Discovery of Selective Inhibitors of Endoplasmic Reticulum Aminopeptidase 1

Article abstract of DOI:10.1021/acs.jmedchem.9b00293

ERAP1 is an endoplasmic reticulum-resident zinc aminopeptidase that plays an important role in the immune system by trimming peptides for loading onto major histocompatibility complex proteins. Here, we report discovery of the first inhibitors selective for ERAP1 over its paralogues ERAP2 and IRAP. Compound 1 (N-(N-(2-(1H-indol-3-yl)ethyl)carbamimidoyl)-2,5-difluorobenzenesulfonamide) and compound 2 (1-(1-(4-acetylpiperazine-1-carbonyl)cyclohexyl)-3-(p-tolyl)urea) are competitive inhibitors of ERAP1 aminopeptidase activity. Compound 3 (4-methoxy-3-(N-(2-(piperidin-1-yl)-5-(trifluoromethyl)phenyl)sulfamoyl)benzoic acid) allosterically activates ERAP1's hydrolysis of fluorogenic and chromogenic amino acid substrates but competitively inhibits its activity toward a nonamer peptide representative of physiological substrates. Compounds 2 and 3 inhibit antigen presentation in a cellular assay. Compound 3 displays higher potency for an ERAP1 variant associated with increased risk of autoimmune disease. These inhibitors provide mechanistic insights into the determinants of specificity for ERAP1, ERAP2, and IRAP and offer a new therapeutic approach of specifically inhibiting ERAP1 activity in vivo.

Full text of DOI:10.1021/acs.jmedchem.9b00293

Subscriber access provided by BIU Pharmacie | Faculté de Pharmacie, Université Paris V
Article
Discovery of Selective Inhibitors of Endoplasmic Reticulum Aminopeptidase 1
Zachary Maben, Richa Arya, Digamber Rane, W Frank An, Shailesh Metkar,
Marc Hickey, Samantha Bender, Akbar Ali, Tina T Nguyen, Irini Evnouchidou,
Roger J. Schilling, Efstratios Stratikos, Jennifer E. Golden, and Lawrence J. Stern
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00293 • Publication Date (Web): 16 Dec 2019
Downloaded from pubs.acs.org on December 17, 2019
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 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Selective Inhibitors of Endoplasmic
Reticulum Aminopeptidase 1
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
Zachary Maben,1,2,§ Richa Arya, Digamber Rane, W. Frank An, Shailesh Metkar, Marc
1,§
3,§
4
4
4
4
2
2,†
1,5,†
Hickey, Samantha Bender, Akbar Ali, Tina T. Nguyen, Irini Evnouchidou,
Roger
4
5
3,†
1,2,*
Schilling, Efstratios Stratikos, Jennifer Golden, and Lawrence J. Stern
1
Department of Pathology, University of Massachusetts Medical School, Worcester,
Massachusetts 01655, United States
2
Department of Biochemistry and Molecular Pharmacology, University of
Massachusetts Medical School, Worcester, Massachusetts 01655, United States
3
Kansas University Specialized Chemistry Center, Lawrence, Kansas 66047, United
States
4
Broad Institute of MIT and Harvard, Cambridge, Massachusetts 02142, United States
5
National Centre for Scientific Research Demokritos, Agia Paraskevi, Athens 15341,
Greece
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 80
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
ACS Paragon Plus Environment
2

Page 3 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ABSTRACT:
ERAP1 is an endoplasmic reticulum-resident zinc aminopeptidase that plays an
important role in the immune system by trimming peptides for loading onto major
histocompatibility complex (MHC-I) proteins. Here we report discovery of the first
inhibitors selective for ERAP1 over its paralogs ERAP2 and IRAP. Compound 1 (N-(N-
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-(1H-indol-3-yl)ethyl)carbamimidoyl)-2,5-difluorobenzenesulfonamide)
compound (1-(1-(4-acetylpiperazine-1-carbonyl)cyclohexyl)-3-(p-tolyl)urea) are
competitive inhibitors of ERAP1 aminopeptidase activity. Compound 3 (4-methoxy-3-
N-(2-(piperidin-1-yl)-5-(trifluoromethyl)phenyl)sulfamoyl)benzoic acid) allosterically
and
2
(
activates ERAP1’s hydrolysis of fluorogenic and chromogenic amino acid substrates but
competitively inhibits its activity towards a nonamer peptide representative of
physiological substrates. Compounds 2 and 3 inhibit antigen presentation in a cellular
assay. Compound 3 displays higher potency for an ERAP1 variant associated with
increased risk of autoimmune disease. These inhibitors provide mechanistic insight into
the determinants of specificity for ERAP1, ERAP2 and IRAP, and offer a new
therapeutic approach of specifically inhibiting ERAP1 activity in vivo.
INTRODUCTION
Endoplasmic reticulum aminopeptidase 1 (ERAP1, also known as ARTS1, ALAP,
ERAAP, or PILS-AP) is a 107 kDa M1-family zinc aminopeptidase (EC 3.4.11.-)1
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 80
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
1
ubiquitously expressed in somatic cells. ERAP1 trims peptides in the endoplasmic
reticulum prior to their presentation on class I major histocompatibility complex (MHC-
I) proteins. ERAP1 has length specificity unique among aminopeptidases, efficiently
trimming peptides longer than ~8 residues but sparing shorter peptides,2, 3 matching the
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
length preferences of MHC-I. ERAP1 exhibits broad but defined substrate sequence
specificity with preference for bulky hydrophobic amino terminal residues and either
bulky hydrophobic or basic C-terminal residues.5, 6 ERAP1 is monomeric in solution but
exhibits allosteric regulation, with short peptides able to activate ERAP1 for hydrolysis
of a model fluorogenic amino acid substrate leucine-7-amido-4-methylcoumarin (L-
2
AMC). The regulatory site bound by these activator peptides is not known, but has
been hypothesized to be distal to the catalytic center although within the overall
binding site for full-length peptide substrates, based on alteration of ERAP1 length
dependence in the presence of allosteric activator, and the ability of some allosteric
activators to inhibit full-length peptide hydrolysis. An allosteric model for ERAP1’s
length dependence has been proposed,2, 7, 8 whereby long but not short peptide
substrates are able to access the regulatory site, in effect allosterically activating their
own hydrolysis.
ERAP1 processing can result in generation or degradation of peptides able to bind MHC-I, and
as a result ERAP1 impacts a large fraction of the overall MHC-I-bound peptide repertoire.9-12
After loading, MHC-I molecules traffic to the cell surface where they are assessed by circulating
T lymphocytes as part of immune surveillance of bodily tissues for infection or malignancy.
ACS Paragon Plus Environment
4

