Novel Tetrahydropyrido[4,3-d]pyrimidines as Potent Inhibitors of Chaperone Heat Shock Protein 90

Article abstract of DOI:10.1021/acs.jmedchem.6b00912

Heat shock protein 90 (Hsp90) is a potential target for oncology therapeutics. Some inhibitors have shown antitumor effects in clinical trials, spurring the discovery of small molecule Hsp90 inhibitors. Here, we describe the structural optimization studies of a hit compound, tetrahydropyrido[4,3-d]pyrimidine-based Hsp90 inhibitor 15, which exhibits inhibitory activity against Hsp90. A series of analogues were synthesized, and their structure-activity and structure-property relationships were analyzed. These explorations led to the discovery of compound 73, which exhibited potent in vitro activities, good physicochemical properties, favorable ADME properties, and a potent antitumor effect in an HCT116 xenograft model. Furthermore, 73 exhibited no ocular toxicity in a rat retinal damage model, suggesting it is a relatively safe Hsp90 inhibitor. As a promising antitumor agent, 73 was progressed for further preclinical evaluation.

Full text of DOI:10.1021/acs.jmedchem.6b00912

Subscriber access provided by University of Otago Library
Article
Novel Tetrahydropyrido[4,3-d]pyrimidines as Potent
Inhibitors of Chaperone Heat Shock Protein 90
Fen Jiang, Hui-Jie Wang, Yu-Hui Jin, Qiong Zhang, Zhi-Hui Wang, Jian-Min Jia,
Fang Liu, Lei Wang, Qi-Chao Bao, Dong-Dong Li, Qi-Dong You, and Xiao-Li Xu
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00912 • Publication Date (Web): 09 Nov 2016
Downloaded from http://pubs.acs.org on November 9, 2016
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 67
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Novel Tetrahydropyrido[4,3ꢀd]pyrimidines as Potent Inhibitors of
Chaperone Heat Shock Protein 90
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a,b
a,b
a,b
a,b
a,b
a,b
Fen Jiang , HuiꢀJie Wang , YuꢀHui Jin , Qiong Zhang , ZhiꢀHui Wang , JianꢀMin Jia , Fang
a,b
a,b
a,b
a,b
a,b*
a,b*
Liu , Lei Wang , QiꢀChao Bao , DongꢀDong Li , QiꢀDong You and XiaoꢀLi Xu
a
State Key Laboratory of Natural Medicines, and Jiang Su Key Laboratory of Drug Design and
Optimization, China Pharmaceutical University, Nanjing 210009, China
b
Department of Medicinal Chemistry, School of Pharmacy, China Pharmaceutical University, Nanjing
2
10009, China
ABSTRACT
Heatꢀshock protein 90 (Hsp90) is a potential target for oncology therapeutics. Some inhibitors have
shown antiꢀtumor effects in clinical trials, spurring the discovery of smallꢀmolecule Hsp90 inhibitors.
Here,
we
describe
the
structural
optimization
studies
of
a
hit
compound,
tetrahydropyrido[4,3ꢀd]pyrimidineꢀbased Hsp90 inhibitor 15, which exhibits inhibitory activity against
Hsp90. A series of analogues were synthesized, and their structureꢀactivity relationships and
structureꢀproperty relationships were analyzed. These explorations led to the discovery of compound 73,
which exhibited potent in vitro activities, good physicochemical properties, favorable ADME properties
and potent antiꢀtumor effect in an HCT116 xenograft model. Furthermore, 73 exhibited no ocular
toxicity in a rat retinal damage model, suggesting it is a relatively safe Hsp90 inhibitor. As a promising
antiꢀtumor agent, 73 was progressed for further preꢀclinical evaluation.
INTRODUCTION
Heatꢀshock protein 90 (Hsp90), which is the most abundant intracellular mammalian protein, is crucial
1
,2
for the correct folding, function, and stability of a wide range of proteins called clients. More than 200
Hsp90 clients have been identified to date, including numerous receptors, kinases, and transcription
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 67
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
3
factors that are necessary for tumor development. Inhibiting the function of Hsp90 can lead to the
degradation of many oncogenic proteins, thereby shutting down multiple signaling transduction
4
,5
pathways in cancer cells and potentially resulting in wideꢀranging antiꢀcancer effects. Thus, Hsp90 has
been widely studied in the past decade as a promising protein target for the discovery of new drugs to
treat cancers.
0
1
2
3
4
5
6
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
Hsp90 is composed of three domains: a Nꢀterminal adenosine triphosphate/adenosine diphosphate
(ATP/ADP)ꢀbinding domain, a middle domain involved in client protein binding, and a Cꢀterminal
6
dimerization domain. The chaperone activity of Hsp90 depends on the dynamic conformational changes
7
of these domains, which are driven by nucleotide binding and hydrolysis. Small molecules can be
designed to occupy the ATPꢀbinding site to inhibit the function of Hsp90. To date, over 20 distinct
Nꢀterminal Hsp90 inhibitors have entered into clinical trials, and many other synthetic inhibitors are
8
−11
12
13
under preclinical evaluation.
Geldanamycin analogues, such as 17ꢀAAG (1) and IPI504 (2)
(Figure 1), entered into clinical trials and showed some activities, providing a proofꢀofꢀconcept for
1
4,15
Hsp90 inhibition in humans.
Despite their clinical effects, the development of geldanamycin
1
6
analogues was limited by formulation and toxicity issues. Hence, inhibitors with improved drugꢀlike
properties are needed, and as a result, synthetic smallꢀmolecule Hsp90 inhibitors are currently in clinical
1
7
development. These smallꢀmolecule Hsp90 inhibitors can be divided into two major classes according
to their structures (Figure 1). The purine class is designed based on the structure of ATP and includes
1
8
19
20
BIIB021 (3) , PUꢀH71 (4) , and MPCꢀ3100 (5) , etc. The resorcinol class is another important class of
2
1
22
23
Hsp90 inhibitors and includes STAꢀ9090 (6) , NVPꢀAUY922 (7) , and AT13387 (8) , etc. Some other
Hsp90 inhibitors that do not belong to these two major classes have also been developed, such as
2
4
25
SNXꢀ5422 (9) and NVPꢀHSP990 (10) . Up to now, a number of clinical trials have been conducted to
evaluate the antiꢀtumor effects of these inhibitors, and clinical efficacy has been observed in combined
therapy. In a phase II study, 1 was proven to be effective in patients with Herꢀ2ꢀpositive metastatic breast
2
6
cancer when used in combination with trastuzumab. The combination of 1 and bortezomib was also
2
7
active in multiple myeloma patients. In a phase I/II study, 8 showed encouraging activity in
combination with crizotinib in patients with nonꢀsmall cell lung cancer, and it may help overcome
ACS Paragon Plus Environment

