Development of Kinase-Selective, Harmine-Based DYRK1A Inhibitors that Induce Pancreatic Human β-Cell Proliferation

Article abstract of DOI:10.1021/acs.jmedchem.8b00658

DYRK1A has been implicated as an important drug target in various therapeutic areas, including neurological disorders and oncology. DYRK1A has more recently been shown to be involved in pathways regulating human β-cell proliferation, thus making it a potential therapeutic target for both Type 1 and Type 2 diabetes. Our group, using a high-throughput phenotypic screen, identified harmine that is able to induce β-cell proliferation both in vitro and in vivo. Since harmine has suboptimal kinase selectivity, we sought to expand structure-activity relationships for harmine's DYRK1A activity, to enhance selectivity, while retaining human β-cell proliferation capability. We carried out the optimization of the 1-position of harmine and synthesized 15 harmine analogues. Six compounds showed excellent DYRK1A inhibition with IC₅₀ in the range of 49.5-264 nM. Two compounds, 2-2 and 2-8, exhibited excellent human β-cell proliferation at doses of 3-30 μM, and compound 2-2 showed improved kinase selectivity as compared to harmine.

Full text of DOI:10.1021/acs.jmedchem.8b00658

Subscriber access provided by READING UNIV
Article
Development of Kinase-Selective, Harmine-Based DYRK1A
Inhibitors that Induce Pancreatic Human #-Cell Proliferation
Kunal Kumar, Peng Wang, Roberto Sanchez, Ethan Swartz, Andrew F. Stewart, and Robert J DeVita
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00658 • Publication Date (Web): 30 Jul 2018
Downloaded from http://pubs.acs.org on July 30, 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 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Development of KinaseꢀSelective, HarmineꢀBased
DYRK1A Inhibitors that Induce Pancreatic Human
βꢀCell Proliferation.
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
1,2
3
1,2
3
3
Kunal Kumar , Peng Wang , Roberto Sanchez , Ethan A Swartz , Andrew F. Stewart and Robert J.
1,2
DeVita *
1
Drug Discovery Institute, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA
2
Department of Pharmacological Sciences, Icahn School of Medicine at Mount Sinai, New York, NY
1
0029, USA
3
Diabetes, Obesity, and Metabolism Institute, Icahn School of Medicine at Mount Sinai, New York, NY
0029, USA
1
*
Corresponding author: Robert J. DeVita. Eꢀmail: robert.devita@mssm.edu. Phone: 212ꢀ659ꢀ5542;
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 51
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
KEYWORDS: Dualꢀspecificity TyrosineꢀRegulated Kinases (DYRKs), Harmine, DYRK1A
inhibitor, Structure Activity Relationship Study, βꢀcell proliferation, diabetes.
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
DYRK1A has been implicated as an important drug target in various therapeutic areas, inluding
neurological disorders and oncology. DYRK1A has more recently been shown to be involved in
pathways regulating human βꢀcell proliferation, thus making it a potential therapeutic target for
both Type 1 and Type 2 diabetes. Our group, using a highꢀthroughput phenotypic screen
identified harmine that is able to induce beta cell proliferation both in vitro and in vivo. Since
harmine has suboptimal kinase selectivity we sought to expand structureꢀactivity relationships
for harmine’s DYRK1A activity, to enhance selectivity, while retaining human βꢀcell
proliferation capability. We carried out the optimization of the 1ꢀposition of harmine and
synthesized 15 harmine analogs. 6 compounds showed excellent DYRK1A inhibition with IC50
in the range of 49.5ꢀ264 nM. Two compounds, 2ꢀ2 and 2ꢀ8, exhibited excellent human βꢀcell
proliferation at doses of 3ꢀ30 µM and compound 2ꢀ2 showed improved kinase selectivity as
compared to harmine.
ACS Paragon Plus Environment

Page 3 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION:
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
The Dualꢀspecificity TyrosineꢀRegulated Kinases (DYRKs) belong to the CMCG family
of eukaryotic protein kinases. The DYRK family consists of five subtypes including 1A, 1B, 2, 3
and 4. Among them, DYRK1A is the most extensively studied subtype. It is ubiquitously
1
expressed and has been shown to play an important roles in brain development and function,
2ꢀ3
4ꢀ5
neurodegenerative diseases, tumorigenesis and apoptosis and other cellular functions.. More
recently, DYRK1A has been shown to be involved in molecular pathways relevant to
proliferation of human insulinꢀproducing pancreatic βꢀcells, making it a potential molecular
6ꢀ9
target for βꢀcell regeneration as a therapy to treat Type 1 and Type 2 diabetes. DYRK1A
inhibition has been proposed to drive pancreatic βꢀcell proliferation in part by inducing
translocation of the nuclear factor of activated T cells (NFAT) family of transcription factors into
the nucleus, allowing access to promoters of genes that subsequently activate human βꢀcell
6
, 8
proliferation.
Because of the prominent role of DYRK1A in neurodegenerative disease and cancer,
1
, 3ꢀ7, 9ꢀ10
numerous studies have attempted to identify DYRK1A inhibitor scaffolds.
Several
DYRK1A inhibitors, like harmine from natural sources, as well as from small molecule drug
discovery programs, have been identified and characterized including, epigallocatechin (EGCG)
1
1ꢀ12
13ꢀ14
15
16
and other flavanꢀ3ꢀols,
leucettines,
quinalizarine, peltogynoids Acanilol A and B,
1
7
benzocoumarins (dNBC),
indolocarbazoles (Staurosporine, rebeccamycin and their
1
8
19
20
21
22ꢀ23
24
analogues), INDY, DANDY, and FINDY, pyrazolidineꢀdiones,
aminoꢀquinazolines,
25ꢀ27
28
29
meriolins,
pyridine and pyrazines,
chromenoidoles,
11Hꢀindolo[3,2ꢀc]quinolineꢀ6ꢀ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 51
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
30
31ꢀ32
33
9,
carboxylic acids, thiazolo[5,4ꢀf]quinazolines (EHT 5372),
CCꢀ401 and 5ꢀiodotubercidin
Among all the DYRK1A inhibitors, harmine and its analogues (βꢀcarbolines) are the most
commonly studied and remain the most potent and orally bioavailable class of inhibitors known
3
4
.
3
, 10
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
to date.
Historically, harmine is also known to exhibit hallucinogenic properties acting as
35ꢀ36
CNS stimulant, due to its affinity for the serotonin, tryptamine and other receptors.
Studies
have revealed multiple CNSꢀbased effects of harmine, including excitation, tremors, convulsion,
3
7
and anxiety.
Also, harmine and several related analogues have been found to inhibit
DYRK1Aꢀmediated phosphorylation of tau protein, thought to be relevant to Alzheimers’s
3
8
Disease (AD) and Down’s Syndrome (DS, trisomy 21). Harmine has also attracted serious
3
9ꢀ43
interest for cancer therapy.
Using systematic structure modifications, several harmine
analogues have been identified to show potent antiꢀproliferative tumor activity, in vitro and in
4
4ꢀ45
vivo through multiple mechanisms of action, including inhibition of topoisomerase I,
46
47
48
inhibition of CDKs, induction of cell apoptosis, and DNA intercalation .
Recently, using a luciferase reporter phenotypic highꢀthroughput smallꢀmolecule screen
HTS), our group identified harmine as a new class of compounds able to induce human βꢀcell
(
proliferation, as measured by immunolabeling of insulin producing βꢀcells, in a range relevant
6
for therapeutic human βꢀcell expansion. DYRK1A is the most likely target of harmine for βꢀcell
proliferation, as demonstrated by genetic silencing, overexpression and other studies, likely
working through the Nuclear Factors of Activated Tꢀcells (NFAT) family of transcription factors
that activate cell cycle machinery to cause proliferation while maintaining differentiation in the
6
human βꢀcell. Using three different mouse, rat and human islet–implant models, it was also
shown that harmine is able to induce βꢀcell proliferation, increase islet mass and improve
ACS Paragon Plus Environment