Page 5 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Indeed ERAP1 inhibition has been proposed as a potential approach to enhance the
immunogenicity of tumors.12, 13 Polymorphisms in ERAP1, in R528K, are associated with
1
4
15, 16
increased susceptibility to T-cell mediated autoimmune diseases and cancer.
ERAP1 also
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
has been implicated in regulation of blood pressure17, 18 by hydrolyzing angiotensin II,19 with
secretion of ERAP1 from the ER regulated by the thiol reductase ERp44.17
In humans the erap1 gene is found together with paralogs ERAP2 and IRAP in a gene cluster
on chromosome 15, and the corresponding proteins form the oxytocinase subfamily of M1
aminopeptidases. Endoplasmic reticulum aminopeptidase 2 (ERAP2, also known as L-RAP, EC
3
.4.11.-) plays a role in antigen processing similar to ERAP1 but with a different substrate
6
peptide sequence preference. Insulin-regulated aminopeptidase (IRAP, EC 3.4.11.3), also
known as placental leucine aminopeptidase (PLAP), oxytocinase, and leucyl-cystinyl
aminopeptidase, plays an analogous function but in endosomes rather than the endoplasmic
reticulum, processing peptides in the MHC-I cross-presentation pathway.20, 21 IRAP also
performs metabolic, neurological and endocrine functions.22-24 Understanding the specific
contributions of the different oxytocinase subfamily aminopeptidases is of interest to both basic
and applied research. Inhibitors with high selectivity for ERAP1 would allow ERAP1 function to
be probed in a manner distinct from genetic deletion,25 expression modulation via RNAi
3
26
knockdown , and from less selective inhibitors. However, identification of highly selective
inhibitors of ERAP1 has met with limited success despite many efforts. Commonly used
aminopeptidase inhibitors bestatin27 and amastatin28 have poor potency for ERAP129 and
promiscuous inhibition of other aminopeptidases.30, 31 Leucinethiol is somewhat more potent,32
but also inhibits many other peptidases.33 Many ERAP1 inhibitors act by interaction with the
catalytic zinc and surrounding residues, which are well conserved among this enzyme family,34,
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 80
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
35
leading to poor specificity for ERAP1 over other M1 aminopeptidases. One notable example
of an ERAP1-specific inhibitor is the compound thiomersal, which is proposed to bind the active
site zinc and surrounding residues but could also inactivate the enzyme by non-specific
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
36
mechanisms that involve interactions with cysteine residue. Small molecule inhibitors specific
for other M1 aminopeptidases have been identified, including sub-micromolar inhibitors of
IRAP,37 human aminopeptidase N / CD13 , and PfA-M1, an enzyme expressed by P.
falciparum during malaria infection.39, 40 Several potent peptide-based inhibitors of M1
aminopeptidases have been developed rationally using a phosphinate group as a substrate
transition state analog.6, 8, 41-45 However, as M1 zinc aminopeptidases reaction mechanisms and
active site geometry are highly conserved, inhibitors in this class exhibit significant potency for
multiple M1 aminopeptidases,44 although a selective ERAP2 inhibitor in this class was
identified6, 8, 46 based on ERAP2’s unique specificity for basic residues at the P1 site.4, 6 These
compounds also may be limited in their in vivo utility due to their similarity to peptides, which
are hampered by proteolysis and cell membrane impermeability. A fragment-based approach has
been applied to the development of small-molecule inhibitors for ERAP1, although selective
inhibitors were not identified.46, 47 As ERAP1, ERAP2 and IRAP each perform different but
related immune functions, the effect of broad inhibition may have dire consequences. The
clinical implications of either specific or broad inhibition of oxytocinase M1 aminopeptidases are
still unknown but may have a significant impact on treating human diseases.48, 49
38
In this study, we developed a high-throughput screening (HTS) strategy to identify
small-molecule inhibitors selective for ERAP1 over the other oxytocinase family
aminopeptidases ERAP2 and IRAP. Compounds with three different chemical
frameworks were identified for optimization and tested using physiologically relevant
ACS Paragon Plus Environment
6

