, P. and Cowtan, K., 2004. Coot: model-building tools for
molecular graphics. Acta crystallographica section D: biological
crystallography, 60(12), pp.2126-2132.
References
1.
Burger, J.A., Tedeschi, A., Barr, P.M., Robak, T., Owen, C., Ghia,
P., Bairey, O., Hillmen, P., Bartlett, N.L., Li, J. and Simpson, D.,
2015. Ibrutinib as initial therapy for patients with chronic
lymphocytic leukemia. New England Journal of Medicine,
373(25), pp.2425-2437.
10. Favalli, N., Bassi, G., Scheuermann, J. and Neri, D., 2018.
DNA‐encoded chemical libraries–achievements and remaining
challenges. FEBS letters, 592(12), pp.2168-2180.
11. Gehringer, M. and Laufer, S.A., 2018. Emerging and re-emerging
warheads for targeted covalent inhibitors: applications in
medicinal chemistry and chemical biology. Journal of medicinal
chemistry, 62(12), pp.5673-5724.
12. Goodnow, R.A., Dumelin, C.E. and Keefe, A.D., 2017. DNA-
encoded chemistry: enabling the deeper sampling of chemical
space. Nature Reviews Drug Discovery, 16(2), pp.131-147.
13. Greig, S.L., 2016. Osimertinib: first global approval. Drugs, 76(2),
pp.263-273.
14. Ghosh, A.K., Samanta, I., Mondal, A. and Liu, W.R., 2019.
Covalent inhibition in drug discovery. ChemMedChem, 14(9),
p.889.
15. Jackson, P.A., Widen, J.C., Harki, D.A. and Brummond, K.M.,
2017. Covalent modifiers: A chemical perspective on the
reactivity of α, β-unsaturated carbonyls with thiols via hetero-
Michael addition reactions. Journal of medicinal chemistry, 60(3),
pp.839-885.
2.
Caldwell, R.D., Qiu, H., Askew, B.C., Bender, A.T., Brugger, N.,
Camps, M., Dhanabal, M., Dutt, V., Eichhorn, T., Gardberg, A.S.
and Goutopoulos, A., 2019. Discovery of evobrutinib: an oral,
potent, and highly selective, covalent Bruton’s tyrosine kinase
(BTK) inhibitor for the treatment of immunological diseases.
Chan, A.I., McGregor, L.M., Jain, T. and Liu, D.R., 2017.
Discovery of a covalent kinase inhibitor from a DNA-encoded
small-molecule library × protein library selection. Journal of the
American Chemical Society, 139(30), pp.10192-10195.
Clark, M.A., Acharya, R.A., Arico-Muendel, C.C., Belyanskaya,
S.L., Benjamin, D.R., Carlson, N.R., Centrella, P.A., Chiu, C.H.,
Creaser, S.P., Cuozzo, J.W. and Davie, C.P., 2009. Design,
synthesis and selection of DNA-encoded small-molecule libraries.
Nature chemical biology, 5(9), pp.647-654
Clark, M., Keefe, A.D., Guilinger, J., Zhang, Y. and Tian, X., X-
CHEM Inc, 2020. Covalent BTK inhibitors and uses thereof. U.S.
Patent Application 16/097,093.
Cuozzo, J.W., Centrella, P.A., Gikunju, D., Habeshian, S., Hupp,
C.D., Keefe, A.D., Sigel, E.A., Soutter, H.H., Thomson, H.A.,
Zhang, Y. and Clark, M.A., 2017. Discovery of a Potent BTK
Inhibitor with a Novel Binding Mode by Using Parallel Selections
with a DNA‐Encoded Chemical Library. ChemBioChem, 18(9),
pp.864-871.
3.
4.
5.
6.
16. Kabsch, W., 2010. Xds. Acta Crystallographica Section D:
Biological Crystallography, 66(2), pp.125-132.
17. Kuglstatter, A., Wong, A., Tsing, S., Lee, S.W., Lou, Y.,
Villaseñor, A.G., Bradshaw, J.M., Shaw, D., Barnett, J.W. and
Browner, M.F., 2011. Insights into the conformational flexibility
of Bruton's tyrosine kinase from multiple ligand complex
structures. Protein Science, 20(2), pp.428-436.
18. Lagoutte, R., Patouret, R. and Winssinger, N., 2017. Covalent
inhibitors: an opportunity for rational target selectivity. Current
Opinion in Chemical Biology, 39, pp.54-63.
7.
Di Paolo, J.A., Huang, T., Balazs, M., Barbosa, J., Barck, K.H.,
Bravo, B.J., Carano, R.A., Darrow, J., Davies, D.R., DeForge,
L.E. and Diehl, L., 2011. Specific Btk inhibition suppresses B
cell–and myeloid cell–mediated arthritis. Nature chemical biology,
7(1), p.41.