Page 3 of 67
Journal of Medicinal Chemistry
1
2
3
2
8
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
crizotinib resistance. Current clinical developments of Hsp90 inhibitors mainly focus on identifying
2
9
which cancer types sensitive to an Hsp90 inhibitor and with which drug combination.
Despite the combined therapy efficacy, limited singleꢀagent activity was observed in clinical trials.
One possible reason for the monotherapyꢀfailure was that most inhibitors exhibited various adverse
effects that limited the clinical dosages. The firstꢀgeneration Hsp90 inhibitors, geldanamycin analogues,
0
1
2
3
4
5
6
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
3
0
suffered from serious hepatotoxicity because of the presence of a benzoquinone moiety. The
secondꢀgeneration synthetic inhibitors showed no hepatotoxicity, however, still suffered from some
1
6
other adverse effects, such as neurological toxicities and gastrointestinal side effects. This may be
31
partially due to the panꢀHsp90 inhibition. Hsp90 has four paralogs, including Hsp90α, Hsp90β, Grp94,
and Trapꢀ1. Reported results indicated that each paralog was responsible for chaperoning a distinct
3
2
spectrum of client proteins. Most of the Hsp90 inhibitors in clinical trials exhibited panꢀHsp90
inhibition and the lack of selectivity led to potential mechanismꢀbased toxicities. To reduce the potential
adverse effects, isoformꢀselective inhibitors have been developed, including the Hsp90α/β selective
3
3−35
inhibitors (such as 11) and Grp94 selective inhibitors (such as 12, 13).
These isoformꢀselective
inhibitors showed promising antiꢀcancer effects and had potential for further development. Ocular
toxicity was the most common adverse effect observed in the clinical trials of Hsp90 inhibitors, such as
3
6−38
7, 8, 9.
Though the ocular disorders were often reversible through doseꢀlimiting, the limited clinical
dosage would lead to insufficient drug exposure and reduced antiꢀtumor efficacy. Interestingly, ocular
disorders were not observed in the clinical trials of some other inhibitors, such as 1 and 6. Recent
research indicated that Hsp90 played a critical role in normal retinal function, and high drug exposure
3
9
level in retina led to structural damage of the retina. Thus, ocular toxicities of Hsp90 inhibitors can be
avoided by reducing the drug exposure level in retina. Reduced ocular toxicity would improve the
therapeutic index and benefit to the clinical development of Hsp90 inhibitors.
Different core structures would contribute to improved efficacy and safety profile, so a study aimed
to identify novel potent and safe Hsp90 inhibitors was conducted. In our previous study, a Rapid
Overlay of Chemical Structures (ROCS) model was constructed and applied to virtually screen the
Topscience database. Compound 14, containing the 5,6,7,8ꢀtetrahydropyrido[4,3ꢀd]pyrimidine core,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 67
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
displayed the most potent Hsp90 inhibition (IC , 45.39 ꢁM) in the fluorescence polarization (FP) assay.
5
0
Preliminary structural modification led to compound 15, which exhibited greater than
4
0
450ꢀfoldꢀimproved Hsp90 inhibition activity (IC , 0.101 ꢁM) compared to 14. Here, we report further
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
optimization studies centered on compound 15. Investigating the structureꢀactivity relationships and
structureꢀproperty relationships led to the identification of compound 73, which showed potent
antiꢀtumor efficacy in an HCT116 human colon carcinoma xenograft model and other favorable
properties. No ocular toxicity in a rat retinal damage model indicated the good safety profile of 73.
Figure 1. Some representative Hsp90 inhibitors and compounds 14 and 15
CHEMISTRY
Derivatives 24−28 were synthesized using the route summarized in Scheme 1. The 5ꢀchlorineꢀ and
5
1
1
ꢀbromineꢀsubstituted benzoic acid intermediates 17a and 17b were prepared from commercial 16a and
6b via benzyl protection of the phenyl groups and subsequent benzyl ester hydrolysis. The intermediate
4
1
7c was prepared according to previously reported procedures. Briefly, FriedelꢀCrafts acylation of the
commercial 18 was conducted in the presence of boron trifluoride, and subsequent protection of the
ACS Paragon Plus Environment

Page 5 of 67
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
phenol groups with benzyl bromide gave 19. Then Wittig reaction was used to obtain the olefin
intermediate, and subsequent hydrolysis of the methyl ester yielded 17c.
NꢀBocꢀ3ꢀPyrrolidinone (20a) or NꢀBocꢀ4ꢀpiperidone (20b) was refluxed in N,Nꢀdimethylformamide
dimethyl acetal to obtain 21a or 21b and then cyclized with acetamidine hydrochloride in the presence
of sodium acetate in refluxed EtOH. Subsequently, the Boc group was deprotected in trifluoroacetic acid
(TFA) to obtain the key intermediates 22a or 22b, which condensed with the substituted benzoic acid
0
1
2
3
4
5
6
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
17a−c
using
standard
1ꢀethylꢀ3ꢀ(3ꢀdimethylaminopropyl)carboiimide/hydroxybenzotriazole
(EDCI/HOBt) coupling conditions to give 23a−e. The benzyl protecting groups in 23a−c were
deprotected using BBr to obtain the target compounds 24−26. For 23d and 23e, Pd/C catalytic
3
hydrogenation simultaneously removed the benzyl protecting groups and reduced the isopropenyl group
to the isopropyl to give 27 and 28.
a
Scheme 1. Synthetic route of compounds 24−28
a
Reagents and conditions: (a) BnBr, K CO , acetone, reflux, 8 h; (b) NaOH, MeOH/H O, 70 °C, 3 h; (c) Ac O,
2
3
2
2
BF ·OEt , 75 ºC, 3 h; (d) MePPh Br, KOC(CH ) , tetrahydrofuran (THF), room temperature (rt), 1 h; (e)
3
2
3
3 3
N,Nꢀdimethylformamide dimethyl acetal, reflux, 1.5 h; (f) acetamidine hydrochloride, Et N, EtOH, reflux, 42 h; (g)
3
trifluoroacetic acid (TFA), 0 °C to rt, 1 h; (h) substituted benzoic acid (17a, 17b, and 17c), EDCI, HOBt, Et N,
3
dichloromethane (DCM), rt, 24 h; (i) BBr , DCM, 0 ºC, 3 h; (j) 10% Pd/C, H , MeOH, rt, 4 h.
3
2
Derivatives 35−65 were synthesized using the route described in Scheme 2. Cyclization of 29 with
urea in the presence of sodium methanolate gave intermediate 30. Subsequent chlorination using
phosphorus oxychloride provided 31. Removal of the 4ꢀchloro atom on the pyrimidine ring using
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 67
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
activated zinc powder produced 32. The benzyl was removed using 1ꢀchloroethyl chloroformate
(ACEꢀCl) to give 33 and then coupled with 17c in the presence of EDCI/HOBt to form the key
intermediate 34. Substitution of the 2ꢀchloro group with different amines or alcohols and subsequent
hydrogenation afforded the target compounds.
0
1
2
3
4
5
6
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
Scheme 2. Synthetic route of compounds 35−65
a
Reagents and conditions: (a) urea, MeONa, EtOH, reflux, 24 h; (b) POCl , N , 3 h; (c) activated zinc powder, NH OH,
3
2
4
EtOH, reflux, 12 h; (d) ACEꢀCl, DCM 0 °C to rt, then reflux, 10 h; (e) MeOH, reflux, 1 h; (f) 17c, EDCI, HOBt, Et N,
,
3
DCM, rt, 24 h; (g) different amines, K CO , dioxane, 85 ºC, 24 h; (h) different alcohols, NaH, THF, rt, 10 h; (i) 10%
2
3
Pd/C, H , MeOH, rt, 4 h.
2
Derivatives 68−78 were synthesized using the route described in Scheme 3. The 4ꢀchloro atom on
the pyrimidine ring of intermediate 31 was substituted with different alcohols and amines, and
subsequent substitution of the 2ꢀchloro atom of the pyrimidine ring with 2ꢀmethoxyethoxy gave the key
intermediates 66a−k. The benzyl group was removed using Pd(OH) under a hydrogen atmosphere, and
2
the resulting amines were converted into the amide intermediates 67a−k via EDCI/HOBtꢀmediated
coupling with the intermediate 17c. Hydrogenation in the presence of 10% Pd/C yielded the desired
compounds 68−78.
a
Scheme 3. Synthetic route of compounds 68−78
ACS Paragon Plus Environment