Page 7 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
substrate processing assays. Compounds 2 and 3 also inhibit ERAP1’s activity in a
cellular antigen processing assay. Molecular docking, mechanism of action studies, and
mutagenesis results indicate that compounds 1 and 2 target ERAP1’s active site,
whereas compound 3 targets an allosteric regulatory site.
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
CHEMISTRY
Compound 1 was synthesized from tryptamine hydrochloride in two steps as
outlined in Scheme 1. Briefly, reaction of tryptamine hydrochloride 4 with 1H-pyrazole-
1-carboximidate hydrochloride 5 in the presence of N,N-diisopropylethylamine (DIEA)
in CH CN provided the corresponding guanidinium intermediate 6. The crude product
3
6
was reacted with 2,5-difluorobenzene sulfonyl chloride in the presence of NaOH in
acetone to afford 1 in 35% yield over two steps.
Scheme 1. Synthesis of Compound 1
Reagents and conditions: (a) DIEA, CH CN, RT, 20 h; (b) NaOH, acetone, RT, 12 h,
5% (over two steps).
3
3
Compound 2 and analogues were prepared from commercially available amino acids
in four steps involving protection of amino group, amide coupling, deprotection, and
urea formation. As shown in Scheme 2, the unnatural amino acid homocycloleucine 7
was treated with di-tert-butyl dicarbonate and NaOH in THF-H O mixture to afford the
2
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 80
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
corresponding Boc-protected derivative 8. The Boc-protected amino acid derivatives 9–
1
0 were prepared using the same conditions. The carboxylic acids 8–10 were coupled
with N-acetylpiperazine, morpholine, or N-phenylpiperazine using HATU and DIEA to
provide amides 11–15. Removal of the Boc protecting group with TFA and reaction of
the resulting amine salts with p-tolyl isocyanate or derivative provided the target urea
compound 2 and analogues 16–23 (Scheme 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
Scheme 2. Synthesis of Compound 2 and Analogues
Reagents and conditions: (a) NaOH, (Boc) O, THF-H O (1:1), RT, 16 h; (b) 4-
2
2
actylpiperazine, morpholine, or 4-phenylpiperazine, HATU, DIEA, CH Cl , RT, 16 h,
2
2
4
5–75% (over two steps); (c) TFA, CH Cl , RT, 1 h; (d) DIEA, p-tolyl isocyanate or
2
2
derivative, CH Cl , RT, 8 h, 60–75% (over two steps).
2
2
The arylsulfonamide 3 and analogues were prepared from the corresponding phenyl-
piperidine and phenyl-piperazine derivatives as outlined in Scheme 3. The required
intermediates were prepared in two steps involving an S Ar reaction of a 5-substituted
N
2
-fluoronitrobenzene derivatives 24–27 with the cyclic amines followed by the
reduction of the nitro group with SnCl .2H O to afford the anilines 32–36. The sulfonyl
2
2
ACS Paragon Plus Environment
8

Page 9 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
chlorides 37–40 were either commercially available or prepared from the corresponding
4
-substituted (3-chlorosulfonyl)benzoic acid derivatives by esterification using thionyl
chloride and MeOH. Finally, reactions of substituted anilines 32–35 with 2,5-substituted
benzenesulfonyl chlorides 37–40 in pyridine followed by ester hydrolysis using
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
LiOH.H O afforded the target compound 3 and analogues 41–46. Treatment of
2
sulfonamide 47 with NaH and MeI in DMF followed by ester hydrolysis provided the
N-methyl derivative 49. The 4-methylpiperazine analogue 50 was synthesized from the
aniline derivative 36 using the above general reaction sequence (Scheme 3).
Scheme 3. Synthesis of Compound 3 and Analogues
Reagents and conditions: (a) Piperidine or N-methylpiperazine, K CO , DMF, 120 °C,
2
3
6 h; (b) SnCl .H O, EtOH, 70 °C, 1 h, 79% (over two steps); (c) SOCl , RT, 16 h, then
2
2
2
MeOH, RT, 2 h, quant.; (d) Pyridine, RT, 16 h; (e) LiOH.H O, THF-H O (1:1), RT, 8 h,
2
2
2
5–35% (over two steps); (f) NaH, MeI, DMF, RT, 18 h, 25%.
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 80
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
RESULTS
MLPCN Library Screen. ERAP1 enzymatic activity can be conveniently assayed
5
0
using the synthetic fluorogenic substrate L-AMC. Since this is shorter than preferred
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
substrates, ERAP1 hydrolysis activity of L-AMC is substantially lower than for more
2
physiologically relevant peptides of 8 or more residues but can be measured easily by
fluorescence in microtiter format HTS assays. In order to identify novel ERAP1-selective
inhibitors, we designed and implemented a HTS of the NIH’s Molecular Libraries Probe
Production Centers Network (MLPCN) >350,000 compound library at the Broad
Institute’s Probe Development Center. We screened for compounds that inhibit ERAP1
L-AMC hydrolysis activity at a single concentration point (10.7 µM) (Figure 1).
Compounds that decreased ERAP1 L-AMC hydrolysis activity by 30% were filtered to
remove promiscuous hit compounds (compounds with greater than 5% hit rate in
PubChem assay records). Additionally, known metal chelators, electrophiles, and
compounds qualifying as PAINS were omitted from the candidate pool. Compounds
containing peptidomimetic motifs were omitted in order to define a pool of hit
compounds with modes of inhibition distinct from previously characterized inhibitors
of M1 family aminopeptidases. Compounds containing coumarinyl moieties were also
omitted to minimize false positive results from autofluorescent compounds, as our
initial activity assay monitored fluorescence of 7-amino-4-methylcoumarin (AMC) as a
reporter of aminopeptidase-catalyzed L-AMC hydrolysis. Remaining hit compounds
ACS Paragon Plus Environment
1
0