Page 5 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
6
glycemic control. Other labs have confirmed these results with other DYRK1A inhibitors
7
ꢀ9
unrelated to the harmine scaffold.
DYRK1A and NFATs are widely expressed outside βꢀcells, and harmine analogs are
known to have offꢀtarget effects, leading to pharmacological side effects thereby limiting the
therapeutic utility and potential for pharmaceutical development of harmine analogues for a
chronic diseases like diabetes. Thus, there is an urgent need to develop strategies to identify
selective harmine analogs specifically designed to induce human βꢀcell proliferation as
pharmacological probes for therapeutic development, with limited offꢀtarget activities. In
addition, optimized selective harmine analogs may also be useful for specific, targeted delivery
to the βꢀcell. One such prototype technology from DiMarchi and colleagues was recently
developed to conjugate estrogen derivatives to GLPꢀ1 analog peptides to effectively target cells
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
9
that express GLP1ꢀreceptor.
With knowledge of this previous work, we explored the binding pockets of known
DYRK1A inhibitorꢀcoꢀcrystal structures and identified areas for expansion of the harmine
scaffold that will make new contacts to DYRK1A to identify kinase selective as well as linkable
harmine analogs for targeted therapy. We carried out the optimization studies of harmine to
expand the structureꢀactivity relationships for harmine and to correlate DYRK1A activity, kinase
selectivity and human βꢀcell proliferation. Herein, we report the optimization of harmine 1ꢀ
position and identification of harmineꢀbased DYRK1A inhibitors that exhibit excellent human βꢀ
cell proliferation with improved kinase selectivity as compared to the hit molecule harmine.
RESULTS AND DISCUSSIONS:
Chemistry: The 1ꢀposition harmine amine analogs were synthesized by the following reaction
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 51
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
sequence outlined in Scheme 1. m-Anisidine underwent classical diazotization to form
corresponding aryldiazonium salt which was coupled with 3ꢀcarboxyꢀ2ꢀpiperidone to give
5
0
arylhydrazone 1ꢀ2. Fisher indole cyclization of the resulting arylhydrazone in the presence of
5
0
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
formic acid afforded 1,2,3,4ꢀtetrahydroꢀ1ꢀoxoꢀβꢀcarboline 1ꢀ3. Oxidation of 1ꢀ3 using DDQ
followed by chlorination with phosphorous oxychloride generated 1ꢀchloroꢀβꢀcarboline 1ꢀ5 in
5
1ꢀ52
7
9% yield.
Finally, 1ꢀaminoꢀβꢀcarbolines 1ꢀ6 (10 analogs) were prepared via amination of 1ꢀ
o
52
chloroꢀ precursors 1ꢀ5 by heating with an excess of neat amine at 170 C in 43ꢀ87% yield.
53
Analog 1ꢀ6a was synthesized by Pd catalyzed amination of 1ꢀ5 (conversion for this amination
was surprisingly low) though provided a sufficient amount of pure compound for in vitro assay.
a
Scheme 1. Synthesis of 1ꢀamino harmine analogs
a
0
Reagents and conditions: (a) NaNO (1.03 eq.), HCl, 10 C, 1 h, (b) Ethyl 2ꢀoxopiperidineꢀ3ꢀ
2
carboxylate (1.05 eq.), KOH (1.2 eq.), pH adjusted to 4ꢀ5, water, rt , 5 h; (c) formic acid, reflux,
o
o
1
h; (d) DDQ (1.2 eq.), 1,4ꢀdioxane, 0 Cꢀrt, 1 h, 23% (4 step); (e) POCl , 150 C, 24 h, 79%; (f)
3
o
R R NH (10 eq.), 170 C, 24 h, 43ꢀ87% (for 1ꢀ6b – 1ꢀ6j); (g) Ruphos (1 mol%), RuPhos Precat
1
2
o
(
1 mol%), Azetidine (1.2 eq.), LiHMDS (1 M in THF, 2.4 eq.), 90 C, 96 h, 31% (based on 32
mg (64%) recovered starting material, for 1ꢀ6a).
ACS Paragon Plus Environment

Page 7 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
ꢀHydroxymethyl and 3ꢀHydroxymethyl substitutedꢀβꢀcarbolines were synthesized using
the sequences shown in Scheme 2. Oxidation of harmine with m-CPBA generated the
corresponding Nꢀoxide 2ꢀ1, which subsequently underwent Boekelheide rearrangement in the
presence of trifluoroacetic anhydride to give the desired 1ꢀhydroxymethyl βꢀcarboline 2ꢀ2 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
54ꢀ55
white solid in 49% yield.
Alternatively, 6ꢀmethoxyindole was reacted with oxime 2ꢀ3,
prepared from ethyl 3ꢀbromoꢀ2ꢀoxopropanoate, in the presence of sodium carbonate at room
56
temperature to give indoleꢀoxime 2ꢀ4. Oxime reduction of 2ꢀ4 with zinc powder in acetic acid
then provided tryptophanyl ester 2ꢀ5 which was underwent PictetꢀSpengler cyclization with
5
6ꢀ57
acetaldehyde to provide 1ꢀmethylꢀ3ꢀhydroxymethyl tetrahydroꢀβꢀcarboline 2ꢀ6.
Aromatization of compound 2ꢀ6 followed by reduction of the ester afforded the final compound
5
7
2
ꢀ8 as white solid. Reduction of 2ꢀ8 using Et SiH and PdCl as catalyst afforded 1,3ꢀdimethylꢀ
3 2
5
8
7ꢀmethoxyꢀβꢀcarboline 2ꢀ9 as white solid.
a
Scheme 2. Synthesis of 1ꢀhydroxymethyl and 3ꢀhydroxymethyl harmine analogs
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 51
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
a
o
Reagents and conditions: (a) mꢀCPBA (3 eq.), MeOH, CHCl , 70 C, 12 h, 47%; (b) TFA
3
anhydride (2.5 eq.), CH Cl , reflux, o/n, 49%; (c) N H OH.HCl (1 eq.), MeOH, CHCl , rt, 24h,
2
2
2
4
3
81%; (d) 2ꢀ3 (1 eq.), K CO (5.5 eq.), DCM, rt, 48 h, 34%; (e) Zn dust, AcOH, rt, o/n, 96%; (f)
2
3
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
Acetaldehyde (1 eq.), TFA (5%), DCM, rt, o/n; (g) S (2 eq.), Xylene, reflux,o/n, 75%; (h)
8
o
LiAlH (2 eq.), THF, rt, 12 h, 91%; (i) Et SiH (16 eq.), PdCl (0.2 eq.), EtOH, 90 C, 5 h, 27%.
4
3
2
Synthesis of 1ꢀ(1ꢀhydroxy)ethyl and 1ꢀacetyl harmine analogs are outlined in Scheme 3. 6ꢀ
methoxy tryptamine underwent Pictetꢀspengler cyclization with pyruvic aldehyde in presence of
5
% TFA to provide 1ꢀacetylꢀ7ꢀmethoxyꢀtetrahydroꢀβꢀcarboline 3ꢀ1. Aromatization of compound
3ꢀ1 followed by reduction of the acetyl group afforded the 1ꢀ(1ꢀhydroxyethyl) harmine analog 3ꢀ
as white solid in 49% yield.
3
a
Scheme 3. Synthesis of 1ꢀ(1ꢀhydroxy)ethyl and 1ꢀacetyl harmine analogs
a
Reagents and conditions: (a) (i) pyruvic aldehyde (1.2 eq.), TFA (5%), DCM, rt, 12 h; (ii)
KMnO (4 eq.), THF, rt, 12 h, 16% (2 steps); (b) NaBH (2 eq.), MeOH, rt, 12 h, 49%.
4
4
StructureꢀActivity Relationship Studies of Harmine Analogs: Harmine is a type I DYRK1A
inhibitor that binds to the ATP binding pocket and forms hydrogen bonding interaction with both
1
9
the side chain of Lys188 and backbone of Leu241 (Figure 1A). Recently, the crystal structure
5
9
of an ATPꢀcompetitive inhibitor DJM2005 bound to DYRK1A was also reported. These data
ACS Paragon Plus Environment