Page 7 of 67
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
Reagents and conditions: (a) different alcohols, NaH, THF, rt to 0 ºC, 0.5 h; (b) different amines, N,
Nꢀdiisopropylethylamine (DIPEA), THF, reflux, 24 h; (c) 2ꢀmethoxyethanol, NaH, THF, reflux, 1 h; (d) Pd(OH) , H ,
2
2
MeOH, rt, 6 h; (e) 17c, EDCI, HOBt, Et N, DCM, rt, 24 h; (f) 10% Pd/C, H , MeOH, rt, 4 h.
3
2
RESULTS AND DISCUSSION
Preliminary SAR study lead to compound 50 with potent activities
In our previous study, compound 15 was identified to be a potent Hsp90 inhibitor (FP IC , 0.101
5
0
4
0
ꢁM). Docking analysis showed that compound 15 inserted deep into the ATP binding pocket of Hsp90
(Figure 2). The resorcinol moiety and carbonyl group formed tight network of hydrogen bonds with
residues Asp93, Thr184 and conserved water molecules, which anchored the compound into the ATP
pocket. The 5’ꢀchlorine atom on the resorcinol moiety incompletely occupied a hydrophobic subꢀpocket
consisting of Leu107, Phe138, Val150 and Val186, and with potential for further optimization.
Relatively hydrophobic groups could occupy the pocket more snugly and improve the affinity to Hsp90.
SAR study at 5’ꢀposition of resorcinol family Hsp90 inhibitors has been well established in previous
study. For instance, during the discovery process of 7 and 8, isopropyl was demonstrated to be the most
22,23
suitable substituent at this position.
The isopropyl occupied the hydrophobic pocket snugly and
resulted in significant improved Hsp90 inhibition and antiꢀproliferative activities. As reference, to
improve the activities of 15, our SAR study firstly focused on the 5’ꢀposition of the resorcinol.
Derivatives of compound 15 were designed in which the chlorine atom was replaced by bromine atom or
isopropyl group. Inspired by the isoindoline core of 8, the tetrahydropyrido[4,3ꢀd]pyrimidine core was
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 67
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
converted into 6,7ꢀdihydroꢀ5Hꢀpyrrolo[3,4ꢀd]pyrimidine to investigate the influence of the ring size on
2
3
activities. These investigations resulted in target compounds 24−28.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Docking analysis of the binding mode between Hsp90 and compound 15 (protein structure from PDB
23
2
XJX). (A) Compound 15 inserts into the active site of Hsp90 with a surface colored in terms of the hydrophobic
state. (B) The detailed interaction between Hsp90 and compound 15. The carbon atoms of compound 15 and the
residues in the Hsp90’s active site are shown in yellow and green, respectively. Conserved water molecules are
presented as red spheres, and hydrogen bonds are indicated by magenta dashed lines.
The Hsp90ꢀinhibition activities of the synthesized derivatives were determined using FP assay.
Antiꢀproliferative effects were also evaluated against two cancer cell lines (HCT116 human colon
carcinoma cell line and MCFꢀ7 human breast adenocarcinoma cell line). As shown in Table 1,
substituting the chlorine atom on the resorcinol moiety with a bromine atom had no significant effects
on the biochemical potency, whereas the isopropyl analogue 27 showed 4ꢀfold more potent activity than
compound 15 in the FP assay. The results indicated that an isopropyl group was the optimal substituent
for the R position, and was similar to the group at this position in clinical resorcinol family inhibitors.
1
The analogues 24, 26, and 28, in which the tetrahydropyrido[4,3ꢀd]pyrimidine core was changed into
pyrrolo[3,4ꢀd]pyrimidine, showed decreased biochemical potencies relative to 15, 25, and 27. Thus, the
tetrahydropyrido[4,3ꢀd]pyrimidine moiety was retained as the molecular core for further optimization.
In this series, compounds 27 and 28 showed comparable Hsp90 inhibition activities with the
reference compound 8, while the antiꢀproliferative activities were much weaker. Permeability is an
important factor that can impact cellular effects of compounds. Results showed that 27 and 28 exhibited
−
6
−6
over 2ꢀfold lower permeability (33.33 × 10 cm/s and 27.03 × 10 cm/s respectively, shown in Table S1)
ACS Paragon Plus Environment

Page 9 of 67
Journal of Medicinal Chemistry
1
2
3
−
6
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
than that of the reference compound 8 (74.32 × 10 cm/s). In consideration of the 2ꢀ or 3ꢀfold lower
enzyme potency, lower permeability should partially lead to the weaker cellular activities of 27 and 28.
Table 1. SAR exploration of the R position and the core
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
Antiꢀproliferative activities
a
a
Compd
n
R1
FP IC (ꢁM)
IC50 (ꢁM)
50
HCT116
2.83 ± 0.92
10.66 ± 0.54
1.53 ± 1.24
1.55 ± 0.35
1.15 ± 0.03
1.63 ± 0.31
0.07 ± 0.02
MCFꢀ7
1
2
5
4
2
1
Cl
Cl
0.101 ± 0.012
0.474 ± 0.203
0.132 ± 0.010
0.262 ± 0.015
0.026 ± 0.011
0.038 ± 0.014
0.013 ± 0.002
1.91 ± 1.21
5.48 ± 2.36
1.73 ± 1.03
3.72 ± 0.64
1.48 ± 0.35
2.14 ± 0.55
0.02 ± 0.01
25
2
Br
2
2
6
7
1
Br
2
isopropyl
isopropyl
—
28
8
1
—
a
Values shown are the mean ± SD (n = 3).
Based on the docking analysis of compound 15, the methyl group at the 2ꢀposition of the pyrimidine
ring was exposed to the solvent area, and substitution at this position would be tolerated. Thus, various
substituents were investigated to elucidate the SARs of this position, leading to target compounds
3
5−58.
Various amines and alcohols were introduced to give derivatives 35−58. Table 2 summarized the
SARs of these compounds. The amine substituted compounds 35−48 showed decreased Hsp90ꢀbinding
affinity with respect to 27. Among the primary amine substituted analogues, 35 incorporating a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 67
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
methylamine substituent possessed the best activity in FP assay. The activities were significantly
reduced when bigger groups were introduced. Similar trends were observed in the analogues bearing
secondary amines. Compared with the amine substituted compounds, 49−58 introduced alcohol
substituents showed higher activities in FP assay, with IC values ranging from 0.056 ꢁM to 0.103 ꢁM.
0
1
2
3
4
5
6
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
These results suggested that alcohol substituents were more suitable than amine substituents at this
position.
Cellular effects of this series showed similar change tendency with the FP data. Most of the amine
substituted compounds exhibited decreased antiꢀproliferative activities, except 35 and 39 which
incorporating small amine substituents. Despite weaker affinity to Hsp90 compared with 27, most of the
alcohol substituted analogues showed improved or retained cellular effects. Among which, 50 with an
ethoxyl group possessed the best antiꢀproliferative activities. Permeability determination showed that
−
6
compound 50 (72.41 × 10 cm/s) exhibited a 2ꢀfold increased permeability compared with that of 27
−
6
(33.33 × 10 cm/s). The increased permeability should contribute to the improved cellular effects.
Table 2. Preliminary exploration of the R position
2
a
Antiꢀproliferative activities IC (ꢁM)
5
0
a
Compd
R2
FP IC (ꢁM)
50
HCT116
MCFꢀ7
3
3
5
6
0.113 ± 0.004
0.167 ± 0.008
0.238 ± 0.007
0.222 ± 0.003
0.164 ± 0.009
0.74 ± 0.07
1.56 ± 0.48
2.34 ± 0.64
2.43 ± 0.38
1.02 ± 0.17
1.25 ± 0.15
1.58 ± 0.11
2.37 ± 0.13
2.68 ± 0.35
1.39 ± 0.06
37
38
39
ACS Paragon Plus Environment