Page 11 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
were subsequently tested for dose-dependent efficacy against ERAP1 and candidates
with half-maximal inhibitory concentration (IC ) for L-AMC hydrolysis weaker than 10
5
0
µM were removed from consideration. Potency was quantified by fitting data with
sigmoidal functions anchored to an uninhibited control as curve maximum and a curve
minimum greater than zero activity. Compounds exhibiting dose-dependent inhibition
were then counter-screened for hydrolysis inhibition of IRAP (L-AMC substrate) or
ERAP2 (arginine-7-amido-4-methylcoumarin (R-AMC) substrate), removing hits with
IC lower than 1 mM for either ERAP2 or IRAP. We identified two compounds as
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
selective inhibitors of ERAP1-catalyzed hydrolysis of L-AMC, designated here as 1 and
2
.
ACS Paragon Plus Environment
1
1

Journal of Medicinal Chemistry
Page 12 of 80
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
352,341 compounds
ERAP1 activity test
>30%
inhibition
>40%
activation
1
0.7μM [compound]
543
658
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
omit peptidomimetics,
PAINS, chelators, electrophiles
163
51
ERAP1 EC < 10μM
5
0
42
21
9
omit coumarin
containing structures
10
ERAP2 and IRAP IC > 1mM
5
0
2 compounds
1 compound
1
3
2
Figure 1. Screening pathway. The MLPCN library was screened for alteration of ERAP1
activity. Inhibitors and activators were examined to remove known problematic
scaffolds and moieties (PAINS) as well as peptidomimetic compounds. Remaining
compounds were then tested for potency with a cutoff half-maximal effective
concentration (EC ) of 10 µM. We observed several candidate compounds contained
50
coumarinyl groups and removed these out of concern for false positives. We then
counter-screened the compound pools for activity on ERAP2 and IRAP. Two L-AMC
hydrolysis inhibitors (1 and 2) and one L-AMC hydrolysis activator (3) were identified.
ACS Paragon Plus Environment
1
2

Page 13 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Based on previous work demonstrating allosteric activation of ERAP1,2, 7 we also
screened for compounds in the MLPCN library that increased ERAP1-catalyzed L-AMC
hydrolysis. Previous studies have shown that certain short peptides and non-
hydrolyzable substrates can activate ERAP1 L-AMC hydrolysis but inhibit processing
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
of full-length peptides, which are more physiologically relevant ERAP1 substrates. The
mechanism proposed for such behavior was binding to an allosteric regulatory site that
promoted a domain closure required for active site reorganization and L-AMC
hydrolysis; in the hypothesized mechanism this site overlapped with the binding site
2
for full-length peptide substrates. Thus occupancy of the regulatory site would
competitively inhibit processing of full-length peptide substrates (see Supplemental
Figure S1). We hypothesized that compounds that activate ERAP1-catalyzed L-AMC
hydrolysis might similarly act as inhibitors for full-length peptide processing by
occupying a regulatory site within the overall substrate binding envelope. We initially
selected compounds that increased L-AMC hydrolysis by 40% (3 standard deviations)
over untreated control for further characterization (Figure 1, right side). Candidate
activator compound screens proceeded as with inhibitors, with manual identification of
metal chelators, electrophiles, PAINS compounds, peptidomimetics, and promiscuous
positive hit compounds (using the same 5% positive PubChem hit rate cutoff as before).
Activator dose-dependent potency was then assayed for half-maximal activating
concentration (AC ) better than 10 µM. As compounds exhibited different levels of
50
ACS Paragon Plus Environment
1
3

Journal of Medicinal Chemistry
Page 14 of 80
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
maximum activation, dose-dependent hydrolysis data were fit using sigmoidal
functions anchored to an uninhibited control with no restraint on curve maximum.
Compounds with sufficient activating potency were then counter-screened for ERAP2
or IRAP activity (either activation or inhibition) as before. We identified one ERAP1-
selective activator, compound 3. Compounds 1, 2, and 3 and analogues were
synthesized for confirmatory testing, structure-activity relationship analysis,
mechanism of action, and functional testing.
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
Dose-dependence and Specificity of Hit Compounds. We tested the synthesized
compounds 1, 2, and 3 in a dose-dependent L-AMC hydrolysis assay (Figure 2A).
Compounds 1 and 2 exhibited ERAP1 L-AMC IC of 9.2 µM and 5.7 µM, respectively.
50
We confirmed the specificity of these compounds observed in the MLPCN screen for
ERAP1 over homologs ERAP2 and IRAP. All three compounds are more than 100-fold
more selective for ERAP1 than for ERAP2 (Figure 2B) or IRAP (Figure 2C), and in some
cases no activity at all on ERAP2 and IRAP was detected. Compound 3 activated
ERAP1 hydrolysis of L-AMC (Figure 2A, lower panel), with an AC of 3.7 µM and a
5
0
maximum activity enhancement of 4.1-fold over control. This compound is a very weak
inhibitor of ERAP2 with an IC greater than 200 µM and has no detectable effect on
5
0
IRAP activity (Figure 2B, C, lower panels).
ACS Paragon Plus Environment
1
4