Page 9 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
showed classical binding to the ATP site with the inhibitor forming hydrogen bond interactions
with the backbone of Leu241 along with waterꢀmediated contacts with the side chains Lys188,
Glu203 and the backbone of Asp307 (Figure 1B). Interestingly, the primary amino group of
DJM2005 was also shown to form two additional hydrogen bonds with the side chains of
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
59
Asn292 and DFG motif Asp307. Moreover, it was also observed that movement of Phe170 in
the DYRK1AꢀDJM2005 complex, creating another “inducedꢀfit”pocket which is occupied by the
chlorophenyl group (Figure 1C, 1D). In comparison, harmine binds in ATPꢀbinding site without
accessing this additional space occupied by DJM2005. Hence, we proposed rational
modifications of the 1ꢀposition of harmine to access this pocket to identify new harmine based
DYRK1A inhibitors with the potential to be chemically linked to GLPꢀ1 agonists, for example,
for βꢀcell targeted delivery. Modifications of the 1ꢀposition in harmine have been explored
5
2, 60ꢀ61
previously
, but not in the context of diabetes and βꢀcell proliferation. A total of 15 harmine
analogs were synthesized following the routes described in Schemes 1 2, and 3. DYRK1A
binding activity of these analogs was screened using FRETꢀbased LanthaScreen binding assay
(
Life Technologies), initially at two concentrations:1000 nM and 300 nM. Those compounds
showing ≥ 50% inhibition at 300 nM were titrated using ten serial threeꢀfold dilutions (in
duplicate) for IC determination.
5
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 51
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 1. Comparison of Harmine and DJM2005 binding to DYRK1A. (A) Harmine binds in
the ATPꢀbinding pocket of DYRK1A making hydrogen bonding contacts (yellow dashed lines)
with the side chain of Lys188 and the backbone of Leu241 (PDB 3ANR). (B) DJM2005 also
interacts with the backbone of Leu241 and makes waterꢀmediated contacts with the side chains
Lys188, Glu203 and the backbone of Asp307. Additionally, the primary amino group interacts
with the side chain of Asn292 and Asp307 (PDB 2WO6). (C) (D) The movement of Phe 170 in
the DJM2005 complex (D) creates a pocket (orange mesh) that is occupied by the chlorophenyl
group of DJM2005. (E) Superposition of the harmine (green) and DJM2005 (pink) complexes,
highlighting the equivalent hydrogen bonding interactions with Leu241 and Lys188. Orientation
is rotated 90 degrees with respect to the view in panels C and D.
Based on our computational modeling, we investigated the effects of cycloalkylamines at
the 1ꢀposition of harmine on DYRK1A binding. First, we studied three analogs with azetidine (1ꢀ
6
a), pyrrolidine (1ꢀ6b) and piperidine (1ꢀ6c) substituents at the 1ꢀposition. Compound 1ꢀ6a
showed the best activity with IC of 159 nM. However, this compound was 5ꢀfold less active
5
0
than harmine. Increasing the ring size to 5ꢀmembered (pyrrolidine, 1ꢀ6b) and 6ꢀmembered
ACS Paragon Plus Environment

Page 11 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
piperidine, 1ꢀ6c) ring reduced the DYRK1A inhibitory activity, with DYRK1A IC of 264 nM
5
0
and 1500 nM, respectively. We also synthesized 7 analogues from commercially available
building blocks bearing substituents on the pyrrolidine at harmine 1ꢀposition to interrogate the
induced binding pocket implicated by the structure of DJM2005 complex. As predicted, the
substitution pattern on the pyrrolidine ring was very sensitive to the DYRK1A inhibition activity.
Among these analogues, 1ꢀ6d and 1ꢀ6f showed modest DYRK1A inhibition activity. 2ꢀ
Substituted pyrrolidine analogs were more potent inhibitors as compared to their 3ꢀsubstituted
analogs. Compound 1ꢀ6d with 2ꢀphenyl pyrrolidine substituent showed a DYRK1A IC of 123
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
nM comparable to 1ꢀ6a (azetidine), and a 2ꢀfold improvement over the corresponding analog 1ꢀ
6b. Unfortunately, introduction of a 3ꢀchloro group, analogous to DJM2005, on 2ꢀphenyl of the
pyrrolidine 1ꢀharmine analog was detrimental for the activity (compound 1ꢀ6i, 22% inhibition at
3
00 nM ), thus demonstrating that harmineꢀbased analogs with directly attached 1ꢀposition
.
cycloalkyamines are unable to productively access the induced fit pocket of DJM2005.
Figure 2. Binding Model of 1ꢀhydroxymethyl harmine Analogs to DYRK1A. (A) Induced fit
docking of compound 2ꢀ2 predicts that it adopts a binding mode that is distinct from harmine
(see Figure 1A). The model shows that the hydroxymethyl group at the 1ꢀposition causes the
o
scaffold to flip 180 horizontally, enabling new putative contacts with the side chains of Glu203
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 51
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 Phe308, inaccessible to harmine. The model also shows retention of the original hydrogen
bonds of harmine 2ꢀnitrogen with Lys188 and 7ꢀmethoxy oxygen with Leu241, respectively.
Hydrogen bonds are shown as yellow dashed lines (PDB 3ANR). (B) Induced fit docking of
compound 2ꢀ8 predicts that repositioning of the hydroxymethyl group to positionꢀ3 causes this
compound to revert to the canonical harmine ATPꢀsite binding mode, while creating new
contacts from the 3ꢀhydroxymethyl group with Glu203 and Phe308 proposed for the “flipped”
binding mode for 2ꢀ2 (PDB 3ANR).
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
Simple replacement of the harmine 1ꢀmethyl group with a chlorine atom (1ꢀ5)
significantly improved the DYRK1A inhibition by 3ꢀfold compared to harmine with IC of 8.81
5
0
6
0
nM (reported in literature IC 56 nM) . We also investigated the effect of introducing polar
5
0
groups like hydroxyl, hydroxymethyl and acetyl at the 1ꢀposition of harmine. Replacing 1ꢀ
methyl with a hydroxyl substituent (or its pyridone tautomer) dramatically reduced the DYRK1A
inhibition, rendering the compound inactive, unlike the Clꢀsubsitutent.
Interestingly, in contrast to a directly attatched hydroxyl group, introduction of 1ꢀ
hydroxymethyl group (2ꢀ2) showed potent DYRK1A inhibition with IC of 55 nM (reported in
5
0
60
lietrature IC50 106 nM) , comparable but slightly less potent than harmine and a 2ꢀfold
improvement over 1ꢀ(2ꢀphenylpyrrolidinꢀ1ꢀyl) harmine analog 1ꢀ6d. We studied the binding
pose of this compound to DYRK1A by induced fit docking and observed that it adopts a new
binding pose that is distinct from the classical ATPꢀbinding pose for harmine (Figure 1A).
Based on the induced fit modeling, the 1ꢀhydroxymethyl substituent causes the core scaffold to
o
horizontally flip by 180 , creating the possibility for the hydroxyl group to form two new
hydrogen bonding interactions with the side chain of Glu203 and the backbone of Phe308 while
maintaining the canonical hydrogen bonding interactions with Leu241 and Lys188 (Figure 2A).
ACS Paragon Plus Environment