Page 11 of 67
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
4
4
4
0
1
2
0.201 ± 0.003
0.370 ± 0.020
0.343 ± 0.015
0.113 ± 0.007
0.172 ± 0.003
0.110 ± 0.009
0.271 ± 0.004
0.191 ± 0.003
4.42 ± 1.27
2.16 ± 1.00
3.09 ± 1.83
1.12 ± 0.14
4.55 ± 1.20
4.46 ± 1.05
1.82 ± 0.06
2.18 ± 0.57
1.43 ± 0.66
5.70 ± 1.04
1.29 ± 0.02
1.14 ± 0.28
1.55 ± 0.03
1.47 ± 0.55
1.71 ± 0.05
1.65 ± 0.06
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
43
44
45
4
6
7
4
4
8
0.270 ± 0.012
1.60 ± 0.04
3.62 ± 1.33
49
50
51
52
53
54
0.075 ± 0.015
0.056 ± 0.002
0.060 ± 0.004
0.065 ± 0.005
0.088 ± 0.001
0.065 ± 0.001
0.071 ± 0.007
0.103 ± 0.024
0.056 ± 0.006
1.16 ± 0.17
0.76 ± 0.09
1.08 ± 0.21
1.24 ± 0.05
1.27 ± 0.06
1.23 ± 0.08
1.72 ± 0.13
2.78 ± 0.63
0.60 ± 0.02
0.63 ± 0.08
0.62 ± 0.05
1.34 ± 0.16
0.96 ± 0.10
0.88 ± 0.24
1.33 ± 0.04
1.19 ± 0.11
1.87 ± 0.45
0.82 ± 0.05
55
56
57
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 67
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
8
0.076 ± 0.006
1.70 ± 0.09
1.79 ± 0.34
a
Values shown are the mean ± SD (n = 3).
To accurately analyze the binding modes of the compounds and guide further optimization, 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
crystal structure of 50 was obtained in complex with the Hsp90α Nꢀterminal. As shown in Figure 3A−B,
the crystal structure revealed that the resorcinol moiety bound deep into the ATPꢀbinding pocket of
Hsp90. The 2ꢀhydroxyl and carbonyl group of the amide formed a tight network of hydrogen bonds with
the Asp93 and Thr184 residues and a conserved water molecule. The 4ꢀhydroxyl group formed a
hydrogen bond with another conserved water molecule. This tight network of hydrogen bonds ensured
high binding affinity between the molecule and the protein. The isopropyl moiety at the 5’ꢀposition of
the resorcinol moiety was efficiently embedded within a large hydrophobic cavity, bound by the Leu48,
Met98, Leu107, Phe138, Val150, and Val186 residues.
Interestingly, the tetrahydropyrido[4,3ꢀd]pyrimidine core of the molecule bound to the protein in two
different conformations (referred as conformation A/B), possibly because of the conformational
flexibility of the tetrahydropyrido ring. In order to understand the different binding modes of
conformations A and B better, we obtained the overlaps of Hsp90ꢀ50 complex and Hsp90ꢀ8 complex. As
shown in Figure 3C, conformation A of compound 50 was well overlapped with 8, and the two
compounds adopt same binding modes with the protein. Both the tetrahydropyrido[4,3ꢀd]pyrimidine
core in 50 and the isoindoline core in 8 were adjacent to the H2 helix of the protein and formed
hydrophobic interactions with residues Ala55, Lys58, Ile96, and Met98. The ethoxyl in 50 inserted into
the solvent area and bound nonꢀspecifically, just like the methylpiperazine moiety in 8. Correspondingly,
the protein adopted an openꢀloop conformation which was similar to that of the
42
1
7ꢀdimethylaminoethylaminoꢀ17ꢀdemethoxygeldanamycin (17ꢀDMAG) bound conformation. When
5
0 adopted conformation B, it possessed different binding modes with 8 (Figure 3D). The
tetrahydropyrido[4,3ꢀd]pyrimidine core along with the ethoxyl group induced rearrangement of the
flexible sequence 104−111 in the protein and formed extended hydrophobic interactions with rearranged
residues Leu107 and Ile110. Protein conformational changes were observed in the proteinꢀcompound
43
crystal complexes of some other Hsp90 inhibitors, such as 3. In the two conformations, the ethoxyl
ACS Paragon Plus Environment

