Depsipeptides Featuring a Neutral P1 Are Potent Inhibitors of Kallikrein-Related Peptidase 6 with On-Target Cellular Activity

Article abstract of DOI:10.1021/acs.jmedchem.8b01106

Kallikrein-related peptidase 6 (KLK6) is a secreted serine protease that belongs to the family of tissue kallikreins (KLKs). Many KLKs are investigated as potential biomarkers for cancer as well as therapeutic drug targets for a number of pathologies. KLK6, in particular, has been implicated in neurodegenerative diseases and cancer, but target validation has been hampered by a lack of selective inhibitors. This work introduces a class of depsipeptidic KLK6 inhibitors, discovered via high-throughput screening, which were found to function as substrate mimics that transiently acylate the catalytic serine of KLK6. Detailed structure-activity relationship studies, aided by in silico modeling, uncovered strict structural requirements for potency, stability, and acyl-enzyme complex half-life. An optimized scaffold, DKFZ-251, demonstrated good selectivity for KLK6 compared to other KLKs, and on-target activity in a cellular assay. Moreover, DKFZ-633, an inhibitor-derived activity-based probe, could be used to pull down active endogenous KLK6.

Full text of DOI:10.1021/acs.jmedchem.8b01106

Subscriber access provided by Kaohsiung Medical University
Article
Depsipeptides featuring a neutral P1 are potent inhibitors of
kallikrein-related peptidase 6 with on-target cellular activity
Elena De Vita, Peter Schüler, Scott Lovell, Jasmin Lohbeck, Sven Kullmann,
Eitan Rabinovich, Amiram Sananes, Bernd Heßling, Veronique Hamon, Niv
Papo, Jochen Heß, Edward W. Tate, Nikolas Gunkel, and Aubry K. Miller
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01106 • Publication Date (Web): 13 Sep 2018
Downloaded from http://pubs.acs.org on September 14, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Depsipeptides featuring a neutral P1 are potent
inhibitors of kallikrein-related peptidase 6 with
on-target cellular activity
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a,b
a
c
a
a
Elena De Vita, Peter Schüler, Scott Lovell, Jasmin Lohbeck, Sven Kullmann, Eitan
d
Rabinovich, Amiram Sananes, Bernd Heßling, Veronique Hamon, Niv Papo, Jochen Hess,
d
e,†
f
d
g,h
c a,i
Edward W. Tate, Nikolas Gunkel and Aubry K. Miller*
a,i
a
Cancer Drug Development Group, German Cancer Research Center (DKFZ), Heidelberg
9120, Germany
6
b
Biosciences Faculty, University of Heidelberg, Heidelberg 69120, Germany
c
Department of Chemistry, Imperial College London, Exhibition Road, London SW7 2AZ, UK
d
Department of Biotechnology Engineering and the National Institute of Biotechnology in the
Negev, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel
e
Center for Molecular Biology, University of Heidelberg, Heidelberg 69120, Germany
f
European Screening Centre, Biocity Scotland, University of Dundee, Newhouse ML1 5UH, UK
g
Department of Otorhinolaryngology, Head and Neck Surgery, Heidelberg University Hospital,
Heidelberg 69120, Germany
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
h
Research Group Molecular Mechanisms of Head and Neck Tumors, German Cancer Research
Center (DKFZ), Heidelberg 69120, Germany
i
German Cancer Consortium (DKTK), Heidelberg 69120, Germany
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ABSTRACT
Kallikrein-related peptidase 6 (KLK6) is a secreted serine protease that belongs to the family of
tissue kallikreins (KLKs). Many KLKs are investigated as potential biomarkers for cancer as
well as therapeutic drug targets for a number of pathologies. KLK6, in particular, has been
implicated in neurodegenerative diseases and cancer, but target validation has been hampered by
a lack of selective inhibitors. This work introduces a class of depsipeptidic KLK6 inhibitors,
discovered via high-throughput screening, which were found to function as substrate mimics that
transiently acylate the catalytic serine of KLK6. Detailed structure–activity relationship studies,
aided by in silico modeling, uncovered strict structural requirements for potency, stability, and
acyl-enzyme complex half-life. An optimized scaffold, DKFZ-251, demonstrated good
selectivity for KLK6 compared to other KLKs, and on-target activity in a cellular assay.
Moreover, DKFZ-633, an inhibitor-derived activity-based probe, could be used to pull down
active endogenous KLK6.
INTRODUCTION
The largest protease gene cluster in the human genome encodes a family of 15 secreted serine
proteases commonly known as the tissue kallikreins (KLKs), which encompass human tissue
kallikrein (KLK1) and 14 kallikrein-related peptidases (KLK2–15). Once considered to function
ACS Paragon Plus Environment
2