Page 13 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In order to retain this new hydrogen bond interaction while maintaining the canonical binding
pose of harmine, a “nonꢀflipped” version of compound 2ꢀ2 with a hydroxymethyl group at the 3ꢀ
carboline position of harmine (compound 2ꢀ8) was synthesized as shown in the (Figure 2B).
Induced fit docking of compound 2ꢀ8 predicted that repositioning of the hydroxymethyl group to
the 3ꢀposition would cause it to revert to the canonical harmine binding mode. We were gratified
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
to learn that 3ꢀhydroxymethyl substituted analog 2ꢀ8 showed IC of 49.5nM, comparable to
50
compound 2ꢀ2. Removal of the hydroxyl group (3ꢀmethyl carboline analog 2ꢀ9) caused a
significant decrease in the DYRK1Abinding affinity with an IC = 971 nM. These data indicate
5
0
the importance of the hydroxyl group in 2ꢀ8, further corroborating our modeling results.
Table 1. DYRK1A inhibition and βꢀCell proliferation of harmine analogs
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 51
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
a = IC values are determined using ten serial three fold dilutions (in duplicate)
nd = not determined
50
Introduction of an alpha methyl group to the 1ꢀhydroxymethyl substituent of harmine 3ꢀ3
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
led to a complete loss of DYRK1A binding activity at the screening concentration while the
planar 1ꢀacetyl substituted analog (3ꢀ2) had an IC (66.7 nM) similar to compounds 2ꢀ2 and 2ꢀ8.
5
0
This indicates that the 1ꢀhydroxymethyl group of 2ꢀ2 occupies constrained space that does not
accommodate further steric bulk, a possibility to have binding mode similar to either 2ꢀ2 or 2ꢀ8.
Figure 3. Effects of harmine analogs on human beta cell proliferation. (A) Initial screening of
harmine analogs on human beta cell proliferation at 10 µM. DMSO was used as negative control and
harmine was used as positive control (n = 4ꢀ5). (B) A representative example from A of a Kiꢀ67 and
insulin double positive cells induced by analog 2ꢀ8. (C) and (D) Doseꢀresponse curves for 2ꢀ2 and 2ꢀ8 in
ACS Paragon Plus Environment

Page 15 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
human β cells (n = 4 for each dose, harmine concentration 10 µM). (E) Quantification of nuclear
frequency of NFATC1ꢀGFP in R7T1 rodent beta cell lines treated with harmine, 2ꢀ2 and 2ꢀ8 (10 ꢁM, 24
hr; n = 3 for each compound). (F) A representative example of 2ꢀ8 (10 ꢁM, 24 hr) increasing the nuclear
frequency of NFATC1ꢀGFP in R7T1 rodent beta cells. In all relevant panels, error bars indicate SEM and,
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
*
indicates P < 0.05. A minimum of 1,000 beta cells was counted for each graph.
Human βꢀcell proliferation assay: Eight compounds, 1ꢀ5, 1ꢀ6a, 1ꢀ6b, 1ꢀ6d, 1ꢀ6f, 2ꢀ2, 2ꢀ8 and 3ꢀ
which showed DYRK1A inhibition IC < 250 nM were assessed for their ability to induce
2
5
0
human βꢀcell proliferation in vitro. Among these harmine analogs, compounds 2ꢀ2, 2ꢀ8 and 1ꢀ5
exhibited human βꢀcell proliferation at 10 µM comparable to that of harmine, as measured by
Ki67 immunolabeling (Figure 3A). A representative example of a Kiꢀ67 and insulin double
6
positive cells induced by analog 2ꢀ8 is shown in Figure 3B as per our literature protocols.
Compound 3ꢀ2 with acetyl substituent at the harmine 1ꢀposition, despite having an IC50
comparable to 2ꢀ2 and 2ꢀ8, showed reduced βꢀcell proliferation at 10 µM. However, when the
compound was tested at higher concentration of 30 µM, proliferation in the range comparable to
harmine was observed. Both the compounds 2ꢀ2 and 2ꢀ8 caused proliferation in a doseꢀ
dependent manner (Figure 3C, 3D) with compound 2ꢀ8 showing superior human βꢀcell
proliferation at lower dose as compared to 2ꢀ2. Unlike harmine, which causes beta cell toxicity at
doses higher than 10 µM, compound 2ꢀ2 and 2ꢀ8 did not cause toxicity even at 30 µM.
Compounds 2ꢀ2 and 2ꢀ8 also drove NFAT2 translocation in rat cell line that transiently
transduced NFATꢀGFP fusion genes to nucleus as observed for harmine. Collectively, these
observations indicate that these compounds drive βꢀcell proliferation qualitatively like harmine,
by inducing translocation of NFATs to the nucleus, likely allowing access to promoters of genes
that subsequently drive human βꢀcell proliferation (Figure 3E, 3F).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 51
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
Interestingly, except for 1ꢀ6a, the 1ꢀamino substituted harmine analogs 1ꢀ6b, 1ꢀ6d and 1ꢀ
f did not show any βꢀcell proliferation despite showing DYRK1A inhibition < 270 nM, which
6
may indicate a DYRK1A potency threshold for βꢀcell proliferation. An alternate rationale may
be that some of the compounds with reduced βꢀcell proliferation, such as 1ꢀ5 and 1ꢀ6a, may have
activity on antiꢀproliferative kinases in addition to DYRK1A that limit their βꢀcell proliferative
capability. Furthermore, we also may suggest that the difference between in vitro DYRK1A
binding and the human beta cell proliferation assay might be attributed to physicochemical
properties such as, cell permeability and other physical chemical properties of the 1ꢀ6 harmine
analogues. Studies to address these rationales are onꢀgoing.
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
Table 2. Kinome scan of compound 2ꢀ2, 2ꢀ8 and harmine
ACS Paragon Plus Environment

Page 17 of 51
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
a = compounds were screened 10 µM against 468 kinases using Discoverx Kinome Scan. Results
for primary screen binding interactions are reported as '% DMSO Control’, where lower values
indicate stronger affinity (see the Supporting Information for details).
Kinome scan profile: As previously mentioned above, harmine is known to exhibit varying
degrees of inhibition for kinases other than DYRK1A. In order to understand kinase selectivity,
we carried out a kinome profiling of compound 2ꢀ2, 2ꢀ8 and harmine on 468 kinases at 10 µM
concentration (Table 2, more potent activities < 20% remaining colored as shown). Harmine
inhibited 17 kinases (<20% activity remaining) in addition to DYRK1A at a screening
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 51
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
concentration of 10 ꢀM similar to earlier reports. Compound 2ꢀ2 exhibited a cleaner kinome
profile as compared to harmine with significantly reduced inhibition against DYRK1B,
CSNK1G2, CSNK2A1, HIPK2, HIPK3, IRAK1, IRAK3 and VPS3 at 10 µM. Additionally, in
comparison to harmine, it showed affinity greater than harmine to only PIK4CB at the screening
concentration (3.7 vs 19 % activity remaining for 2ꢀ2 and harmine, respectively; see
Supplementary Information for complete Discoverx screening data). Interestingly, the kinome
profile observed for compound 2ꢀ8 was much less elective than that of both harmine and 2ꢀ2. In
addition to the kinases for which harmine and 2ꢀ2 showed inhibition, compound 2ꢀ8 exhibited
inhibition of several additional kinases including CDKs, DAPKs and PIMs (<20% activity
remaining). Despite having poor kinome selectivity profile, it is interesting to note that
compound 2ꢀ8 showed βꢀcell proliferation activity and DYRK1A inhibition comparable to
harmine and 2ꢀ2. One probable explaination could be that some of these kinases have
compensatory mechanism to either induce or inhibit βꢀcell proliferation. The kinome profile data
indicates that between the two harmine scaffold analogs 2ꢀ2 and 2ꢀ8, which have similar βꢀcell
proliferation activity and DYRK1A binding, compound 2ꢀ2 has improved kinase selectivity thus
making it the lead compound for further studies.
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
CONCLUSIONS:
We have reported the structureꢀactivity relationships of DYRK1A inhibition and βꢀcell
proliferation for new harmine analogs using an integrated, structureꢀbased drug design,
computational modeling and medicinal chemistry optimization approach. In order to access the
putative induced binding pocket occupied by DJM2005, we synthesized several 1ꢀamino harmine
analogs (1ꢀ6a to 1ꢀ6j) to investigate their effect on DYRK1A kinase binding and human βꢀcell
proliferation. Most of the 2ꢀamino harmine compounds showed reduced DYRK1A inhibition
ACS Paragon Plus Environment