Page 13 of 67
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
group at the 2ꢀposition of the pyrimidine ring exposed to the solvent surface of the protein, with
potential for further optimization.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Xꢀray structure of compound 50 bound to the active site of Hsp90 (PDB: 5GGZ). (A) Compound 50 adopts
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 67
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
conformation A to bind with the openꢀloop conformation of Hsp90. (B) Compound 50 adopts conformation B to bind
with the extended αꢀhelix conformation of Hsp90. The active site of Hsp90 is surfaced according to the hydrophobic
state, and detailed interaction is shown. The carbon atoms of compound 50 and the residues in the active site of Hsp90
are yellow and green, respectively. Conserved water molecules are shown as red spheres, and hydrogen bonds are
indicated by magenta dashed lines. The omit F – F electron density for 50 is shown at the 3.0 σ contour level. (C)
o
c
0
1
2
3
4
5
6
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
Overlaps of Hsp90ꢀ50 complex and Hsp90ꢀ8 complex when 50 adopts conformation A. (D) Overlaps of Hsp90ꢀ50
complex and Hsp90ꢀ8 complex when 50 adopts conformation B. The protein in Hsp90ꢀ50 complex and Hsp90ꢀ8
complex are colored as green and grey, respectively. Compound 50 is colored yellow and compound 8 is colored
magenta.
Lead optimization to improve activities and drugꢀlike properties
Though the lead compound 50 displayed high binding affinity to Hsp90 and potent antiꢀproliferative
activities, its biological potency was still much weaker than that of the clinical compounds, such as
2
2,23
compounds 7 and 8.
In addition, the solubility of 50 was relatively low (66 ꢁg/mL). In consideration
that resorcinol class Hsp90 inhibitors were intravenously administrated in clinical trials, higher
solubility should be favorable. Hence, further optimization should focus on both the biological activities
and physicochemical properties of these compounds. Based on the Xꢀray structural analysis of
compound 50, the ethoxyl group at the 2ꢀposition of the pyrimidine ring was exposed to the solvent area,
indicating this position should be suitable for introducing solubilizing groups. Thus, compounds that
incorporated polar and flexible moieties as solubilizing functionalities were designed and synthesized.
Various substituted alcohols were introduced to the 2ꢀposition of the pyrimidine ring afforded
analogues 59−65. Data concerning the biochemical activities and physicochemical properties of the
analogues were summarized in Table 3. All compounds of this series exhibited FP IC values lower than
5
0
30 nM, indicating higher affinity to Hsp90 than that of 50. Compound 59 (2ꢀmethoxyethoxy group)
showed significantly improved antiꢀproliferative activities with respect to 50, with IC values of 0.17
5
0
ꢁM in HCT116 cells and 0.12 ꢁM in MCFꢀ7 cells. Unfortunately, the other compounds (60−65) in this
series showed decreased cellular activities. The inconsistency between the FP activities and
antiꢀproliferative activities may be attributable to the unsuitable physicochemical properties of these
compounds, such as low permeability.
To explain this inconsistency, physicochemical properties of these compounds were determined. The
partition coefficient (CLogP) values were calculated using ChemBioDraw software (CambridgeSoft).
ACS Paragon Plus Environment

Page 15 of 67
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
Distribution coefficient (Log D ) and intrinsic aqueous solubility were obtained using a Gemini
7.4
Profiler instrument (Pion Inc., MA, USA) by the “gold standard” Avdeef−Bucher potentiometric
4
4
titration method. Permeability was determined using the doubleꢀsink parallel artificial membrane
permeability assay (PAMPA) on a PAMPA Explorer instrument (Pion Inc., MA, USA). As shown in
Table 3, the most potent compound (59) showed the most favorable Log D (2.14) and membrane
0
1
2
3
4
5
6
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
7
.4
−
6
permeability (55.55 × 10 cm/s). Additionally, 59 showed 2ꢀfold improved intrinsic solubility compared
to that of the lead compound 50. The favorable physicochemical properties of compound 59 indicated
that it possessed balanced hydrophilic and hydrophobic properties. Compounds 60−65, which had more
polar basic substituents, exhibited improved intrinsic solubility but obviously decreased Log D values
7
.4
−
6
ranging from 1.04 to 1.48 and correspondingly lower permeability ranging from 3.06 × 10 cm/s to
−6
30.25 × 10 cm/s. The lower Log D7.4 and membrane permeability should lead to the inconsistent
activities of compounds 60−65 between enzymatic and cellular levels.
Table 3. Further exploration of the R position
2
Antiꢀproliferative activities
b
Physicochemical properties
a
IC50 (ꢁM)
a
Compd
R2
FP IC50 (ꢁM)
a
Log D Solubility
Pe, pH 7.4
HCT116
MCFꢀ7
CLogP
−
6
pH 7.4
2.95
(ꢁg/mL)
66.0
(10 cm/s)
72.41 ± 5.36
55.55 ± 3.56
3.25 ± 0.17
25.70 ± 2.71
5
5
6
6
0
9
0
1
0.056 ± 0.002
0.019 ± 0.001
0.030 ± 0.005
0.032 ± 0.004
0.76 ± 0.09
0.17 ± 0.04
2.38 ± 0.19
0.89 ± 0.09
0.62 ± 0.05
0.12 ± 0.01
1.58 ± 0.05
2.90 ± 0.83
2.70
2.06
2.27
3.32
2.14
139.8
148.2
151.4
1.04
1.17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 67
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
6
6
6
6
2
3
4
5
0.036 ± 0.004
0.029 ± 0.004
0.030 ± 0.002
0.028 ± 0.006
0.92 ± 0.17
0.98 ± 0.19
0.78 ± 0.09
2.37 ± 0.05
1.93± 0.08
2.33 ± 0.82
1.23 ± 0.14
2.62
2.94
3.50
2.29
1.09
1.12
1.21
1.48
155.6
145.5
174.6
139.9
4.15 ± 0.12
3.06 ± 0.11
20.46 ± 2.20
30.25 ± 3.53
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
1.06 ± 0.10
a
b
Values shown are the mean ± SD (n = 3). The partition coefficient (CLogP) values were calculated using
ChemBioDraw software (CambridgeSoft). The distribution coefficient (Log D ) and intrinsic aqueous solubility were
7.4
obtained using a Gemini Profiler instrument (Pion Inc., MA, USA) by the “gold standard” Avdeef−Bucher
potentiometric titration method. Permeability was determined using the doubleꢀsink parallel artificial membrane
permeability assay (PAMPA) on a PAMPA Explorer instrument (Pion Inc., MA, USA).
According to the Xꢀray structural analysis of compound 50, the 4ꢀposition of the pyrimidine ring is
surrounded by the hydrophobic side chains of the Lys58, Ile96 and Met98 residues, and hydrophobic
substituents at this position may facilitate improved activities. Thus, a series of compounds with
substituents introduced at the 4ꢀposition of the pyrimidine ring were investigated. Table 4 summarized
the activities and physicochemical properties of compounds 68−78. Alkoxyl groups were introduced to
yield compounds 68−71. Compounds 68−70 bearing small hydrophobic residues (methoxyl, ethyoxyl,
and nꢀpropoxyl group) retained the Hsp90 affinity and cellular activities. Compound 70 (nꢀpropoxyl
group) showed better cellular activities than compound 59, possibly because of its higher permeability
−6
(
87.71 × 10 cm/s). Compound 71 (isopropoxyl group) showed slightly decreased activities in the FP
and antiꢀproliferative assays relative to those of 68−70. Alkylamino moieties were introduced to yield
2−78. Compounds 72−77 retained their activities at enzymatic and cellular levels, among which 73
exhibited the best activities with IC values of 0.14 ꢁM in HCT116 cells and 0.10 ꢁM in MCFꢀ7 cells.
7
5
0
Compound 78 (IC , 0.102 ꢁM) with a diethylamino substitution exhibited markedly decreased potency
5
0
with respect to the dimethylamino analogue 77 (IC , 0.016 ꢁM) in the FP assay. Indeed, the
5
0
diethylamino group was too big for the hydrophobic pocket around the 4ꢀposition of the pyrimidine ring.
Correspondingly, the cellular effects of 78 were dramatically decreased.
Physicochemical properties determination of this series revealed that compounds 68−77 possessed
acceptable CLogP values ranging from 2.99 to 4.05, Log D values ranging from 2.21 to 3.84, and
7
.4
ACS Paragon Plus Environment