Page 3 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
largely as redundant extracellular matrix-degrading enzymes, it is now well recognized that
1
KLKs are critical regulators of finely tuned and complex cascade systems (e.g. skin
desquamation, brain function, seminal liquefaction, and immune control), with tissue-dependent
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
expression and function. Beyond their physiological roles, aberrant KLK expression and activity
have been implicated in a variety of pathologies, including skin diseases, neurological disorders,
3
4
and cancer. In the latter case, KLKs are extensively studied as biomarkers. For example,
KLK3, more commonly known as prostate-specific antigen (PSA), is one of the most
5
well-known of all cancer biomarkers. In addition to their usefulness as clinical markers, KLKs
are now considered to directly contribute to a variety of cancer hallmarks, such as
chemo-resistance, cell growth, invasion, migration, and modulation of the tumor
3a
microenvironment. Due to a lack of high quality, selective activity-based probes (ABPs),
however, few attempts to quantify the active enzyme fraction of endogenous KLK proteins have
6
been made.
Despite being one of the most abundant proteinases in the central nervous system, very little is
known about kallikrein-related peptidase 6 (KLK6). Four different groups isolated KLK6 in the
mid-1990s, and, not immediately recognizing that they were working with the same protein, gave
7
it multiple names, i.e. Protease M, Zyme, Neurosin, and Myelencephalon-specific protease.
After the sequencing of the human genome, KLK6 was correctly classified as a member of the
kallikrein family, and by then had captured attention for being highly expressed in breast and
7
a
ovarian cancer. KLK6 is secreted in its zymogen pro-form and activated via proteolytic
cleavage (Lys15↓Leu16). Auto-inactivation can occur via hydrolysis at Arg76↓Glu77 and this
8
has been postulated as a negative-feedback inhibition mechanism. In vitro studies have shown
that this protease is able to cleave myelin basic protein, alpha-synuclein, and the beta-amyloid
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
7b,9
precursor protein.
Moreover, it can activate protease-activated receptors (PARs), thus
1
0
triggering intracellular signaling and influencing cellular processes. Aberrant expression of
KLK6 has been reported in various neurodegenerative diseases such as multiple sclerosis,
9
b, 11
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Alzheimer’s, and Parkinson’s disease.
Notably, two studies involving auto-immunization of
a multiple sclerosis mouse model with KLK6-neutralizing antibodies showed a significantly
1
2
delayed onset and reduced severity of the disease. Recent reports have also suggested that
KLK6 is an active driver of tumor development. Keratinocytes within the malignant melanoma
microenvironment, but not primary tumor cells, produce high amounts of KLK6, which then
promotes tumor migration and invasiveness via the activation of protease-activated receptor 1
1
3
(PAR1). In ovarian cancer, KLK6 overexpression negatively correlates with survival and has
14
15
also been postulated to play a role in tumor invasiveness. In gastric cancer and
1
6
glioblastoma, KLK6 has been shown to contribute to therapy resistance. In head and neck
cancer on the other hand, KLK6 silencing was found to promote cell proliferation and an
1
7
epithelial to mesenchymal transition (EMT).
While the field of KLK inhibitor discovery is relatively underdeveloped, recent reviews
1
a, 18
highlight the interest and need for further work.
A common approach for the development of
inhibitors has been to utilize natural peptidic inhibitors as starting points, exploiting
combinatorial or designed mutations in order to achieve selectively toward the targeted KLK.
1
9
Still, the development of sub-type selective KLK inhibitors remains challenging. Better success
has been achieved with the chymotryptic-like KLK7, which has unique substrate specificity in
20
comparison to other KLKs. To the best of our knowledge, four accounts of small molecule
KLK6 inhibitors exist to date, including two classes of small molecules discovered by virtual
2
1
22
screen, a series of pseudopeptides (based on the activating peptide of PAR2), and a recent
ACS Paragon Plus Environment
4

Page 5 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
23
coumarin-derived mechanism-based inhibitor (Chart 1). As most of these reports show no data
for activity against related proteases, the selectivity of these compounds remains unknown at
present. KLK6, like most other KLKs, has trypsin-like activity, governed by an acidic Asp189
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
trypsin numbering) in the S1 pocket, which dictates the preference for a basic arginine or lysine
residue in P1 of KLK6 substrates. KLK6 also contains a Ser190, a feature that has been found
2
4
particularly difficult to target with neutral P1 inhibitors, with few examples in the literature.
Usually, inhibitors of Ser190-containing proteases require the presence of a strongly basic
residue in P1 for binding, but this can lead to unfavorable pharmacological properties due to the
2
5
inherent charged state of the molecules. In addition, the highly conserved nature of Asp189
(and Ser190 when present) in the S1 pocket of trypsin-like serine proteases has been recognized
26
to complicate the development of selective inhibitors. Motivated by the pressing need for
selective KLK6 chemical probes for use as validation tools for in depth biological studies, and
aware of the existing challenges, we set out to discover potent and selective inhibitors of KLK6.
Chart 1. Known KLK6 inhibitors.
RESULTS AND DISCUSSION
High Throughput Screening Provides a Neutral Depsipeptidic Hit Scaffold
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In order to discover novel KLK6 inhibitors, we chose a high-throughput screen (HTS)
approach at the outset of this project. A triple mutant of KLK6 (R74G, R76Q, and N132Q),
resistant to autocleavage and lacking a glycosylation site, was recombinantly expressed in insect
2
1a,27
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
cells, and purified as previously reported (Figure S1).
An enzymatic assay, which monitors
the activity of KLK6 by cleavage of the fluorogenic substrate N-Boc-Phe-Ser-Arg-7-amino-4-
methylcoumarin (FSR-AMC), was optimized for HTS and validated with the known inhibitors
AEBSF and gabexate. After a successful pilot screen (Figure S2), a full HTS of a library of more
than 80,000 substances tested at 40 ꢀM gave a hit rate of 0.05% when applying a threshold
cut-off at 30% inhibition of substrate turnover (Figure 1A).
Dose response curves of 124 hits were determined in duplicate and 18 substances were found
to have IC values between 10–100 µM. These 18 hits were grouped into similarity clusters and
5
0
singletons, and 123 structurally related substances were ordered from commercial vendors. The
resulting 141 compounds were then tested in dose response against KLK6, and commercially
sourced trypsin, thrombin and factor Xa. These other three serine proteases were selected as a
counter selectivity screen due to their close structural similarity with KLK6 and to their
physiological importance. The selectivity screen revealed only one class of substances, a series
of depsipeptides (e.g. 1–5), which exhibited at least low micromolar IC values against KLK6
5
0
and some selectivity against all three other peptidases (Figure 1B). We found that while some
activity against trypsin was observed with these substances, they were generally quite inactive
against both factor Xa and thrombin. Encouragingly, some structurally similar substances (e.g.,
1
, 2 and 5) had very different IC values against KLK6, giving us some confidence that the hits
50
were “true” positives. The general structure of these substances can be dissected into three main
components: the depsipeptidic glycolic linker (L, Figure 1B, black), which is flanked by a
ACS Paragon Plus Environment
6