Page 19 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
activity at the screening concentration of 1 µM with the exceptions of harmine analogs 1ꢀ6a, 1ꢀ
6
b, 1ꢀ6d. These analogs showed a range of 5ꢀ to 9ꢀfold reduced DYRK1A inhibitory activity as
compared to harmine, indicating that harmineꢀbased analogs with cycloalkylamines directly
attached at the 1ꢀposition are unable to productively access the induced fit pocket of DJM2005,
suggesting alternate linking groups need to be explored.
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
In contrast, we have shown that harmine analogs 2ꢀ2, 2ꢀ8 and 3ꢀ2 bearing small polar
groups like, hydroxymethyl or acetyl exhibited potent DYRK1A inhibition with IC in the range
5
0
of 49ꢀ67 nM, 2ꢀfold more active than 1ꢀazetidine harmine analog 1ꢀ6b. However, introduction
of the directly attached 1ꢀhydroxy substituent (or its pyridone tautomer) was detrimental for
DYRK1A inhibition. Interestingly among all the harmine analogs synthesized, the 1ꢀchloro
substituted harmine analog (1ꢀ5) significantly improved DYRK1A inhibition and was the most
potent DYRK1A inhibitor compound, with an IC of 8.8 nM, though it was not explored further
5
0
as a lead due to its potential for nucleophilic substitution. Among the 8 compounds with IC <
5
0
2
50 nM against DYRK1A, (1ꢀ5, 1ꢀ6a, 1ꢀ6b, 1ꢀ6d, 1ꢀ6f, 2ꢀ2, 2ꢀ8 3ꢀ2) only 1ꢀ5, 2ꢀ2, 2ꢀ8 and 3ꢀ2
exhibited human βꢀcell proliferation comparable to that of harmine at similar or reduced
concentrations. Importantly, harmine analogs 2ꢀ2 and 2ꢀ8 were most effective for inducing
human βꢀcell proliferation in vitro, indicating that introduction of polar groups like
hydroxymethyl at 1ꢀ and 3ꢀpositions of harmine improves the βꢀcell proliferation, possibly by
improving kinase selectivity, as shown for 2ꢀ2. None of the 1ꢀamino harmine analogs caused any
βꢀcell proliferation. Kinome scan of 468 kinases for compound 2ꢀ2 and 2ꢀ8 indicated that
compound 2ꢀ2 exhibits a cleaner kinome profile as compared to harmine and 2ꢀ8 at a
concentration of 10ꢀM. In contrast, compound 2ꢀ8 exhibited much lower kinase selectivity,
inhibiting several additional kinases including CDKs, DAPKs and PIMs (<20% activity remaining).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 51
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
These data indicate the potential for improvement of the harmine scaffold for kinase selectivity,
resulting in βꢀcell proliferation at lower concentration in vitro, with the potential for a cleaner
offꢀtarget profile. We have been able to successfully modify harmine to identify a novel kinase
selective DYRK1A inhibitor 2ꢀ2 with improved kinase selectivity and βꢀcell proliferation ability.
These observations suggest that further improvements in DYRK1A binding affinity as well
pharmacological studies in addition to DYRK1A binding of the harmine analog series is an
attractive approach to the development of effective and novel diabetes therapeutics for βꢀcell
proliferation. The results of our onꢀgoing structureꢀbased drug design, medicinal chemistry SAR
and pharmacological studies will be reported in due course.
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
EXPERIMENTAL:
1
13
Materials and Methods. H and C NMR spectra were acquired on a Bruker DRXꢀ600
1
13
spectrometer at 600 MHz for H and 150 MHz for C. TLC was performed on silica coated
aluminum sheets (thickness 200 µm) or alumina coated (thickness 200 µm) aluminum sheets
supplied by Sorbent Technologies and column chromatography was carried out on Teledyne
ISCO combiflash equipped with a variable wavelength detector and a fraction collector using a
RediSep Rf high performance silica flash columns by Teledyne ISCO. LCMS/HPLC analysis for
purity determination and HRMS was conducted on an Agilent Technologies G1969A highꢀ
resolution APIꢀTOF mass spectrometer attached to an Agilent Technologies 1200 HPLC system.
Samples were ionized by electrospray ionization (ESI) in positive mode. Chromatography was
performed on a 2.1 × 150 mm Zorbax 300SBꢀC18 5ꢀꢁm column with water containing 0.1%
formic acid as solvent A and acetonitrile containing 0.1% formic acid as solvent B at a flow rate
of 0.4 mL/min. The gradient program was as follows: 1% B (0−1 min), 1−99% B (1− 4 min),
ACS Paragon Plus Environment

Page 21 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and 99% B (4−8 min). The temperature of the column was held at 50 °C for the entire analysis.
The purity of all the compounds were ≥ 95%. The chemicals and reagents were purchased from
Aldrich Co., Alfa Aesar, Enamine, TCI USA. All solvents were purchased in anhydrous from
Acros Organics and used without further purification.
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
Molecular Modeling: All molecular modeling procedures were carried with the smallꢀmolecule
6
2
drug discover suite of Schrödinger LLC, release 2016ꢀ3. The Induced Fit Docking (IFD)
protocol combining Glide and Prime was used to dock the compounds using the crystal structure
5
9
of DYRK1A/harmine complex (PDB 3ANR) as a starting point. The IFD protocol modifies
both the ligand and protein conformations to account for receptor flexibility. Within this protocol
the proteinꢀligand complex conformations are ranked using the IFDScore, which is an empirical
scoring function that combines Prime_Energy (the energy of the protein conformation),
GlideScore (an approximation of the ligand binding free energy), and Glide_Ecoul (Coulombic
6
2
interaction energy). While no explicit water molecules are included in our starting model due to
the uncertainty in positioning these molecules, GlideScore does include solvation terms that
63
simulate the effects of solvent. Glide docks explicit water molecules into the binding site for
each energetically favorable ligand pose and employs empirical scoring terms that measure the
exposure of various groups to the explicit waters. This explicit approach overcomes some of the
limitations of continuum solvation models, particularly in the highly constrained environment of
a binding site containing a bound ligand. Ligands and receptor were prepared using LigPrep and
64
the Protein Preparation Wizard. MM/GBSA binging energy calculations were carried out with
6
5
Prime MMꢀGBSA using the VSGB 2.0 solvation model. Figures were prepared using PyMOL
2
.0 (Schrödinger, LLC).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 51
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
DYRK1A Binding Assays: Compounds were tested for DYRK1A binding activity by a
commercial kinase profiling services, Life Technologies which uses the FRETꢀbased
6
6
LanthaScreen® Eu Kinase Binding Assay . Compounds were screened for DYRK1A activity at
concentrations of 1000 nM and 300 nM in duplicates. The IC50 was determined by 10 point
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
66
LanthaScreen® Eu Kinase Binding Assay in duplicates.
Βꢀcell proliferation assay: Human pancreatic islets were obtained from the NIH/NIDDKꢀ
supported Integrated Islet Distribution Program (IIDP) (https://iidp.coh.org). Islets were first
6
dispersed with Accutase (Sigma, St. Louis, MO) onto coverslips as described earlier. Dispersed
islets were treated with vehicle (0.1% DMSO) or with harmineꢀrelated compounds in RPMI1640
6
complete medium for 96 hours. Then the cells were fixed and stained for insulin and Ki67.
Total insulin positive cells and cells double positive Ki67 and insulin positive cells were imaged
and counted. Islets from 3ꢀ5 different adult donors were tested as descriebed in Figure 3, and at
least 1000 beta cells cells were counted for human islet donor.
NFAT2 translocation assay: The immortalized R7T1 rat beta cell line was infected with aan
6
adenovirus expressing NFAT2ꢀGFP for 48 hours, then the cells were treated with different
compounds at 10 µM for another 24 hours. The cells were imaged and counted for the NFAT2ꢀ
GFP nuclear translocation. At least 1000 cells were counted.
Kinome Scan Profile: Compounds were screened against 468 kinases at single concentration of
6
7
1
0 ꢀM in duplicates at DiscoverX using their proprietary KINOMEscan® Assay. The results
for primary screen binding interactions are reported as '% DMSO Ctrl', where lower values
indicate stronger affinity.
Synthetic Procedures:
ACS Paragon Plus Environment