Page 15 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
ERAP1
B
ERAP2
C
IRAP
2
1
1
00
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
0
5
00
3
00
3
100
-8 -7 -6 -5 -4 -3 -8 -7 -6 -5 -4 -3 -8 -7 -6 -5 -4 -3
Log [Compound] (M)
Figure 2. Compounds 1, 2, and 3 are highly specific for ERAP1. Enzyme activity was
measured using dipeptide substrate analog X-7-amido-4-methylcoumarin (X-AMC),
where the amino-terminal residue X is leucine (ERAP1, IRAP) or arginine (ERAP2),
2
1
corresponding to the substrate preference of each enzyme. Hydrolysis was measured
over a range of compound concentrations for (A) ERAP1, (B) ERAP2, or (C) IRAP.
Activity was normalized to the activity of the enzyme in the presence of DMSO.
Representative data (mean ± SD, n = 3) from one of two experiments are shown. Data
points were fit using a sigmoidal curve with the top or bottom value for inhibitor or
activator curves, respectively, constrained to 100%. ED values shown in Supplemental
5
0
Table S1.
ACS Paragon Plus Environment
1
5

Journal of Medicinal Chemistry
Page 16 of 80
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 tested the effect of these inhibitors on ERAP1-catalyzed hydrolysis of a decamer
51
peptide WRCYEKMALK (WK10), which represents a more physiologically-relevant
ERAP1 substrate than L-AMC. ERAP1 efficiently hydrolyzed the N-terminal
tryptophan and released the nonamer product peptide RCYEKMALK (Figure 3A).
Removal of the subsequent amino acid arginine was inefficient, consistent with
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
1
ERAP1’s amino-terminal substrate specificity. Here we quantified product peptide
production by LC-MS rather than fluorescence, which does not require introduction of
the non-natural fluorescence-quenching dinitrophenyl maleimide as in the original
assay.51, 52 Both compounds 1 and 2 inhibited peptide hydrolysis in a dose-dependent
manner (Figure 3B). Compound 3, identified as an activator of ERAP1 L-AMC
hydrolysis, also inhibited peptide hydrolysis in a dose-dependent manner (Figure 3B),
2
as previously observed for short peptides that activate L-AMC hydrolysis. The IC
5
0
values for WK10 peptide hydrolysis showed the same rank order and relative potency
as the respective IC values for L-AMC hydrolysis (Supplemental Table S1).
5
0
WRCYEKMALK: 1327.66 Da
RCYEKMALK: 1141.41 Da
B
A
1
2
3
1327.972
100
100
50
1141.924
0
7
0
-8
-7
-6
-5
-4
00
900
1100
m/z
1300
1500
Log [Compound] (M)
ACS Paragon Plus Environment
1
6

Page 17 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 3. Compounds 1, 2, and 3 inhibit ERAP1-catalyzed peptide hydrolysis. (A)
ERAP1 hydrolysis of the N-terminal tryptophan from the peptide WRCYEKMALK
(WK10). MALDI spectrum after 30 minute incubation with ERAP1. Calculated masses
are shown at top and observed peak m/z are displayed on the graph. (B) Compound
dose-response of ERAP1-catalyzed peptide hydrolysis inhibition. Representative data (n
= 2) normalized to DMSO control condition shows inhibition of peptide hydrolysis. IC50
values shown in Supplemental Table S1.
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
Structure-Activity Relationship Studies. Of the compounds identified as inhibitors
or activators of ERAP1 hydrolysis of L-AMC, compound 2 and 3 were selected for
structure-activity relationship (SAR) analysis. For compound 2 modification of the N-
acetylpiperazine and homocycloleucine moieties of compound 2 generally resulted in
loss of potency of one log or more; however, substitutions on the aromatic ring were
better tolerated (Table 1, Supplemental Figure S2). Compound 16, wherein the p-tolyl
group is replaced with a 4-chlorylphenyl group, showed a small improvement in
potency in both L-AMC and WK-10 peptide hydrolysis assays. The N-acetyl group on
piperazine was critical was activity, as analogues with morpholine (compound 20) and
N-phenylpiperazine (compound 21) had no detectable inhibitory activity (Table 1).
Table 1. IC Values of Compound 2 and Analogues
5
0
ACS Paragon Plus Environment
1
7