Page 7 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
“
western” aromatic group (W, Figure 1B, blue) and an “eastern” N-phenyl-4-amino-3,5-
dimethylpyrazole (E, Figure 1B, red). Compound 1, containing a western N-linked
2-methylindole stood out as having the best combination of potency against KLK6 and
selectivity over trypsin (IC s of 640 nM and 70 µM, respectively). We were further intrigued by
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
the conspicuous lack of a strongly basic group in 1, and were cautiously optimistic about the
possibility to develop a selective inhibitor for a Ser190-containing serine protease.
Figure 1. HTS and selectivity screen results. (A) Graphical representation of an HTS, which
measured reduced turnover of FSR-AMC by KLK6. Inhibitors were tested at 40 µM. Orange line
represents 30% inhibition cut-off. (B) Structures of five depsipeptides and their measured IC50
values against four proteases. Main structural features are defined by color code: W (blue) =
“western” aromatic group; L (black) = depsipeptidic linker; E (red) = “eastern” aromatic group.
(C) Structure and IC data of the previously reported inhibitor 6. Try = trypsin; Thr = thrombin;
5
0
FXa = factor Xa.
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
We also prepared the known zwitterionic KLK6 inhibitor 6 as a validation compound, and
2
1a
determined its selectivity profile (Figure 1C).
The measured KLK6 IC50 value was in
accordance with the literature (measured: 890 nM; reported: 1.8 µM) and the substance was
found to potently inhibit trypsin (IC = 180 nM), while being ineffective against both thrombin
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
and factor Xa. The fact that 6 inhibits trypsin well is perhaps not surprising considering its high
homology with KLK6 and that 6 contains a basic polar benzylamine P1 group.
Depsipeptides Act as Transient Quiescent Affinity Labelers of KLK6
The depsipeptidic structure of 1, which contains an electrophilic ester functional group, gave
2
8
rise to the question whether the substance might function as a covalent inhibitor of KLK6. We
2
9
therefore prepared a fully peptidic analog (7) as well as carbamate 8 (Figure 2A). These were
found to be inactive against KLK6, suggesting that reaction of the ester bond is required for
compound activity. Moreover, methyl ester 9 and acetate 10 were prepared so as to “isolate” the
western and eastern portions, respectively (Figure 2A). Both 9 and 10 were inactive against
KLK6, showing that while the central ester bond of 1 is required for activity, both the western
and eastern moieties must also be in place. Nano-flow reverse-phase chromatography combined
with electrospray ionization mass spectrometry (ESI-MS) of the intact protein (7 µM KLK6,
Figure S3) incubated with excess 1 (200 µM) was performed, and, after 1 h, a mass shift of
1
71 Da was observed with no un-modified protein detected (Figure 2B). This mass shift
corresponds to the carboxy part of ester 1. Conducting this experiment with analogs of 1 bearing
different W groups gave different mass shifts (Figure S4). In addition, 11 (DKFZ-251), a
regioisomer of 1, produced the same 171 Da shift (Figure 2B). Tryptic digest of the modified
enzyme showed a 171 Da shift only on those peptides containing Ser195 (Figure S5), and ESI-
MS of an Ser195A mutant indicated no mass shift when treating with 1 or 11 (Figure 2B).
ACS Paragon Plus Environment
8

Page 9 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Depsipeptides are transient covalent inhibitors. (A) Structures of 7–10. (B) ESI-MS of
KLK6 (24,374 Da) treated with 1 or 11 (DKFZ-251) fully shifts the mass to 24,545 Da (ꢁDa =
1
71) after 1 h. Longer incubation times show recovery of unmodified KLK6 with 1, but not with
1
1. An S195A mutant of KLK6 (24,358 Da) is neither modified by 1 nor 11. (C) Reaction of 1
with the catalytic serine of KLK6, followed by the hydrolysis of the resulting acyl-enzyme
complex.
Observation of the covalent adduct of 1 with KLK6 via intact protein MS at incubation times
longer than 1 hour revealed that the covalent modification is transient. While the enzyme was
still almost completely modified after 24 h, most of it was in an unmodified form after 48 h
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
Figure 2B). Interestingly, the transient nature of the modification was found to be quite sensitive
to changes in the western moiety structure. For example, regioisomer 11 (DKFZ-251) showed no
recovery of native enzyme even after 72 h under the same conditions (Figure 2B).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
These data indicate that the inhibitors act as substrate mimics, undergoing cleavage of the ester
bond via nucleophilic attack by Ser195 to produce an acyl-enzyme complex that is resistant, but
not impervious, to hydrolysis (Figure 2C). The fact that 1 exhibits selectivity for KLK6 over
related serine proteases suggests that it is not simply a reactive, nonspecific affinity labeler.
Moreover, the finding that both the western and eastern moieties are important for inhibition
indicates that non-covalent interactions between the enzyme and inhibitor are required before
reaction can occur. Taken together these results are consistent with the inhibitors acting via
transient quiescent affinity labeling, as described by eq 1.
First, equilibrium is rapidly (presumed) established between free enzyme (E) and inhibitor (I)
with a non-covalent enzyme inhibitor complex (E-I) in a reversible manner. This step is guided
by an intrinsic affinity of the compound to the KLK6 active site. Once brought into close
proximity to the catalytic serine, the ester bond covalently reacts with KLK6, forming an
acyl-enzyme complex (E-I*). This step is irreversible as the alcohol part of the ester is lost to the
bulk solvent. The acyl-enzyme complex is hydrolyzed by water, after which, similarly to
*
endogenous protein substrates, the corresponding carboxylic acid (I ) is released. This step is
also irreversible, as the acid is not able to bind again to KLK6 (data not shown). When the
half-life of E-I* is long enough, the enzyme accumulates as E-I*, resulting in effective inhibition
of the enzymatic function.
ACS Paragon Plus Environment
10

Page 11 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Structure–Activity Relationship Studies Reveal Strict Requirements for Effective KLK6
Inhibition
Having successfully elucidated the mechanism of action (MOA) of hit compound 1, we set out
to explore the structure–activity relationship (SAR) of this class of compounds with respect to its
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
three components (W, L, and E, as defined in Figure 1B). We decided to utilize kinact/K values as
I
a metric to compare derivatives, working under the assumption that enzyme acylation would be
3
0
essentially irreversible during such an experiment. DKFZ-251 (11) behaved as expected,
showing a time-dependency between residual enzyme activity and pre-incubation time of
–
1 –1
enzyme with inhibitor (Figure 3A). From this data, a kinact/K of 1883 M s (95% C.I. 1593 to
I
2172) could be determined (Figure 3B). When the same assay was performed with 1, however,
no time-dependency, up to a 40 min pre-incubation, was observed. When the pre-incubation time
range was extended to 4 h, ester 1 appeared to “lose” potency over the assay time course
(Figure 3C). By testing a larger panel of structures (Figures S6–8), we observed that for some
compounds it was similarly inconclusive to measure kinact/K . In the case of such substances, the
I
half-life of E-I* is presumably too short to allow such a measurement. Unable to utilize kinact/KI
values to compare all structures, we decided to utilize “apparent” IC values (referred to as IC
50
50
from now on) after a relatively long pre-incubation time (30 min), as a compromise meant to
capture as much of the relevant features of eq. 1 for each substance as possible.
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Time dependent behavior of 1 and 11 (DKFZ-251). (A) KLK6 residual activity after
different pre-incubation times with, and at different concentrations of, 11. (B) Plot for
determination of kinact/K for 11. (C) KLK6 residual activity after different pre-incubation times
I
with, and at different concentrations of, 1. Residual activity in (A) and (C) was measured in
triplicate at each time point and concentration. kobs for each concentration in (A) was determined
by fitting to an exponential decay function.
All compounds were tested against both KLK6 and trypsin in order to monitor selectivity. A
steep SAR was identified for the W part of the molecule, as anticipated from the HTS, with
indoles representing the best scaffold (Table 1). A switch in indole substitution from N-linked
ACS Paragon Plus Environment
12