Page 23 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
7
ꢀMethoxyꢀ2,9ꢀdihydroꢀβꢀcarbolinꢀ1ꢀone (1ꢀ4)
m-Anisidine (1.096 mL, 38.56 mmol) was dissolved in 15 mL of concentrated hydrochloric acid
and, after cooling 15 mL of water was added to the solution. Sodium nitrite (2.74 g, 39.71 mmol)
in 15mL of water was added dropwise to the cold suspension and stirred for 1 hour while
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
o
maintaining the temperature below 10 C. This solution was added to a cold solution of ethyl 3ꢀ
carboxyꢀ2ꢀpiperidone (7.0 g, 40.88 mmol) and potassium hydroxide (2.63 g, 47.01 mmol) in 20
mL of water which had been kept overnight at room temperature. The pH of the reaction mixture
was adjusted to 4ꢀ5 by saturated aqueous solution of sodium acetate. The resultant mixture was
stirred at room temperature for 5 hours. The yellow solid was filtered and washed with small
amount of water and ethanol to give 6ꢀmethoxyphenylhydrazone 1ꢀ2 which was immediately
taken to the next step. The phenylhydrazone 1ꢀ2 was refluxed in 25 mL of formic acid for 1 hour.
The solution was neutralized with saturated aqueous solution of sodium carbonate and extracted
with 100 mL of ethyl acetate three times. The organic layer was collected, dried over magnesium
sulfate, filtered and solvent was rotary evaporated to give the desired tetrahydroꢀoxoꢀβꢀcarboline
2
ꢀ3. To a solution of Compound 1ꢀ3 in 60 mL of 1,4ꢀdioxane was added DDQ (2.65 g, 11.7
o
mmol) in 40 mL 1,4ꢀdioxane dropwise at 0 C and stirred for 1 hour at room temperature. Upon
completion of the reaction, 100 mL of water was added to the reaction. The product mixture was
transferred to a separatory funnel and extracted with ethyl acetate (100 mL X 3) three times. The
organic layer was washed with 0.1 N sodium hydroxide solution (50 mL X 3). The combined
organic layer was then dried over magnesium sulfate, filtered and concentrated to provide the
desired 7ꢀmethoxyꢀ2,9ꢀdihydroꢀβꢀcarbolinꢀ1ꢀone 1ꢀ4 (1.91 g, 23% in 4 steps) as yellow solid.
1
HꢀNMR (600 MHz, CD OD): δ 7.85 (d, 1H, J = 9 Hz), 7.12 (d, 1H, J = 4.8 Hz), 7.06ꢀ7.08 (m,
3
1
3
2H), 6.86 (d, 1H, J = 6.6 Hz), 3.89 (s, 3H); CꢀNMR (150 MHz, d ꢀDMSO): δ159.40, 155.72,
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 51
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
40.74, 127.75, 125.12, 122.58, 116.40, 110.78, 99.85, 94.69, 55.61; HRMS (ESI): m/z [M +
H]+ calcd for C12H11N2O2+: 215.0815, found: 215.0806.
1ꢀChloroꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ5):
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 solution of 7ꢀmethoxyꢀ2,9ꢀdihydroꢀβꢀcarbolinꢀ1ꢀone (1.80 g, 8.49 mmol) in POCl (20 mL)
3
o
was stirred at 150 C for 24 hours. The mixture was neutralized with the saturated aqueous
solution of sodium carbonate. The solution was transferred to separatory funnel and extracted
with ethyl acetate (50 mL X 3). The organic layer was collected, dried over magnesium sulfate,
filtered and concentrated to provide the desired 1ꢀChloroꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline 1ꢀ5 (1.56 g,
1
7
9%) as yellow solid. HꢀNMR (600 MHz, CDCl ): δ 8.31 (s, 1H), 8.19 (d, 1H, J = 5.4 Hz),
3
1
3
7.96 (d, 1H, J = 8.4 Hz), 7.78 (s, 1H, J = 5.4 Hz), 7.0 (s, 1H), 6.95 (m, 1H), 3.93 (s, 3H); Cꢀ
NMR (150 MHz, d ꢀDMSO): δ161.14, 142.87, 138.10, 132.93, 132.83, 130.73, 123.44, 115.06,
6
1
14.39, 110.60, 95.28, 55.78; HRMS (ESI): m/z [M + H]+ calcd for C12H10ClN2O+: 233.0476,
found: 233.0471. Purity >95%
General procedure for the synthesis of 1ꢀaminoꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6)
A solution of 1ꢀchloroꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1 mmol) and amine (10 mmol) in a sealed
o
pressure vessel was heated to 170 C for 24 hour. Upon the completion of the reaction monitered
by LCMS, the reaction mixture was concentrated and purified by flash column chromatography
using DCM/MeOH as eluent to afford the desired 1ꢀaminoꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline 1ꢀ6 as
white solid.
53
1
ꢀ(azetidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6a)
A screwꢀcap pressure vessel, equipped with a magnetic stir bar, was charged with 1ꢀ5 (50 mg,
0.21 mmol) RuPhos (1 mg, 1 mol %) and RuPhos precatalyst (P1) (1.75 mg, 1 mol %). The vial
ACS Paragon Plus Environment