Journal of Medicinal Chemistry
Page 18 of 80
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
ERAP1
WK10
peptide
IC50 (µM)
ERAP1
L-AMC
IC50 (µM)
ERAP2
R-AMC
IC50 (µM)
IRAP1
L-AMC
IC50 (µM)
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
3
Compound
X
R1
R /R
6.9a
2
N-Ac
N-Ac
N-Ac
N-Ac
N-Ac
O
Me
Cl
-(CH
-(CH
-(CH
-(CH
-(CH
-(CH
-(CH
2
2
2
2
2
2
2
)
5
)
5
)
5
)
5
)
5
)
5
)
5
-
-
-
-
-
-
-
6.9
3.6
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
>200
1
6
4.3
31.8
17
OMe
F
115a
>200
20.6a
>200
>200
126
1
8
~165a
~135a
>200
~124
150
19
H
2
2
0
1
Me
Me
Me
Me
N-Ph
N-Ac
N-Ac
22
Me/Me
-(CH
2
3
2
)
2
-
>200
170a
a
Partial inhibition, with less than 90% inhibition observed at the highest concentration
tested (see Supplemental Figure S2).
SAR exploration of compound 3 provided several analogues with similar or slightly
improved potency (Table 2, Supplemental Figure S3). Substitutions of the
1
2
trifluoromethyl (R ) and methoxy (R ) groups on the two aromatic rings were largely
tolerated but resulted in lower potency. Compound 41, where the methoxy group was
replaced with a chloro group showed a 5-fold loss in potency, whereas compound 42
with a bromo substitution was only 2-fold less active than 3. Replacement of the
ACS Paragon Plus Environment
1
8

Page 19 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
trifluoromethyl group with a methyl (compound 44) or fluoro group (compound 45)
resulted in about 8-fold loss in potency, though the unsubstituted analogue (compound
46) maintained better potency. The aromatic carboxylic acid appeared to be essential for
activation of ERAP1’s L-AMC hydrolysis activity, as the methyl ester 47 had no
detectable activity in this assay, although surprisingly WK10 peptide hydrolysis
inhibition was largely intact. This was the only compound for which such discordant
activity was observed. Methylation of the sulfonamide nitrogen gave analogue 49,
which was also about 10-fold less active. Compound 50, with N-methylpiperazine
moiety instead of piperidine, had equivalent activity to the parent compound 3 (Table 2,
Supplemental Figure S3).
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
Table 2. IC Values of Compound 3 and Analogues
5
0
ERAP1
WK10
peptide
ERAP1
L-AMC
ERAP2
R-AMC
IRAP1
L-AMC
R1
R2
3
4
Compound
X
R , R
AC50 (µM)
(
AC50 (µM)
IC50 (µM)
IC50 (µM)
>200
IC50 (µM)
>200
IC50 (µM)
5.3
fold)a
(fold)a
3
CH
CH
CH
2
2
2
CF
CF
CF
3
OMe
Cl
H, H
H, H
H, H
4.1 (4.1)
21.2 (3.7)
10.9 (3.3)
4
1
3
3
>200
>200
15.2
42
Br
>200
>200
8.2
ACS Paragon Plus Environment
1
9

Journal of Medicinal Chemistry
Page 20 of 80
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
43
CH
CH
CH
CH
CH
CH
2
2
2
2
2
2
CF
3
H
H, H
35.9 (3.5)
32.9 (1.5)
29.1 (1.4)
11.6 (1.4)
~140
NDb
159
>200
>200
>200
>200
>200
~160
>200
20.2
7.4
4
4
4
5
Me
F
OMe
OMe
OMe
OMe
OMe
OMe
H, H
H, H
8.6
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
46
H
H, H
81.2
14.4
6.9
4
7
CF
CF
CF
3
Me, H
H, Me
H, H
>200
~42 (1.2)
49
3
50.3 (2.7)
2.9 (3.1)
~160
ND
32.6
3.0
5
0
N-Me
3
aFold increase in hydrolysis activity.
bND: not determined.
Mechanism of Action Studies. To help understand the mechanism of action of these
compounds, we evaluated ERAP1 enzyme activity at various concentrations of
substrate and inhibitors. We used the chromogenic substrate leucine p-nitroanilide (L-
pNA) instead of fluorogenic L-AMC, so that a wider range of substrate concentrations
could be investigated. We measured initial rates under steady state conditions and
analyzed the data using a Michaelis-Menten framework with a cooperativity term (Hill
coefficient) to account for the allosteric behavior of ERAP1 towards short substrates (see
2
Experimental Section). L-pNA hydrolysis in the absence of inhibitor had the following
-
1
parameters: h = 1.45  0.15, Khalf = 1.41  0.16 mM, and Vmax = 17.25  1.05 sec , where h
is the Hill coefficient, Khalf is the concentration of substrate at which half-maximal rate is
observed and is the allosteric equivalent of the conventional Michaelis constant K , and
m
ACS Paragon Plus Environment
2
0

Page 21 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Vmax is the maximum reaction rate (Supplemental Table S2). For compound 1, increasing
inhibitor concentrations resulted in decreased L-pNA hydrolysis activity (Figure 4A).
The apparent Michaelis constant Khalf increased linearly with increasing inhibitor
concentration, without apparent change in Vmax, consistent with a competitive inhibition
mechanism. The full set of curves were described well by an equation describing
competitive inhibition in an allosteric setting (see Experimental Section, competitive
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
allosteric inhibition) K = 51.7  4.3 μM (Supplemental Table S1). For compound 2
i
(Figure 4B), similar behavior was observed, with fit values for Khalf increasing linearly
with inhibitor concentration, fit values for Vmax not significantly changing. As with
compound 1, this behavior is consistent with a competitive mode of inhibition. Best fit
value for K was 12.8  1.0 μM (Supplemental Table S1).
i
ACS Paragon Plus Environment
2
1