Page 13 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
1) to C3-linked (11, DKFZ-251) resulted in an increase in potency of almost 5 fold. The free
NH of 11 is required for potency as neither benzimidazole 12 nor N-methylindole 13 are
comparable in potency (not even to 1, hinting to both steric and electronic effects being
important). Remarkable was the effect of the methyl group in the 2-position of the indole, the
removal of which (14) decreased the activity relative to 11 more than 100 fold. The small methyl
group seems to represent the best fit, as a larger isopropyl substituent (15) also strongly
decreased potency. Further substitution of the indole with a chlorine atom produced moderate to
significant potency losses (16 and 17, respectively). While substitution of indole in 11 for
azaindole has little effect on KLK6 potency for 6- (18) and 7- (19) aza derivatives, it is
moderately to extremely detrimental for activity at positions 5 (20) and 4 (21), respectively.
Remarkably, the nitrogen in 7-azaindole 19 causes a dramatic loss of activity against trypsin
relative to 11, while 6-azaindole 18 loses almost all selectivity between the two enzymes. The
basic nature of the introduced nitrogen may in fact start triggering interactions with the
conserved Asp189 in the S1 pocket, and the placement of the nitrogen atom in 18 seems to be
geometrically optimal (v. infra).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. Structure–activity relationship data
a
IC (ꢀM)
KLK6
0.64 ± 0.29
0.13 ± 0.04
8.9 ± 4.3
6.2 ±2.4
50
Cmpd
1
Trypsin
38 ± 3
21 ± 6
>200
>200
150 ± 25
>200
34
1
1
1
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
13
14
15
16
17
18
15 ± 6
13 ± 8
0.25 ± 0.02
7.6 ± 1.4
0.10 ± 0.05
0.19 ± 0.04
0.49 ± 0.15
6.0 ± 3.9
>200
0.33
>200
25
1
2
2
9
0
1
61
2
2
2
2
2
2
2
2
3
2
3
4
5
6
7
8
9
0
>200
58 ± 4
110 ± 25
23 ± 4
0.018 ± 0.015
28 ± 2
>200
>200
>200
>200
0.38
>200
>200
61
>200
2.31 ± 0.04
>200
>200
31
32
33
34
35
36
4.4 ± 2.9
17.0 ± 0.4
6.99 ± 0.01
0.29 ± 0.14
7.5 ± 3.2
1.4 ± 0.1
1.2 ± 0.6
2.8 ± 0.1
>200
>200
>50
34
>200
>50
3
3
7
8
50
8.6 ± 0.6
a
IC values on KLK6 represent the mean of at least two independent experiments performed in
5
0
triplicate ± standard deviation after 30 min pre-incubation; IC values on trypsin represent the
5
0
mean of at least one experiment performed in triplicate (± standard deviation when multiple
independent experiments were performed) after 30 min pre-incubation.
The depsipeptidic scaffold (L) of DKFZ-251 (11) was explored in order to define the
requirements for its structure and reactivity (Table 1). Deletion (22) or insertion (23) of a
methylene between the indole and the ester resulted in inactive or significantly less potent
ACS Paragon Plus Environment
14

Page 15 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compounds, respectively. Moreover, the peptidomimetic nature of the depsipeptide was shown to
be critical, as the introduction of a further methylene between the ester and the amide almost
completely abolished activity (24). On the basis of the MOA elucidation, we next tried to
enhance the compound potency by increasing the reactivity of the ester. An electron-
withdrawing trifluoromethyl group was introduced on the carboxy (25) and on the alkoxy (26)
part of the ester. Only the latter compound resulted in a potency increase, while 25 showed
reduced potency that was comparable to the introduction of a simple methyl group (27). The di-
methylated derivative 28 was found to be completely inactive. A hybrid substance (29), based on
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
6 and 27, only somewhat counteracted the loss of potency given by the steric hindrance of the
methyl group. The electron-withdrawing effect of the trifluoromethyl group was also not able to
activate peptidic analogue 30 for KLK6 acylation. Altogether, these results suggest that inductive
effects, which result in a better leaving group in E (e.g. 26), can affect the compound potency by
increasing its reactivity toward KLK6. On the other hand, steric hindrance of the ester (e.g. 27),
even when combined with inductively activating effects (e.g. 25) reduces reactivity toward
KLK6.
Complete substitution of the E part of DKFZ-251 (11) with other aromatic systems (e.g. 4-
morpholinophenyl (31) or pyrimidine (32)) resulted in a significant loss of potency (Table 1).
Thus, we started systematically investigating the different features of the highly substituted
pyrazole ring. The methyl groups, which are suggestive of a constrained rotation for the amide
and the N-phenyl bonds, are critical for KLK6 affinity, as shown by bis-demethylated 33.
Compounds 34 and 35 were synthesized to probe this effect, ascribing it mostly to the methyl in
position 5. In fact, 34 was found to be only about 2-fold less active than 11, while 35 was over
50 fold less potent and essentially equipotent to 33. The phenyl ring also contributes to the
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
affinity to KLK6, and its substitution with a smaller methyl group (36) resulted in a 10-fold less
potent inhibitor. Compounds bearing different 5-membered heterocycles such as isoxazole 37
were also found to be less potent, implying that the 5-methyl and the phenyl group in the absence
of the pyrazole are not sufficient for retaining the affinity of 34 to KLK6. Finally, 38 showed that
the presence of an NH in the peptidic bond seems to be required in order to interact with KLK6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(likely to build H bond interactions, typical of the peptidic substrate recognition), and it is
intriguing albeit not fully clear to notice that 38 also has decreased selectivity toward trypsin. In
summary, the E part of 11 plays an important role for inhibitor efficacy, in agreement with a 2-
step inhibition mechanism, but in general the substances are still active if the 2-methyl-3-
substitued indole is maintained in W.
Acyl-enzyme Complex Stability is Sensitive to Structural Changes
Given the inability to measure kinact/K for all the different structures and having no means to
I
directly measure k , we decided to measure k . KLK6 was incubated with an excess of inhibitor
on
off
for 1 h, so that complete labeling of the enzyme was achieved (as observed in the ESI-MS
experiments). Upon excess inhibitor removal (performed with desalting columns), immediate
recovery of KLK6 enzymatic activity was observed for 1 as opposed to DKFZ-251 (11), where a
t1/2 of 6.43 h (95% C.I. 5.93–7.03) was measured (Figure 4). While consistent with the kinact/KI
measurements (see Figure 3), the difference in t1/2 values is surprising for such an apparently
small change in the structure. A potential explanation for this difference is that the indole
N-linkage of 1 tends to decrease the stability of the acyl-enzyme ester intermediate toward water
hydrolysis via inductive electron-withdrawing effects, while the indole C3-linkage in 11 tends to
inductively increase the stability of the acyl-enzyme intermediate due to the relatively high
electron density found at C3 of an indole. Also surprising is the fact that 14, which differs from
ACS Paragon Plus Environment
16