Page 25 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
was evacuated and backfilled with argon, and sealed with a Teflon screw cap. LiHMDS (1M in
THF, 2.4 eq.) was added via syringe, followed by azetidine (0.25 mmol, 1.2 equiv.). The reaction
mixture was heated at 90 ˚C for 96 h. Conversion for the reaction was low. The solution was
allowed to cool to room temperature, then quenched by the addition of 1M HCl (1 mL), diluted
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
with EtOAc and poured into sat. NaHCO . After extracting with EtOAc, the combined organic
3
layers were washed with brine, dried over MgSO4, then concentrated and purified by flash
column chromatography using MeOH/DCM (10/90) as eluent to afford compound 1ꢀ6a (6.5 mg,
1
3
1% based on 32 mg (64%), recovered starting material) as white solid. HꢀNMR (600 MHz,
CDCl ): δ 7.98 (m, 1H), 7.90 (d, 1H, J = 9 Hz), 7.30 (m,1 H), 6.94 (m,1 H), 6.89 (d, 1H, J = 8.4
3
1
3
Hz), 4.38 (t, 4 H, J = 7.2 Hz), 3.90 (s, H), 2.52 (m, 2 H); CꢀNMR (150 MHz, d ꢀDMSO): δ
6
1
5
59.83, 149.05, 141.87, 136.84, 128.30, 123.52, 122.33, 115.48, 109.56, 106.39, 95.22, 55.61,
1.82, 17.42; HRMS (ESI): m/z [M + H]+ calcd for C15H16N3O+: 254.1288, found: 254.1284;
Purity >95%
1
ꢀ(Pyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6b)
1
White solid. Yield 64%. HꢀNMR (600 MHz, d ꢀDMSO): δ 8.32 (s, 1H), 7.94 (d, 1H, J = 5.4
6
Hz), 7.89 (d, 1H, J = 8.4 Hz), 7.22 (d, 1H, J = 5.4 Hz), 6.93 (s, H), 6.88 (d, 1H, J = 8.4 Hz), 3.90
1
3
(
m, 7H), 2.07 (s, 4 H); CꢀNMR (150 MHz, d ꢀDMSO): δ 159.64, 146.72, 141.52, 136.88,
6
1
28.60, 123.63, 122.09, 115.48, 109.43, 104.82, 95.28, 55.58, 48.39, 25.35; HRMS (ESI): m/z
[
M + H]+ calcd for C16H18N3O+: 268.1444, found: 268.1460; Purity >95%
1ꢀ(Piperidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6c)
1
White solid. Yield 87%. HꢀNMR (600 MHz, d ꢀDMSO): δ 8.05 (d, 1H, J = 5.4 Hz), 7.90 (d,
6
1
H, J =8.4 Hz), 7.48 (d, 1H, J = 4.8 Hz), 7.04 (s, 1H), 6.90 (m, 1H), 3.90 (s, 3H), 3.48 (broad s,
1
3
4H), 1.84 (broad s, 4H), 1.70 (m, 2H); CꢀNMR (150 MHz, d ꢀDMSO): δ159.98, 149.23,
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 51
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
2
41.74, 136.57, 129.05, 126.62, 122.33, 115.66, 109.43, 108.75, 95.20, 55.60, 49.92, 26.09,
4.82; HRMS (ESI): m/z [M + H]+ calcd for C17H20N3O+: 282.1601, found: 282.1590; Purity
>95%
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
ꢀ(2ꢀPhenylpyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6d)
1
Brown solid. Yield 80%. HꢀNMR (600 MHz, CDCl ): δ 7.98 (d, 1H, J = 5.4 Hz), 7.82 (d, 1H, J
3
=
8.4 Hz), 7.61 (s, 1H), 7.51 (d, 2H, J = 7.2 Hz), 7.44 (t, 2H, J = 7.8 Hz), 7.34 (t, 2H, J = 7.2
Hz), 7.24 (d, 1H, J = 5.4 Hz), 6.79 (m, 1H), 6.49 (d, 1H, J = 1.8 Hz), 5.48 (d, 1H, J = 8.4 Hz),
13
4
.06 (m, 1H), 4.01 (m, 1H), 3.82 (s, 3 H), 2.46 (m, 1H), 2.04 (m, 3H); CꢀNMR (150 MHz, d ꢀ
6
DMSO): δ 159.75, 146.24, 145.92, 141.60, 136.78, 128.85, 128.39, 126.45, 126.28, 124.43,
122.15, 115.49, 109.43, 105.63, 95.25, 61.35, 55.62, 50.48, 24.53; HRMS (ESI): m/z [M + H]+
calcd for C22H22N3O+: 344.1757, found: 344.1761; Purity >95%
1
ꢀ(3ꢀPhenylpyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6e)
1
Brown solid. Yield 59%. HꢀNMR (600 MHz, d ꢀDMSO): δ 8.17 (s, 1H), 7.98 (d, 1H, J = 6
6
Hz), 7.91 (d, 1H, J = 9 Hz), 7.37 (m, 4H), 7.27 (m, 2H), 6.91 (s, 1H), 6.88 (d, 2H, J = 9 Hz), 4.33
(t, (d, 1H, J = 8.4 Hz), 4.09 (m, 1H), 4.01 (m, 1H), 3.94 (t, 1H, J = 8.4 Hz), 3.81 (s, 3H), 3.60 (m,
1
3
1
1
3
H), 2.49 (m, 1H), 2.24 (m, 1H);
CꢀNMR (150 MHz, CDCl ): δ161.19, 143.50, 141.54,
3
39.47, 130.03, 128.60, 126.97, 121.83, 114.64, 112.61, 104.69, 95.30, 55.53, 50.12, 43.47,
2.24; HRMS (ESI): m/z [M + H]+ calcd for C22H22N3O+: 344.1757, found: 344.1750; Purity
>
95%
1ꢀ(2ꢀBenzylpyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6f)
1
Brown solid. Yield 65%. HꢀNMR (600 MHz, CDCl ): δ 8.05 (d, 1H, J = 5.4 Hz), 7.91 (d, 2H, J
3
=
9 Hz), 7.28 (m, 5H), 6.88 (m, 2H), 4.68 (s, 1H), 3.90 (m, 5H), 3.22 (m,1H), 2.79 (m, 1H), 1.97
1
3
(m, 3H), 1.84 (m, 1H); CꢀNMR (150 MHz, CDCl ): δ160.66, 142.04, 138.44, 129.63, 128.26,
3
ACS Paragon Plus Environment