Page 17 of 67
Journal of Medicinal Chemistry
1
2
3
−
6
−6
45
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
permeability ranging from 56.35 × 10 cm/s to 87.71 × 10 cm/s. Alkoxyl substituted compounds
68−71 possessed decreased solubility compared with 59. Most of the alkylamino bearing compounds
showed improved or retained solubility, except for 74. In this series, compound 73 exhibited the most
favorable physicochemical properties.
0
1
2
3
4
5
6
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 4. R expansion
3
Antiꢀproliferative activities
b
Physicochemical properties
a
IC50 (ꢁM)
a
Compd
R3
FP IC50 (ꢁM)
a
Log D
pH 7.4
2.14
2.56
2.74
3.59
3.84
2.21
2.32
2.92
2.98
2.45
Solubility
(ꢁg/mL)
139.8
93.5
Pe, pH 7.4
HCT116
MCFꢀ7
CLogP
−6
(10 cm/s)
55.55 ± 3.56
75.75 ± 3.17
77.26 ± 4.69
87.71 ± 5.45
79.43 ± 3.37
57.50 ± 6.86
71.47 ± 1.68
78.67 ± 12.10
62.44 ± 2.86
56.35 ± 7.13
5
9
H
0.019 ± 0.001
0.030 ± 0.001
0.026 ± 0.002
0.036 ± 0.011
0.058 ± 0.008
0.017 ± 0.001
0.028 ± 0.002
0.032 ± 0.002
0.030 ± 0.003
0.019 ± 0.002
0.17 ± 0.04
0.25 ± 0.05
0.32 ± 0.06
0.12 ± 0.03
0.74 ± 0.07
0.27 ± 0.01
0.14 ± 0.02
0.40 ± 0.06
0.79 ± 0.05
0.23 ± 0.02
0.12 ± 0.01
0.14 ± 0.03
0.18 ± 0.02
0.04 ± 0.01
0.25 ± 0.04
0.29 ± 0.03
0.10 ± 0.01
0.35 ± 0.11
0.25 ± 0.03
0.16 ± 0.01
2.06
2.99
3.52
4.05
3.83
2.88
3.40
3.93
3.71
3.46
68
6
7
9
0
86.1
73.6
71
40.2
7
7
2
3
177.7
192.3
114.3
137.4
157.0
74
7
7
5
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 67
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
7
7
0.016 ± 0.001
0.102 ± 0.003
0.26 ± 0.02
3.57 ± 0.04
0.33 ± 0.12
2.89 ± 0.35
2.96
4.02
2.35
3.08
152.7
95.9
82.96 ± 7.07
107.14 ± 9.65
78
a
b
Values shown are the mean ± SD (n = 3). The partition coefficient (CLogP) values were calculated using
ChemBioDraw software (CambridgeSoft). The distribution coefficient (Log D ) and intrinsic aqueous solubility were
0
1
2
3
4
5
6
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
7
.4
obtained using a Gemini Profiler instrument (Pion Inc., MA, USA) by the “gold standard” Avdeef−Bucher
potentiometric titration method. Permeability was determined using the doubleꢀsink parallel artificial membrane
permeability assay (PAMPA) on a PAMPA Explorer instrument (Pion Inc., MA, USA).
Hsp90α selectivity versus Grp94 of selected compounds
Reported results indicated that flexible sequence 104−111 on Hsp90α/β could adopt a
compoundꢀinduced rearrangement and form an extended αꢀhelix conformation. The equivalent sequence
in Grp94 and Trapꢀ1 were disordered and unable to accommodate the extended αꢀhelix conformation.
This unique conformation rearrangement was chemotypeꢀdependent and believed to contribute to the
4
3
Hsp90α/β selective inhibition. The clinical compounds, including 3 and 9, induced conformation
rearrangement of sequence 104−111 on Hsp90α/β and exhibited moderate selectivity versus Grp94 and
Trapꢀ1. However, compound 7 which adopted similar bound conformation with the four isoforms,
showed no isoform selectivity. The crystal structure of 50 with Hsp90α revealed that the
tetrahydropyrido ring in the molecule could adopt two different conformations (conformation A/B) to
bind to the protein. When the tetrahydropyrido ring adopt conformation B, the flexible sequence
1
04−111 on Hsp90α was induced to form the extended αꢀhelix conformation, indicating the compound
might show some Hsp90α selectivity. Thus, the Hsp90α selectivity versus Grp94 of some selected
compounds were determined. The results were shown in Table 5.
Compound 27 with a tetrahydropyrido[4,3ꢀd]pyrimidine core, showed a 10ꢀfold selectivity.
Correspondingly, compound 28 with a reduced ringꢀsize core showed no selectivity, indicating that the
conformation flexibility of the tetrahydropyrido[4,3ꢀd]pyrimidine core contributed to the Hsp90α
selective inhibition. Substitution at R position were tolerated for the Hsp90α selectivity, compounds
2
4
9−54 and 59−61 showed maintained selectivity (6.6ꢀfold to 10.9ꢀfold). The influence of substituents at
R position on selectivity was also investigated. Results showed that analogues 69−71 and 73−75
3
bearing relatively big substituents possessed improved selectivity (over 20ꢀfold), suggesting that
ACS Paragon Plus Environment