Journal of Medicinal Chemistry
Page 22 of 80
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. Mechanism of action studies for (A) compound 1 and (B) compound 2. Top,
initial rates of L-pNA hydrolysis by ERAP1 in the presence of various concentration of
inhibitor. Initial rate versus substrate concentration curves obtained at various inhibitor
concentrations were fit to an allosteric Michaelis-Menten equation to obtain values for
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
Vmax and for Khalf, an allosteric equivalent to K (See Experimental Section for details).
m
Bottom, the resulting values for Khalf and Vmax with uncertainly estimates are plotted as a
function of inhibitor concentration to help determine the inhibitor mechanism of action.
For compound 3, increasing concentrations resulted in increased rather than
decreased L-pNA hydrolysis activity, across a range of substrate concentrations (Figure
5
A). This behavior indicates that compound 3 is acting as a non-essential activator of L-
pNA hydrolysis, with activity (Vmax/Khalf) increasing up to 1.8-fold over the
concentration range tested. However, different behavior was observed for compound 3
5
when a full-length peptide (LF9, LVAFKARKF) was used as a substrate instead of L-
pNA (Figure 5B). Here we used a coupled-enzyme assay to follow removal of the N-
terminal leucine (see Experimental Section). ERAP1 hydrolysis of full-length peptides is
not cooperative.2, 53 and initial rate curves in the absence of inhibitor were fit with K =
m
0
.024 ± 0.006 mM. K values increased linearly with increasing inhibitor concentration
m
while Vmax remained essentially unchanged (Figure 5B). These data were described best
by an equation defining conventional competitive inhibition (see Experimental Section,
competitive inhibition) with K = 3.8 ± 0.7 µM (Supplemental Table S1).
i
ACS Paragon Plus Environment
2
2

Page 23 of 80
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 5. Mechanism of action studies for compound 3. Initial rates of (A) L-pNA or (B)
LF9 peptide hydrolysis by ERAP1 in the presence of various concentrations of
compound 3. Top, initial rate versus substrate concentration curves obtained at various
inhibitor concentrations were fit to an allosteric Michaelis-Menten equation for L-pNA
hydrolysis or a conventional Michaelis-Menten equation for LF9 peptide hydrolysis.
Bottom, the resultant values for Vmax/Khalf (for L-pNA) or for K and V
(for LF9
m
max
peptide) with uncertainty estimates are plotted as a function of compound 3
concentration to help determine the mechanism of action.
Molecular Docking and Validation. To help rationalize the different behaviors
observed for compounds 1, 2, and 3, and to identify possible binding sites, we used an
in silico docking approach. Compound docking was performed using the Glide module
ACS Paragon Plus Environment
2
3

Journal of Medicinal Chemistry
Page 24 of 80
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
from the Schrödinger software package, and binding site solutions were further
analyzed by running Embrace minimization on the docked structure. ERAP1 has been
proposed to cycle between open and closed conformations during the catalytic cycle,2, 53,
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
4
and so we used as targets for the in silico docking procedure coordinates from crystal
structures of ERAP1 in both open (PDB 6MGQ) and closed (PDB 2YD0) conformations.
For compound 1, the top docking solution for either open or closed conformations
was at the active site, partially overlapping with substrates modeled into these
structures based on bound inhibitors (bestatin for 2YD0, phosphinic peptide for
6MGQ). In the top predicted binding pose for the open conformation docking solution
(Figure 6A, B), the inhibitor’s sulfonamide group engaged the catalytic zinc, similarly to
the binding mode observed or predicted for other sulfonamide-based M1-family
peptidase inhibitors.55,
56
In the top closed conformation docking solution, the
sulfonamide was displaced from the zinc center by approximately 4 Å, and the
positions of the indole and difluorobenzene moieties were flipped (Supplemental
Figure S4). To help distinguish these possibilities, we compared the active sites of
ERAP1 with ERAP2 (PDB 6EA4, closed conformation) and IRAP (PDB 4Z7I and PDB
5
MJ6, partially open and closed conformations, respectively) because these proteins
were >1000-fold less sensitive to inhibition by compound 1. We identified residues in
contact with compound 1 for each of the predicted binding poses and replaced these
either with homologous ERAP2 or IRAP residues or with alanine. Thus we prepared
ACS Paragon Plus Environment
2
4

Page 25 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ERAP1 proteins with a substitution at position T350, predicted to contact compound 1
indole ring if the open-conformation docking mode were correct, at position S869,
predicted to contact the indole ring if the closed-conformation docking mode were
correct, or at position S316, predicted to interact with one of the fluoro groups on the
phenyl ring in the closed but not open docking mode. ERAP1 mutants T350A, S869Y,
and S316A retained L-pNA hydrolysis activity, although S869Y activity was reduced 5-
fold (Supplemental Figure S5, Supplemental Table S2). We measured sensitivity of these
mutant ERAP1 proteins for compound 1, using a L-AMC inhibition assay (Figure 6C,
Supplemental Figure S6, Supplemental Table S3). A small reduction in potency was
observed for T350A. This is consistent with the docking mode identified for the open
conformer shown in Figure 6A, as well as the observed competitive mode of action, but
given the uncertainties in docking and the lack of robust effects observed with
mutagenesis the predicted binding site should be considered provisional.
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
2
5