Page 27 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
26.31, 121.94, 115.48, 110.80, 105.33, 94.95, 60.45, 55.54, 50.04, 39.36, 29.77, 23.65; HRMS
(ESI): m/z [M + H]+ calcd for C23H24N3O+: 358.1914, found: 358.1900; Purity >95%
1ꢀ(3ꢀBenzylpyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6g)
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
Brown solid. Yield 71%. HꢀNMR (600 MHz, CDCl ): δ 8.14 (s, 1H), 7.96 (d, 1H, J = 5.4 Hz),
3
7
.90 (d, 2H, J = 9 Hz), 7.32 (m, 2H), 7.25 (m, 4H), 6.93 (s,1H), 6.87 (d, 1H, J = 5.4 Hz), 4.01
(m, 2H), 3.90 (m, 4H), 1.08 (t, 1H, J = 9 Hz), 2.87 (m, 1H), 2.81 (m, 1H), 2.69 (m,1H), 2.17 (m,
13
1
1
3
H), 1.83 (m, 1H); CꢀNMR (150 MHz, d ꢀDMSO): δ159.62, 146.71, 141.52, 141.15, 137.03,
6
29.15, 128.76, 126.38, 123.56, 122.08, 115.49, 109.41, 104.81, 95.29, 55.58, 53.80, 47.95,
1.18; HRMS (ESI): m/z [M + H]+ calcd for C23H24N3O+: 358.1914, found: 358.1901; Purity
>95%
1
ꢀ(2ꢀ(2ꢀChlorophenyl)pyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6h)
1
Yellow solid. Yield 77%. HꢀNMR (600 MHz, CDCl ): δ 7.98 (d, 1H, J = 5.4 Hz), 7.83 (d, 1H, J
3
=
8.4 Hz), 7.52 (m, 3H), 7.25 (m, 3H), 6.80 (m, 1H), 6.59 (d, 1H, J = 2.4 Hz), 5.80 (d, 1H, J =
1
3
8.4 Hz), 4.10 (m, 2H), 3.84 (s, 3 H), 2.54 (m, 1H), 2.04 (m, 3H); CꢀNMR (150 MHz, d6ꢀ
DMSO): δ159.82, 145.77, 142.92, 141.72, 136.88, 131.91, 129.73, 129.08, 128.20, 127.30,
1
24.39, 122.17, 115.53, 109.47, 105.73, 95.33, 59.41, 55.62, 50.31, 32.65, 24.60; HRMS (ESI):
m/z [M + H]+ calcd for C22H21ClN3O+: 378.1368, found: 378.1366; Purity >95%
1
ꢀ(2ꢀ(3ꢀChlorophenyl)pyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6i)
1
Yellow solid. Yield 68%. HꢀNMR (600 MHz, CDCl ): δ 7.96 (d, 1H, J = 5.4 Hz), 7.86 (d, 1H, J
3
=
8.4 Hz), 7.69 (s, 1H), 7.39 (m, 1H), 7.37 (d, 1H, J = 7.8 Hz), 7.33 (t, 1H, J = 7.2 Hz), 7.28 (m,
2
1
1
H), 6.83 (m, 1H), 6.65 (m, 1H), 5.51 (d, 1H, J = 8.4 Hz), 4.07 (m, 2H), 3.84 (s, 3H), 2.46 (m,
13
H), 2.04 (m, 3H); CꢀNMR (150 MHz, d ꢀDMSO): δ159.84, 148.77, 146.08, 141.72, 136.73,
6
33.18, 130.33, 129.03, 126.48, 125.14, 122.20, 115.53, 109.51, 105.94, 95.30, 61.51, 55.62,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 51
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
5
3
0.60, 35.23, 24.75; HRMS (ESI): m/z [M + H]+ calcd for C22H21ClN3O+: 378.1368, found:
78.1377; Purity >95%
1ꢀ(2ꢀ(4ꢀChlorophenyl)pyrrolidinꢀ1ꢀyl)ꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (1ꢀ6j)
1
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
Yellow solid. Yield 43%. HꢀNMR (600 MHz, CDCl ): δ 7.97 (d, 1H, J = 5.4 Hz), 7.84 (d, 1H, J
3
=
8.4 Hz), 7.66 (s, 1H), 7.44 (d, 2H, J = 8.4 Hz), 7.38 (d, 2H, J = 8.4 Hz), 7.27 (m, 1H), 6.83 (m,
1
3
1
H), 6.64 (m, 1H), 5.50 (d, 1H, J = 8.4 Hz), 4.06 (m, 2H), 3.84 (s, 3H), 2.47 (m, 1H), 2.04 (m,
1
3
H); CꢀNMR (150 MHz, d ꢀDMSO): δ159.80, 146.08, 144.98, 141.66, 136.78, 130.85, 128.95,
6
28.33, 124.47, 122.16, 115.51, 109.46, 105.81, 95.28, 60.99, 55.61, 50.49, 35.26, 24.63; HRMS
(ESI): m/z [M + H]+ calcd for C22H21ClN3O+: 378.1368, found: 378.1365; Purity >95%
βꢀCarbolineꢀNꢀoxide (2ꢀ1)
To a solution of harmine (500 mg, 2.35 mmol) in 8 mL of chloroform was added 3ꢀ
chloroperoxybenzoic acid (1.22 g, 7.05 mmol) and refluxed for 24 hours. The reaction mixture
was concentrated and purified by flash chromatography using DCM/MeOH as eluent to give the
1
desired product βꢀcarbolineꢀNꢀoxide 2ꢀ1 (255 mg, 47%) as white solid. HꢀNMR (600 MHz, d ꢀ
6
DMSO): δ 11.60 (s, 1H), 8.04 (d, 1H, J = 6.6 Hz), 8.00 (d, 1H, J = 8.4 Hz), 7.86 (d, 1H, J = 6.6
1
3
Hz), 6.97 (s, 1H), 6.85 (m, 1H), 3.85 (s, 3H), 2.61 (s, 3H); CꢀNMR (150 MHz, d ꢀDMSO):
6
δ159.82, 143.39, 136.55, 132.79, 131.31, 122.35, 119.13, 115.62, 113.67, 109.57, 95.18, 55.78,
1
3.00; HRMS (ESI): m/z [M + H]+ calcd for C13H13N2O2+: 229.0972, found: 229.0962; Purity
95%
ꢀHydroxymethylꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline (2ꢀ2)
>
1
Trifluoroacetic anhydride (2.20 mL, 15.87 mmol) was added to a stirred mixture of βꢀcarbolineꢀ
o
Nꢀoxide (2ꢀ1) (724 mg, 3.17 mmol) and CH Cl (20 mL) at 0 C. After being stirred for 30 min,
2
2
the mixture was refluxed 12 h. Upon completion of the reaction monitored by TLC, the mixture
was concentrated and purified by flash column chromatography using DCM/MeOH as eluent to
ACS Paragon Plus Environment

Page 29 of 51
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
afford the 1ꢀhydroxymethylꢀ7ꢀmethoxyꢀ9Hꢀβꢀcarboline 2ꢀ2 (358 mg, 49%) as white solid. Hꢀ
NMR (600 MHz, CD OD): δ 8.39 (d, 1H, J = 6 Hz ), 8.27 (d, 1H, J = 6 Hz ), 8.22 (d, 1H, J = 9
3
1
3
Hz ), 7.15 (s, 1H), 7.06 (m, 1H), 5.31 (s, 2H), 3.96 (s, 3 H); CꢀNMR (150 MHz, d ꢀDMSO):
6
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
δ163.11, 145.98, 140.04, 133.09, 131.95, 129.49, 124.80, 115.17, 113.83, 112.77, 94.96, 58.27,
5
6.10; HRMS (ESI): m/z [M + H]+ calcd for C13H13N2O2+: 229.0972, found: 229.0967; Purity
95%
ꢀBromoꢀ2ꢀhydroxyiminoꢀpropionic acid ethyl ester (2ꢀ3)
>
5
6
3
Hydroxylamine hydrochloride (1.81 g, 25.63 mmol) was added to a stirred solution of ethyl
bromopyruvate (5.01 g, 25.63 mmol) in chloroform (50 mL) and methanol (70 mL) at room
temperature. The reaction mixture was stirred for 24 hours and concentrated under reduced
pressure. The crude was dissolved in dichloromethane (300 mL) and washed with 0.1 N of
hydrochloric acid and brine. The organic layer was collected, dried over magnesium sulfate,
filtered, and concentrated to give 3ꢀbromoꢀ2ꢀhydroxyiminoꢀpropionic acid ethyl ester 2ꢀ3 (4.36
1
g, 81%) as white solid. HꢀNMR (600 MHz, CDCl ): δ 10.08 (bs, 1H), 4.38 (q, J=7.2 Hz, 2H),
3
4
.27 (s, 2H), 1.39 (t, J=7.2 Hz, 3H); ESI: m/z [M + H]+ 209.97.
3
ꢀ(6ꢀMethoxyꢀ1Hꢀindolꢀ3ꢀ yl)ꢀpropionic acid ethyl ester (2ꢀ4)
A solution of 3ꢀbromoꢀ2ꢀhydroxyiminoꢀpropionic acid ethyl ester 2ꢀ3 (1.75 g, 8.35 mmol) in
CH Cl (20 mL) was slowly added dropwise to a stirring mixture of 6ꢀmethoxyindole (1.23 g,
2
2
8.35 mmol) and Na CO (4.86 g, 45.92 mmol) in CH Cl (20 mL) at room temperature. The
2
3
2
2
mixture was stirred for 48 h, filtered through Celite, concentrated, and purified by flash column
chromatography using ethyl acetate\Hexanes as eluent to give 3ꢀ(6ꢀmethoxyꢀ1Hꢀindolꢀ3ꢀ yl)ꢀ
1
propionic acid ethyl ester 2ꢀ4 (800 mg, 34%) as white solid. HꢀNMR (600 MHz, d ꢀDMSO): δ
6
ACS Paragon Plus Environment