Page 17 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1 only by the lack of an indole C2 methyl group, produced an acyl-enzyme intermediate with a
half-life shorter than 10 min. This suggests that a combined effect of the C2 methyl group (which
presumably increases acyl-enzyme stability through steric hindrance, but also potentially through
electron-donating effects) and the substitution pattern of the indole ring, is responsible for the
stability seen with 11 in comparison to structurally related 1 and 14. For the acyl-enzyme
intermediate formed from 27, KLK6 enzymatic activity was not fully recovered within the
experiment timescale (predicted t1/2 = 19.1 h, 95% C.I. 16.1–23.4). Notably, 33 which has a
much lower potency (higher IC value) than 11, but bears an identical W group, was found to
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
give a similar acyl-enzyme stability profile, consistent with only W influencing this step of the
inhibition mechanism. The lower IC of 33 likely results from decreased affinity to the KLK6
5
0
active site in the first reversible recognition step. Although 27 forms the longest-lived covalent
adduct with KLK6, its higher IC value relative to 11 (likely due to a smaller k ) renders it less
5
0
on
suitable for use as a chemical probe. Altogether DKFZ-251 (11) is the best candidate to probe
KLK6 as it exhibits a balance between rapid on-rate coupled with a long lived acyl-enzyme
complex.
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 4. Acyl-enyzme complex stability determinations for a panel of selected inhibitors. (A)
Aliquots from a solution of acyl-enzyme complex, after excess inhibitor removal, were tested in
triplicate at different time points (10 min, 1 h, 2 h and 3 h for 1 and 14, and 10 min, 1 h, 3 h, 6 h,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9
2
h, 15 h and 24 h for 33, 11 and 27) by addition of FSR-AMC and fluorescence detection at
5 °C. Data were fit to an exponential growth equation, which was used to calculate half-life
values. The values are shown in (B) (± standard deviation).
Docking Studies are Supported by the Synthesis of Amidine Derivatives
Intrigued by the lack of a clearly basic moiety that could act as a P1, we performed a docking
study to analyze the most important interactions within the KLK6 active site. A restricted
docking of DKFZ-251 (11) was performed using a previously solved crystal structure of KLK6
(PDB: 4D8N), and a model of the tetrahedral intermediate formed between Ser195 and 11 was
built (Figure 5A, Figure S9). In this model, the amide bond of DKFZ-251 (11) engages in an
H-bond with L41 in the surroundings of the S1’ pocket, an interaction that is consistent with
3
1
somewhat structurally related compounds co-crystallized with KLK7. In addition, the indole
occupies the S1 pocket and the negatively charged oxygen of the tetrahedral intermediate is
stabilized in the canonical oxyanion hole, where it is coordinated by the backbone N-Hs of S193
and Ser195 in two bridged H-bonds. A previously reported KLK6 crystal structure with a
non-covalent inhibitor (PDB: 3VFE) indicates a lipophilic interaction of the P1 group with I218,
21b
which could also be made by the indole ring of DKFZ-251 (11) (Figure S10). This amino acid
is less conserved in serine proteases and could at least partially account for the selectivity shown
by 11 for KLK6 against other proteases such as trypsin. Superimposing the tetrahedral
intermediate pose with the crystal structure of KLK6 in complex with benzamidine (PDB: 1L2E)
suggested that the introduction of an amidine moiety in the indole ring of 11, specifically at the
ACS Paragon Plus Environment
18

Page 19 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
6
, but not the 5, position, would enable H-bond interactions with Asp189. To test this hypothesis,
we synthesized 39 and 40 (Figure 5B), bearing an amidine in positions 6 and 5, respectively.
As predicted, 39 is highly potent against KLK6, reaching the IC assay limit of 2.5 nM, while
5
0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
0 is essentially equipotent with DKFZ-251 (11). Interestingly, 39 proved to also be a highly
potent inhibitor of trypsin (IC ≤ 1 nM), while 40 showed only a slightly improved potency
5
0
against trypsin relative to 11, implying that the amidine in position 6, but not 5, of the indole can
engage in the Asp189 H-bond network, similarly to canonical basic P1 groups used for
trypsin-like serine proteases inhibitors. The peptidic analog of 39, bis-amide 41 (Figure 5B), was
found to be a reasonably good KLK6 inhibitor (IC = 1.2 ± 0.4 ꢀM), but selectivity over trypsin
50
was lost, thus supporting the notion that polar basic P1 residues, if correctly placed, make strong
interactions with the S1 pocket of both KLK6 and trypsin. These data are consistent with the
specific MOA of the inhibitor class being critical for their selectivity as well.
Figure 5. Docking studies and inhibition data of amidine-containing derivatives. (A) Covalent
docking of KLK6 (light green, PDB: 4D8N) with DKFZ-251 (11) (orange) superimposed with
KLK6 (hidden) co-crystallized with benzamidine (turquoise, PDB: 1L2E). H-bonds given by
each compound are shown (white). (B) Structures and IC values of amidine derivatives 39–41.
5
0
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
KLK Sub-type Selectivity Screening
Aware of the high degree of homology amongst tissue-kallikreins, we tested DKFZ-251 (11)
against a panel of 6 other KLKs (KLK4, 5, 7, 8, 13, and 14 – see Experimental Section for
details on the production and purification of these enzymes). Intrigued by the effect of the
amidino group in position 6 of the indole, 39 was also tested (Table 2). In these assays the
compounds were pre-incubated with the enzymes for 10 min, giving as expected slightly higher
IC s against KLK6 than in Table 1. KLK7 has chymotryptic-like specificity, and recognizes
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
non-basic P1 residues as substrates, but both 11 and 39 are much less active against KLK7 as
compared to KLK6. DKFZ-251 (11) shows KLK6 selectivity against KLK4 (>400 fold), and
KLK8, KLK13, and KLK14 (all three >20 fold). On the other hand, selectivity is partially
retained by 39 against KLK8 (9 fold) and KLK13 (>20 fold), but not in the case of KLK14.
KLK5, which is very similar in structure and active site shape to its cognate KLK6, is the second
favored target of 11 with only about a 2 fold difference in activity relative to KLK6, while 39 is
about 6 fold less active. The selectivity of 11 against secreted serine proteases (Figure S11) and
recombinant KLKs (Figure S12) was also tested in a more complex environment (derived from
conditioned cell culture medium) using broad spectrum activity-based probes to visualize the
proteins by in-gel fluorescence. At concentrations as high as 50 µM, pre-incubation with 11 only
prevents probe binding to KLK6, leaving the other protease levels unchanged.
ACS Paragon Plus Environment
20