19. Lombardo, L.J., Lee, F.Y., Chen, P., Norris, D., Barrish, J.C.,
Behnia, K., Castaneda, S., Cornelius, L.A., Das, J., Doweyko,
A.M. and Fairchild, C., 2004. Discovery of N-(2-chloro-6-methyl-
phenyl)-2-(6-(4-(2-hydroxyethyl)-piperazin-1-yl)-2-
protocols, HAT performed the non-covalent selections, ADK and
JPG designed the covalent selection protocols, JPG performed the
non-covalent selections, MAG processed the selection output
sequencing data, ADK and HAT analyzed the non-covalent
selection output data, ADK and JPG analyzed the covalent
selection output data, YL and YZ coordinated the outsourced off-
DNA compound synthesis, YL performed compound synthesis,
JWC coordinated the sourcing of BTK, JWC and JL established
the initial TR-FRET (data not shown), JL ran the initial TR-FRET
(data not shown), AA interpreted the crystallography results, MA
established purification protocols for BTK, AB provided scientific
and technical support for RDA and nanoDSF assays, MD
performed the RDA assay, RK optimized crystallization
conditions, SK coordinated the crystallography, AL solved the
Compound 3 complex crystal structure, LL designed and
coordinated the RDA and nanoDSF experiments, KM purified the
wild-type and C481S BTK proteins, MM solved the Compound 1
complex crystal structure, KP performed nanoDSF and RDA
assays, MS performed and analyzed all intact protein LC-MS
studies, DKM designed overall compound assay strategy and
coordinated all Proteros activities.
methylpyrimidin-4-ylamino) thiazole-5-carboxamide (BMS-
354825), a dual Src/Abl kinase inhibitor with potent antitumor
activity in preclinical assays. Journal of medicinal chemistry,
47(27), pp.6658-6661.
20. Marcotte, D.J., Liu, Y.T., Arduini, R.M., Hession, C.A.,
Miatkowski, K., Wildes, C.P., Cullen, P.F., Hong, V., Hopkins,
B.T., Mertsching, E. and Jenkins, T.J., 2010. Structures of human
Bruton's tyrosine kinase in active and inactive conformations
suggest a mechanism of activation for TEC family kinases. Protein
Science, 19(3), pp.429-439.
21. Neumann, L., von König, K. and Ullmann, D., 2011. HTS reporter
displacement assay for fragment screening and fragment evolution
toward leads with optimized binding kinetics, binding selectivity,
and thermodynamic signature. Methods in enzymology, 493,
pp.299-320.
22. Project, C.C., 1994. The CCP4 suite: programs for protein
crystallography. Acta crystallographica. Section D, Biological
crystallography, 50(Pt 5), pp.760-763.
23. Ray, S. and Murkin, A.S., 2019. New electrophiles and strategies
for mechanism-based and targeted covalent inhibitor design.
Biochemistry, 58(52), pp.5234-5244.
24. Schneider, E.V., Böttcher, J., Huber, R., Maskos, K. and
Neumann, L., 2013. Structure–kinetic relationship study of
CDK8/CycC specific compounds. Proceedings of the National
Academy of Sciences, 110(20), pp.8081-8086.
25. Sequist, L.V., Soria, J.C., Goldman, J.W., Wakelee, H.A.,
Gadgeel, S.M., Varga, A., Papadimitrakopoulou, V., Solomon,
B.J., Oxnard, G.R., Dziadziuszko, R. and Aisner, D.L., 2015.
Rociletinib in EGFR-mutated non–small-cell lung cancer. New
England Journal of Medicine, 372(18), pp.1700-1709.
26. Spaargaren, M., De Rooij, M.F.M., Kater, A.P. and Eldering, E.,
2015. BTK inhibitors in chronic lymphocytic leukemia: a glimpse
to the future. Oncogene, 34(19), pp.2426-2436.
Statement on Conflicts of Interest
All authors were employees of X-Chem Inc. or Proteros
Biostructures GmbH at the time they conducted work on this
project and declare no conflicts of interest.
27. Zambaldo, C., Daguer, J.P., Saarbach, J., Barluenga, S. and
Winssinger, N., 2016. Screening for covalent inhibitors using
DNA-display of small molecule libraries functionalized with
cysteine reactive moieties. MedChemComm, 7(7), pp.1340-1351.
28. Zhang, T., Hatcher, J.M., Teng, M., Gray, N.S. and Kostic, M.,
2019. Recent advances in selective and irreversible covalent
ligand development and validation. Cell chemical biology, 26(11),
pp.1486-1500.
29. Zhu, Z., Grady, L.C., Ding, Y., Lind, K.E., Davie, C.P., Phelps,
C.B. and Evindar, G., 2019. Development of a Selection Method
for Discovering Irreversible (Covalent) Binders from a DNA-
Encoded Library. SLAS DISCOVERY: Advancing Life Sciences
R&D, 24(2), pp.169-174.
30. Zimmermann, G., Rieder, U., Bajic, D., Vanetti, S., Chaikuad, A.,
Knapp, S., Scheuermann, J., Mattarella, M. and Neri, D., 2017. A
Specific and Covalent JNK‐1 Ligand Selected from an Encoded
Self‐Assembling Chemical Library. Chemistry–A European
Journal, 23(34), pp.8152-8155.
Supplementary Material
Supplementary materials exist in a separate document.
Acknowledgments
X-Chem Inc. and Proteros Biostructures GmbH are
acknowledged for their support of this work. This research did not
receive any specific grant from funding agencies in the public,
commercial of not-for-profit sectors.
Author Contributions
MAC, PAC and YZ designed the library chemical schemes,
PAC synthesized libraries 1 and 2, SH synthesized the non-
electrophilic core of libraries 3 and 4, XT installed the
electrophiles, ADK and HAT designed the non-covalent selection