Journal of Medicinal Chemistry
Page 26 of 80
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
Compound 1
A
B
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
Zn
T350
C
-
4.5
-
4.0
Large
effect
Small
effect
No
effect
-
3.5
Figure 6. In silico docking sites and experimental validation for compound 1. (A)
5
7
Environment around preferred docking pose as shown by LigPlot+. Alternate docking
pose (in PDB 2YD0) is shown in Supplemental Figure S4. Boxes indicate positions of
residues mutated to corresponding ERAP2 or IRAP residues in order to validate
docking pose. (B) Docking pose of compound 1 in surface structure of open ERAP1,
with domain I shown in light blue, domain II in green, and mutation position in red. (C)
L-AMC IC values for mutant proteins (Supplemental Table S3). PDB 6MGQ was used
5
0
for docking.
For compound 2, the top docking solutions to open and closed ERAP1 conformers
again were found at the active site, overlapping with the expected substrate binding
ACS Paragon Plus Environment
2
6

Page 27 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
site. Two very similar potential binding poses were identified, with the terminal N-
acetylpiperazine carbonyl group coordinating the catalytic zinc in the open conformer
docking solution (Figure 7A, B) and the central urea oxygen coordinating zinc in the
closed conformer solution (Supplemental Figure S7). Again we identified residues from
ERAP1 that contacted the inhibitor bound in either of the two predicted binding poses,
and prepared mutant proteins with residues from either ERAP2 or IRAP that would be
predicted to impact inhibitor binding: K380T predicted to disrupt contacts with
compound 2 in the open conformer binding solution, and S316A, H873M, and S869Y
predicted to disrupt contacts in the closed conformer binding solution. We also mutated
residue T350 to alanine; in ERAP1 this residue contacts the compound 2 cyclohexyl
group in either of the two predicted binding poses. All the mutants retained 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
although K and/or V
values were altered 10-fold for K380T (Supplemental Figure
m
max
S5, Supplemental Table S2). L-AMC inhibition studies with the mutant proteins
revealed a large effect of K380T, with 80-fold increase in IC value, and a smaller but
5
0
still substantial 10-fold increase in IC for T350A (Figure 7C, Supplemental Figure S6,
5
0
Supplemental Table S3). None of the other mutants appreciably affected the compound
potency. These results are consistent with the docking pose identified using the open
2
conformer of ERAP1 (Figure 7A), but not the pose identified using the closed conformer
(Supplemental Figure S7). The preferred docking pose has the ERAP1 zinc atom
5
8
coordinated by an N-acetyl carbonyl group, substitution of which in compounds 20
ACS Paragon Plus Environment
2
7

Journal of Medicinal Chemistry
Page 28 of 80
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
and 21 substantially reduced potency, and it is possible that replacement of the N-
acetylpiperazine with 4-carboxypiperidine or the corresponding carbamate would
increase potency, similarly to a strategy used to optimize aminopeptidase N
inhibitors.59
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
Compound 2
A
B
K380
T350
Zn
C
-5.0
-4.0
-3.0
Large
effect
Small
effect
No
effect
Figure 7. In silico docking sites and experimental validation for compound 2. (A)
Environment around preferred docking site shown as LigPlot+.57 Alternate docking
pose is (in PDB 2YD0) shown in Supplemental Figure S7. Boxes indicate positions of
residues mutated to corresponding ERAP2 or IRAP residues to validate docking pose.
(B) Docking pose of compound 2 in surface structure of open ERAP1, with domain I
shown in light blue, domain II in green, domain IV in pink, and mutation positions in
ACS Paragon Plus Environment
2
8

Page 29 of 80
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
red. (C) L-AMC IC values for mutant proteins (Supplemental Table S3). PDB 6MGQ
5
0
was used for docking.
For compound 3, the top in silico docking solutions outside the active site were found
at the junction of domains II and IV. (We did not consider docking solutions with any
atom within 6Å of the catalytic zinc, because compound 3 acts to activate L-AMC and
L-pNA hydrolysis, implying that it binds at a site distinct from these substrates). One
potential binding site (Figure 8A, B) was found similarly using both open and closed
ERAP1 conformers as docking targets, another potential binding site with similar
docking energy (Supplemental Figure S8) was found only using the closed conformer.
For each of the predicted binding sites we identified ERAP1 residues interacting with
compound 3 that would be disrupted by substitution of homologous residues from
ERAP2 or IRAP. By this analysis, ERAP1 mutations K551A and T914Y are predicted to
disrupt electrostatic and hydrogen-bonding interactions with compound 3 bound in the
shared predicted docking site, and S340Y and S342Y are predicted to disrupt similar
contacts with compound 3 bound in the docking site predicted only in the closed
conformation. Recombinant proteins carrying these mutations individually had
enzymatic parameters similar to those of wild-type ERAP1 (Supplemental Figure S5,
Supplemental Table S2). A substantial effect was observed only for the K551A mutation
(Figure 8C, Supplemental Figure S6, Supplemental Table S3), favoring the shared site
shown in Figure 8A.
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
2
9