Page 21 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. KLK selectivity screen of 11 and 39.
a
IC (ꢀM)
5
0
Enzyme
1
1 (DKFZ-251)
>200
1.1 ± 0.5
0.47 ± 0.14
73
>10
>10
10.3 ± 0.3
39
N.M.
0.130 ± 0.002
0.023
b
KLK4
KLK5
KLK6
KLK7
KLK8
KLK13
KLK14
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
>0.50
0.21
>0.50
0.008 ± 0.001
a
Data represent the mean of at least one experiment performed in triplicates (± standard deviation
when multiple independent experiments were performed) after 10 min pre-incubation of inhibitor
and enzyme. N.M. = not measured.
b
Compound Stability Correlates with Acyl-Enzyme Complex Stability
Compound stability toward ester hydrolysis in cell culture medium at 37 °C was measured as a
means to assess the suitability of the compounds for cellular assays (Figure 6A). DKFZ-251 (11)
demonstrated an increased stability profile in comparison to hit 1 with a half-life almost 8 times
longer (t1/2 = 4.22 h, 95% C.I. 4.01–4.44, and 0.56 h, 95% C.I. 0.54–0.59, respectively).
Interestingly 14, which lacks the 2-methyl group of 11, was rapidly hydrolyzed
(
t1/2 = 0.45 h, 95% C.I. 0.44–0.46), while, as expected, steric shielding with an additional methyl
group (27) stabilized the ester, (t1/2 = 15.2 h, 95% C.I. 13.7–17.1). The chemical stabilities of 1,
11, 14, and 27 closely parallel the stabilities of the corresponding acyl-enzyme complexes which
they produce (see Figure 4), suggesting that the acyl-enzyme complex lifetimes are governed by
the intrinsic chemical properties of the inhibitor structure, and don’t involve non-covalent
interactions with the enzyme. Inductive activation of the ester results in rapid hydrolysis, e.g. 26
(t1/2 = 0.425 h, 95% C.I. 0.420–0.430), and 29, which combines the stabilizing methyl group of
2
7 with the destabilizing trifluoromethyl group of 26, gives an intermediate stability profile
(
t1/2 = 1.68 h, 95% C.I. 1.60–1.73). Interestingly, amidine 39 also shows a less stable profile
(t1/2 = 1.32 h, 95% C.I. 1.16–1.52), which might result from combined increased proteolytic
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
cleavage of the ester bond (due to its lower selectivity profile) in addition to water hydrolysis.
Substance 40, which also contains an amidine group, does show a decrease in stability compared
to 11, but not as dramatic (t1/2 = 2.88 h, 95% C.I. 2.57–3.26), in agreement with its retained
selectivity.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. (A) Stability of a selection of compounds in DMEM at 37 °C. (B) In vitro results of
DKFZ-251 (11) on FaDu cells after a daily treatment for 48 h. Both proliferation (measured via
BrdU incorporation) and phenotypic EMT are affected by 11 in a dose-dependent fashion as
1
7
previously reported for KLK6 shRNA. BrdU incorporation results are the average from 5
independent experiments ± standard deviation.
Cellular Assay Demonstrates On-Target Activity
To assess the activity of the KLK6 inhibitors in cells we used a recently published phenotypic
1
7
assay, which correlates the expression of KLK6 in FaDu cells to proliferation and morphology.
In this assay, a daily treatment of FaDu cells for 48 h with DKFZ-251 (11) (Figure 6B), but not
with inactive carbamate analogue 8 (Figure S13), copied the shRNA-mediated KLK6 silencing
phenotype in a dose-dependent fashion. The same effect was not triggered in other head and neck
tumor cell lines (Cal27, Detroit562), which do not express high basal KLK6 levels and do not
show growth- and morphological-dependence to this protein (Figure S14). These data indicate
that DKFZ-251 (11) is able to effectively inhibit endogenous KLK6 in a cellular context.
ACS Paragon Plus Environment
22