Page 19 of 67
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
hydrophobic substituents at R position were favorable for the compound adopting conformation B.
3
Table 5. Hsp90α selectivity versus Grp94
0
1
2
3
4
5
6
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
Hsp90α
Grp94
Fold
Compd
n
R2
R3
a
a
FP IC (ꢁM)
FP IC (ꢁM)
Grp94/ Hsp90α
5
0
50
2
2
7
8
2
1
2
2
2
2
2
2
2
2
2
2
2
CH
CH
3
3
H
H
H
H
H
H
H
H
H
H
H
0.026 ± 0.011
0.038 ± 0.014
0.075 ± 0.015
0.056 ± 0.002
0.060 ± 0.004
0.065 ± 0.005
0.088 ± 0.001
0.065 ± 0.001
0.019 ± 0.001
0.030 ± 0.005
0.032 ± 0.004
0.030 ± 0.001
0.026 ± 0.002
0.245 ± 0.016
0.050 ± 0.003
0.498 ± 0.007
0.561 ± 0.016
0.498 ± 0.009
0.518 ± 0.015
0.670 ± 0.017
0.706 ± 0.005
0.198 ± 0.005
0.290 ± 0.013
0.217 ± 0.004
0.232 ± 0.006
0.608 ± 0.015
9.4
1.3
49
50
51
6.6
10.0
8.3
52
53
54
59
60
61
68
69
8.0
7.6
10.9
10.4
9.7
6.8
7.7
23.4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 67
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
7
7
7
7
0
1
2
3
2
2
0.036 ± 0.011
0.058 ± 0.008
0.017 ± 0.001
0.028 ± 0.002
0.032 ± 0.002
0.030 ± 0.003
0.019 ± 0.002
0.016 ± 0.001
0.102 ± 0.003
0.929 ± 0.009
1.316 ± 0.018
0.196 ± 0.006
0.729 ± 0.018
0.796 ± 0.009
0.787 ± 0.008
0.216 ± 0.003
0.195 ± 0.006
1.100 ± 0.016
0.022 ± 0.007
25.8
22.7
11.5
26.0
24.9
26.2
11.4
12.2
10.8
1.7
0
1
2
3
4
5
6
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
2
74
75
76
2
2
2
7
7
7
8
2
2
8
—
—
—
0.013 ± 0.002
a
Values shown are the mean ± SD (n = 3).
In vitro antiꢀcancer effects
The antiꢀproliferative activities of compounds 59, 70 and 73 were evaluated using a panel of human
cancer cell lines from diverse tissue origins. As summarized in Table 6, the three compounds exhibited
broadꢀspectrum activities and potently inhibited cell proliferation, with IC values ranging from 0.018
5
0
to 0.434 ꢁM. To evaluate the selective antiꢀproliferative effects of the compounds on cancer cells, we
determined their effects on normal human liver cells (Lꢀ02). Despite the potent antiꢀproliferative effects
against various cancer cell lines, the three compounds showed much lower cytotoxicity on Lꢀ02 cells.
The compounds’ good cellular selectivity supporting a potentially safer in vivo profile.
Table 6. Antiꢀproliferative effects of 59, 70 and 73 in a panel of human cancer cell lines and normal
human liver cells (Lꢀ02)
a
Cell line
Tissue
Antiꢀproliferative activities IC (ꢁM)
50
ACS Paragon Plus Environment

Page 21 of 67
Journal of Medicinal Chemistry
1
2
3
4
5
6
5
9
70
73
8
MDAꢀMBꢀ231 Breast carcinoma
MDAꢀMBꢀ468 Breast carcinoma
0.129 ± 0.023 0.122 ± 0.015 0.119 ± 0.007 0.038 ± 0.004
0.172 ± 0.033 0.139 ± 0.025 0.135 ± 0.015 0.066 ± 0.009
0.175 ± 0.056 0.220 ± 0.033 0.243 ± 0.025 0.071 ± 0.017
0.183 ± 0.034 0.178 ± 0.032 0.195 ± 0.024 0.024 ± 0.004
0.144 ± 0.016 0.118 ± 0.019 0.095 ± 0.017 0.032 ± 0.007
0.107 ± 0.067 0.171 ± 0.030 0.052 ± 0.012 0.067 ± 0.010
0.325 ± 0.073 0.407 ± 0.029 0.434 ± 0.093 0.129 ± 0.009
0.168 ± 0.002 0.060 ± 0.014 0.194 ± 0.017 0.050 ± 0.006
0.119 ± 0.037 0.222 ± 0.024 0.153 ± 0.009 0.018 ± 0.005
0.109 ± 0.003 0.150 ± 0.009 0.119 ± 0.005 0.038 ± 0.005
0.031 ± 0.001 0.018 ± 0.002 0.027 ± 0.003 0.011 ± 0.003
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
SKBr3
T47D
Breast carcinoma
Breast carcinoma
BT474
U2OS
MKN28
A549
Breast carcinoma
Human osteosarcoma
Human gastric carcinoma
Human pulmonary carcinoma
Hepatocellular carcinoma
Chronic myeloid leukemia
Acute monocytic leukemia
Normal human liver cells
HepG2
K562
MV4ꢀ11
Lꢀ02
27.62 ± 3.02
30.94 ± 3.91
33.95 ± 7.33
11.64 ± 5.04
a
Values shown are the mean ± SD (n = 3).
To further confirm the mechanisms underlying the potent antiꢀproliferative activities, effects of the
three compounds on cellular markers were investigated. As expected, Western blot analysis after 24 h of
exposure to various concentrations of the compounds indicated the depletion of the Hsp90 client
proteins Akt, Erk, and phosphorylated Erk (pꢀErk) and significant induction of Hsp70 in a
doseꢀdependent manner. What’s more, Hsp90 was slightly induced. The results confirmed that these
compounds inhibit cell growth by an Hsp90ꢀinhibitionꢀdependent mechanism.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 67
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 4. (A) Western blot analysis of the biomarker levels in HCT116 cells after incubation for 24 h with 59, 70 and
73. (B) Densitometry analysis of the protein levels are showed as normalized (to actin) ratios. Actin is used as the
ACS Paragon Plus Environment

Page 23 of 67
Journal of Medicinal Chemistry
1
2
3
control for protein loading. 8 is used as the reference compound.
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In vitro metabolic stability, CYP450 inhibition, in vivo pharmacokinetics and tissue distribution
Preliminary ADME properties assessments of 59, 70, 73 were conducted to evaluate the potential for
further development. The in vitro metabolic stability of the three compounds were assessed in human
liver microsome. As shown in Table 7, compounds 59 and 73 possessed good human microsomal
stability, especially 73. By contrast, compound 70 showed poor human microsomal stability.
0
1
2
3
4
5
6
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 7. In vitro metabolic stability in human liver microsome
Compd
Parameters
5
9
70
59.5
69
73
60 min remaining (%)
81.5
194
90.4
212
4.45
t1/2 (min)
Clint (mL/kg)
4.88
13.74
Then compounds 59 and 73 were selected for further in vivo pharmacokinetics study in healthy rats.
As shown in Figure 5, the two compounds showed low bioavailability after oral administration. For
intravenous dosing, 59 and 73 exhibited acceptable halfꢀlives (4.16 h and 3.95 h). In addition, 73
showed a higher exposure level than 59. Both compounds showed high volumes of distribution (86.62
L/kg and 29.59 L/kg) after intravenous administration, indicating high penetration into tissues. Then
compound 73, which possessed higher exposure level, was selected for tissue distribution study in an
HCT116 xenograft model. As shown in Figure 6, after intravenous administration of 60 mg/kg, 73 was
rapidly cleared from blood, spleen and liver. In contrast, the compound showed higher concentration and
longer retention in tumor, significant levels of compound were detected in tumors after 48h (>3 ꢁM/L).
This specific accumulation of 73 in tumors indicated the long action duration in xenograft models.
Further CYP450 inhibition determination showed 73 had no serious inhibition on the five major
isozymes (data shown in Table 8).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 67
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. Pharmacokinetic analysis in healthy rats. (A) Plasma concentration vs. time profile and pharmacokinetic
parameters of compound 59. (B) Plasma concentration vs. time profile and pharmacokinetic parameters of compound
73. The pharmacokinetic parameters are calculated using DAS2.0 software.
Figure 6. Tissue distribution of 73 in an HCT116 xenograft model. The plasma, tumors, livers, and spleens were
collected at the indicated times after intravenous (iv) administration of 60 mg/kg 73.
ACS Paragon Plus Environment