Page 23 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Development of Inhibitor-based ABPs for Research Application
Having validated the MOA, selectivity profile, and cellular activity of the substance class, we
utilized the scaffold of DKFZ-251 (11) to develop activity-based probes (ABPs). Distinguishing
3
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
active from inactive KLKs (and proteases in general) in tissue-derived samples is highly
33
desirable as it could enhance the prognostic value of KLKs as biomarkers. Moreover ABPs are
3
4
powerful tools with wide applicability in basic research and drug discovery. To functionalize
our inhibitors for ABP development, an alkyne was introduced in the W part of the structure via
Sonogashira coupling, and 4 derivatives were synthesized (42–45, Figure 7A). These showed
SAR in accordance with previous results; notably, 43 has excellent selectivity against trypsin.
DKFZ-633 (44), which showed the best labeling in a preliminary assay, was chosen for further
validation. The probes were initially tested in concentrated conditioned medium from the KLK3-
3
5
rich LNCap cell line, to which recombinant KLK6 was added. After incubation with the probe,
and copper-catalyzed alkyne–azide cycloaddition (CuAAC) with a TAMRA-labeled azide, the
samples were visualized “in-gel”. DKFZ-633 (44) selectively labeled KLK6 at all probe
concentrations tested and within a 10 min incubation (Figure S15). Encouragingly, the labeling
was persistent up to 6 h and resistant to the denaturing conditions used. Labeling of KLK6 (10
µM 44, 1 h incubation) was inhibited in a dose-dependent fashion by pre-incubation for 30 min
with 11 but not inactive analog 8 (Figure 7B). Optimized conditions (2.5 µM, 1 h incubation)
were then used to determine the efficiency of KLK6 detection, with 42 and 44 being able to
detect KLK6 at levels as low as 0.05% of the total protein in the sample (Figure S16). Finally, by
utilizing a biotin-labeled azide as an enrichment handle, we were able to pull down endogenous
KLK6 in a dose-dependent fashion in FaDu cell conditioned medium (Figure 7C). Encouraged
by these results we also tested the activity of KLK6 in cell culture medium from a panel of cell
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
lines, and interestingly found that the amount of active protein does not correlate to its relative
3
6
abundance (Figure 7D), for example in SCC9 cells. Interestingly, two glycoforms of KLK6,
which showed cell-dependent relative abundance, could be enriched with DKFZ-633 (44).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Activity-based probe studies. (A) Structure and IC values (30 min pre-incubation) of
50
42–45. (B) In-gel fluorescent labeling of KLK6 (60 ng in 20 µg total protein) by 44 (after
CuAAC with a TAMRA-labeled azide) is prevented by prior incubation with parent inhibitor 11,
but not by inactive carbamate 8, in a dose-dependent fashion. (C) Western blot of endogenous
KLK6 pulled down by 44 (after CuAAC with a biotin-labeled azide) in conditioned FaDu cell
medium shows dose-dependent behavior. (D) Western blot of active KLK6 pulled down from
different cell line conditioned medium. BPD: aliquots from the total mixture before the pull
down; PD: pull down samples; SN: aliquots from the supernatant after the pull down.
Chemical Synthesis
The preparation of most of the substances in this study is depicted in Scheme 1. For the
synthesis of substances with an N-phenyl-3,5-dimethyl-2-aminopyrazole
E
group,
3,5-dimethylpyrazole (46) was converted to highly substituted pyrazole 47 via a three step
ACS Paragon Plus Environment
24

Page 25 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
nitration/N-arylation/reduction sequence. Amide formation with acid 48, or with acid chloride 49
(which gave 10, see Figure 2) followed by acetate removal, gave alcohols 51 and 50,
respectively. The critical ester bond could then be formed with a variety of different substituted
carboxylic acids. For compounds with a 2-methyl-3-acetylindole W group, acid 52 was
condensed with benzyl glycolate (53) to give ester 54. After hydrogenolysis of 54 to give acid
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
5, amide coupling with various primary amines proceeded to give final inhibitors for testing.
Substances 7–9, 24, 30, 33, 38, and 41 required slightly different preparations, which are detailed
in the supporting information.
a
Scheme 1. General synthetic scheme toward depsipeptide KLK6 inhibitors
a
Reagents and conditions: (a) KNO (1.6 equiv), H SO , 100 °C, 18 h, 83%; (b) CuI (5 mol%),
3
2
4
PhI (3.7 equiv), trans-N,N´-dimethylcyclohexane-1,2-diamine (20 mol%), 180 °C (microwave),
h, 96%; (c) H (4 atm), 10% Pd/C, EtOH/CHCl 7:1, r.t., 48 h, quantitative; (d) 48 (1 equiv),
4
2
3
DCC (1.2 equiv), i-Pr Net (1.2 equiv), MeCN, r.t., 24 h, 92%; (e) 49 (1.2 equiv), Et N
2
3
(
1.5 equiv), –10 °C, 5 h, 78%; (f) NaOMe (3 mol%), MeOH, r.t., 3 h, then HCl, 96%; (g)
2
R CO H, DCC or EDCI, i-Pr NEt, DMAP, MeCN or CH Cl or THF, r.t., 26–99%; (h) 53
2
2
2
2
(
1 equiv), DCC (1 equiv), DMAP (0.2 equiv), CH Cl , r.t., 4 h, 88%; (i) H (1 atm), 10% Pd/C,
2 2 2
3
EtOAc, r.t., 48 h, 96%; (j) R NH , DCC, EDCI, or HATU, i-Pr NEt, CH Cl or THF, r.t., 42–
2
2
2
2
5
8%.
The preparation of 39 and 40 is of note, as it required the synthesis of compounds containing a
polar amidine group in the presence of an ester. Amidines are often prepared from the
corresponding cyano derivative by reaction with hydroxyl amine and then reductive cleavage of
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the N-O bond. Both reactions require relatively harsh conditions, which would normally not be
compatible with an ester. Moreover, the intermediates and products in such a reaction sequence
are typically very polar and can be challenging to purify. Our approach to 39 and 40 circumvents
these problems by revealing the polar amidine in the final step, and the synthesis of 39 is
depicted in Scheme 2. Fischer indole synthesis using 3-bromophenylhydrazine (56) and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
-oxopentanoic acid (57) in EtOH gave bromide 58, which was converted to boronic acid 59 via
Miyaura borylation. The fully protected amidine was installed with a modified Liebeskind–Srogl
37
coupling of 59 and pseudothiourea 60. The resulting amidinoindole (61) was saponified to give
acid 62, which was then coupled with alcohol 50 to give ester 63. TFA-mediated deprotection
then provided amidine 39. The synthesis of 40 proceeded analogously.
a
Scheme 2. Synthesis of amidine 39
a
Reagents and conditions: (a) 57 (1 equiv), H SO , EtOH, reflux, 66 h, 36%; (b)
2
4
tetrahydroxydiboron (1.5 equiv), Xphos Pd G2 (1 mol%), XPhos (2 mol%), KOAc (3 equiv),
NaOt-Bu (1 mol%), EtOH, 80 °C, 16 h, 50%; (c) 60 (1 equiv), Pd dba (5 mol%), CuTC
2
3
(3 equiv), TFP (35 mol%), THF, 70 °C, 17 h, 63%; (d) LiOH, dioxane/DMF/water 4:2:1, r.t., 18
h, 92%; (e) 50 (0.9 equiv), EDCI (2 equiv), DMAP (0.2 equiv), CH Cl , r.t., 18 h, 67%; (f) 1:5
2
2
TFA/CH Cl , r.t., 5 h, 12%.
2
2
CONCLUSIONS
Despite continuing reports that KLK6 plays an important role in different pathologies, the
molecular mechanisms by which it affects cancer and neurodegenerative diseases is not yet fully
ACS Paragon Plus Environment
26

Page 27 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
understood. Indeed, the lack of selective and potent inhibitors has slowed down the progression
of such studies, limiting target validation tools to protein silencing and overexpression, with no
pharmacological means to investigate what role KLK6 enzymatic activity plays in disease. To
the best of our knowledge, we report here the first class of inhibitors suitable for investigating
the biology of KLK6. The compounds are potent and selective within the tissue kallikrein family
as well as related proteases, and they inhibit KLK6 via transient quiescent affinity labeling. The
inhibitors must contain an ester bond for reactivity, and this ester bond is susceptible to water
hydrolysis. The intermediate which is formed after reaction of an inhibitor with the enzyme also
undergoes hydrolysis, and the relative stabilities of the parent esters and their resulting acyl-
enzyme intermediates correlate. An appropriate balance between rapid reactivity and long lasting
inhibition is therefore required for an effective inhibitor. Potency is also modulated by
non-covalent interactions that govern the initial affinity for the inhibitor toward the enzyme, and
which have no effect on the acyl-enzyme complex lifetime. On the other hand, incorporation of a
basic P1 residue, correctly placed to interact with Asp189 in the S1 pocket of KLK6, was found
to increase this initial affinity, but was detrimental for selectivity. Our data show that DKFZ-251
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(11) presents the best balance of KLK6 activity, selectivity, and stability, and is an effective
chemical probe for cellular experiments, as demonstrated in a phenotypic assay with FaDu cells.
Moreover, we exploited the inhibition MOA to develop ABPs, which were validated via “in-gel”
fluorescence experiments. The best ABP, DKFZ-633 (44), was used to pull down active
endogenous KLK6 from conditioned medium in a variety of different cell lines. We preliminarily
found that two cell lines, while containing KLK6, had no active enzyme. The probes should be
suitable also for use on patient-derived samples. We envision that this new class of compounds
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
together with the generated ABPs will help advance our understanding of the physiological and,
importantly for translation, pathological role of KLK6.
EXPERIMENTAL SECTION
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Chemical Synthesis
All organic synthesis experiments were performed in solution under air unless otherwise
specified. Reagents and solvents were purchased from Sigma Aldrich, Santa Cruz, VWR,
EnzoLifeSciences and AlfaAesar and used without purification. Anhydrous CH Cl , toluene,
2
2
MeCN and THF were prepared with an MBraun SPS800 Solvent Purification System. Thin layer
chromatography (TLC) was carried out on Merck glass silica plates. TLC visualization was
accomplished using 254 or 366 nm UV light or charring solutions of KMnO , cerric ammonium
4
molybdate, or p-anisaldehyde. Purification was performed by flash column chromatography
using SiliCycle SiliaFlash® P60 (40–63 µm, 60 Å particle size) or via RP-HPLC (Agilent 1260
Infinity and an ES quadrupole Agilent 6120, column: Kinetex® 5µm C18 100 Å, AXIA Packed
LC Column 250 x 21.2 mm; Temperature = 40 ºC; Solvent A = H O, 0.05% TFA; Solvent B =
2
MeCN, 0.05% TFA; Flow Rate = 15.0 mL/min; method: Gradient: 95% A ꢀ 70% A [over 4
min], then 70% A ꢀ 40% A [over 8 min], then 40% A ꢀ 5% A [over 4 min]).
Analytical LC/MS was performed on an Agilent 1260 Infinity system using reverse phase.
Column: Kinetex® 2.6 µm C18 100 Å, LC Column 50 x 2.1 mm; Temperature = 40 ºC; Solvent
A = H O, 0.01% HCO H; Solvent B = MeCN, 0.01% HCO H; Flow Rate = 0.60 mL/min;
2
2
2
method: Gradient: 99% A ꢀ 10% A [over 6 min] then 10% A ꢀ 1% A [over 2 min]). High
resolution mass spectrometry was recorded on a Bruker ApexQe FT-ICR instrument, (Organic
Chemistry Institute, University of Heidelberg). NMR spectra were recorded on Bruker 400 MHz
or 600 MHz instruments at 298.1 K. Melting points were determined on a semi-automated
ACS Paragon Plus Environment
28

Page 29 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
apparatus (MPM-H2, Coesfeld Materialtest). All final compounds were found to have ≥ 95%
purity, controlled by analytical (HPLC/UV/ELSD/MS) with the above described method, and
1
confirmed by H NMR.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Detailed synthesis procedures, characterization data, and NMR spectra can be found in the
Supporting Information.
Protein Production and Purification
Trypsin was purchased from Polymun scientific. Thrombin was purchased from Biopur. Factor
Xa was purchased from Enzo Life Sciences.
2
1b
Recombinant KLK6. Recombinant KLK6 (R74G, R76Q, and N132Q) and KLK6 S195A
3
8
(R74G, R76Q, and N132Q) proteins were produced as previously described with minor
changes. A pFastBac™ donor plasmid including cDNA encoding the sequence for the mature
form of KLK6 and KLK6 S195A (both with additional R74G, R76Q, and N132Q mutations)
preceded at 5´ from the gp64 signal peptide, a 6x His-tag and an enterokinase (EK) recognition
sequence of (Asp) Lys was purchased from Life Technologies™. This system was transformed
4
in competent E.coli DH10MultiBac by electroporation. Transformed cells were grown overnight
at 37 °C in 2x TY medium. 5 µL of 1:10 and 1:100 dilution of these cells were cultured on
selection plates [YTE Agar + Carbenicillin (100 g/ml), Kanamycin (30 µg/ml), Gentamycin
(7 µg/ml), Tetracycline (10 µg/ml), X-gal (400 ꢀg/ml) and IPTG (5 mM)]. White colonies were
re-streaked for monoclonal growth on the same plates overnight at 37 °C. Seven white cultures
thus obtained were grown in a 3 mL culture for 24 h at 37°C in 2x TY supplemented with the
above-mentioned antibiotics. Bacmid DNA containing the KLK6 constructs was isolated using
Qiagen’s miniprep kit following producer instructions and sequenced. DNA concentrations were
determined using NanoDrop spectrophotometer (Thermo Scientific, MA, USA) and the construct
ACS Paragon Plus Environment
29