Page 25 of 67
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
Table 8. CYP450 isozymes inhibition of 73
a
%
inhibition
Isozymes
b
7
3
Positive controls
0
1
2
3
4
5
6
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
CYP1A2
CYP2C9
CYP2C19
CYP2D6
CYP3A4
8
76
82
88
89
72
36
21
48
22
a
b
Performed at 10 ꢁM concentration. αꢀnaphthoflavone (CYP1A2), sulfaphenazole (CYP2C9), ticlopidine (CYP2C19),
quinidine (CYP2D6), and ketoconazole (CYP3A4) are used as the positive control.
In vivo antiꢀtumor efficacy
The in vivo antiꢀtumor efficacy of compound 73 was evaluated in mice bearing HCT116 xenograft
tumors. In this study, the hydrochloride salt of 73 was administered intraperitoneally (ip) at 40 and 60
mg/kg according q.o.d. for 16 days. 8 (60 mg/kg) was chosen as the reference compound. As shown in
Figure 7A, compound 73 caused significant inhibition of tumor growth. The T/C values of the 40 mg/kg
and 60 mg/kg doses were 45% and 23%, respectively. The 60 mg/kg dose showed equivalent tumor
growth inhibition to that of the reference compound 8 (60 mg/kg). Compound 73 was well tolerated, as
evidenced by the absence of serious weight loss (Figure 7B). Hematoxylinꢀeosin (H&E) staining (Figure
7C) revealed that tumor cells in the xenografts decreased significantly after treatment with compound 73.
The pharmacodynamic biomarker changes in tumor xenografts of compound 73 were assessed by
Western blot 12 h after the final treatment. As shown in Figure 7D−E, compound 73 significantly
upregulated Hsp90 and Hsp70 via a feedback response to Hsp90 inhibition and induced an obvious
decrease of the client proteins Akt and Erk, confirming that its potent in vivo antiꢀtumor efficacy was
Hsp90 dependent.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 67
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 7. Antiꢀtumor efficacy of compound 73 in HCT116 human colon cancer xenografts. Compound 73 was
3
intraperitoneally (ip) administrated according q.o.d. for 16 days once the mean tumor volume reached 100 mm . (A)
Relative tumor volume (RTV) changes after treatment. Tumor diameters were measured every three days and used to
calculate the tumor volumes. RTV = V /V , where V is the tumor volume measured at each time point, and V is the
t
0
t
0
tumor volume at the beginning of treatment. Bars represent SEM. ** p < 0.01, Student’s tꢀtest (n = 6). (B)
Pharmacodynamic biomarker changes assessed by Western blot (12 h after final dose). (C) Representative H&E images
of the sacrificed tumors. (D) Western blot analysis of the pharmacodynamic biomarker changes in tumor xenografts.
Actin is used as the control for protein loading. (E) Densitometry analysis of the protein levels are showed as
normalized (to actin) ratios.
Ocular toxicity evaluation
Visual disturbance is the most common adverse event of Hsp90 inhibitors in clinical trials. It was
ACS Paragon Plus Environment

Page 27 of 67
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
reported that ocular toxicity is associated with the drug exposure level in retina. High retina exposed
Hsp90 inhibitors would induce photoreceptor cell death and retinal damage. The inhibitors with lower
retina/plasma exposure ratios and faster retinal elimination, including 6 and 11, showed no ocular
33, 39
0
1
2
3
4
5
6
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
toxicity either in clinical trials or preclinical models, indicating the ocular toxicity was drugꢀrelated.
In order to assess the safety profile, ocular toxicity of 73 was evaluated in a rat retinal damage model. 7
(NVPꢀAUY922) was chosen as the reference compound because visual disorders were observed both in
3
6,39
clinical trials and the preclinical model.
Rats were intravenously treated with 2ꢀday consecutive
dosing of vehicle, 73 (20 mg/kg), or 7 (10 mg/kg), then retinas were harvested for further biological
evaluation. Hsp70 upregulation is a surrogate marker for Hsp90 inhibition, so we detected the
expression level of Hsp70 in retinal tissue by Western blot. As shown in Figure 8A−B, 73 treatment
showed weak effect on Hsp70 expression while the reference compound 7 caused robust upꢀregulation
of Hsp70. This indicated that 73 had minimal influence on retina. The Hsp70 upꢀregulation was also
confirmed by immunohistochemistry assay (Figure 8C). At the same time, apoptotic induction effect on
retinal cells was evaluated by TUNEL staining (Figure 8D). No obvious retina cell death was observed
after treatment with 73. In contrast, 7 treatment induced extensive retina cell death, which was consistent
with the clinical visual disorders. Taken together, weak Hsp70 upꢀregulation and no retinal cell apoptotic
induction indicated compound 73 exhibited no ocular toxicity.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 67
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 8. Ocular toxicity evaluation of compound 73 in a rat retinal damage model. SD rats (n = 3) were intravenously
(iv) treated with 2 consecutive daily doses of vehicle, 73 (20 mg/kg) or 7 (NVPꢀAUY922, 10 mg/kg). The retina tissues
were harvested 24 h after the last treatment. (A) Hsp70 regulation in retina tissues after treatment assessed by Western
blot. Actin is used as the control for protein loading. (B) Densitometry graph of Hsp70 regulation. (C) Hsp70
immunohistochemistry in the retina after treatment. Scale bar, 50 ꢁm. (D) Apoptosis induction in the retina after
treatment evaluated by TUNEL staining. Scale bar, 50 ꢁm.
CONCLUSION
The structureꢀactivity relationships and structureꢀproperty relationships of novel Hsp90 inhibitors
were extensively studied, focusing on the lead compound 15. A series of derivatives were developed and
exhibited improved Hsp90ꢀinhibition and antiꢀproliferative activities. In addition, most of the synthetic
compounds showed minimal to moderate Hsp90α selectivity versus Grp94, should be because of the
conformation flexibility of the tetrahydropyrido[4,3ꢀd]pyrimidine core. Compounds 59, 70 and 73 were
selected for further evaluation. Potent antiꢀproliferative activities of the three compounds against a panel
of human cancer cell lines from diverse tissue origins indicated the broadꢀspectrum antiꢀcancer effects.
Modulating the in vitro biomarkers in HCT116 cells resulted in significant degradation of the Hsp90
clients Akt, Erk and pꢀErk associated with a dramatic induction of Hsp70, which is a commonly used
biomarker for Hsp90 inhibition. In the ADME assessment, 59 and 73 showed good human microsomal
ACS Paragon Plus Environment

Page 29 of 67
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
stability. After iv administration in healthy rats, both compounds showed acceptable halfꢀlives.
Considering the higher exposure level in blood, 73 was chosen for further in vivo study. In HCT116
xenografts, compound 73 showed significant tumor growth inhibition with no serious weight loss.
Potent tumor accumulation of 73 was also observed. In a rat retina damage model, 73 showed no ocular
toxicity, indicating its good safety profile. Overall, these results indicate that compound 73 is a
promising candidate agent for the treatment of cancer.
0
1
2
3
4
5
6
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 SECTION
1
. Biology.
1
.1. FP competition assay.
N
FP competition assay was used to evaluate the binding affinities of the compounds to Hsp90α and
N
Grp94 . The compounds (in DMSO) were prepared into various concentration solutions by serial diluted
in assay buffer (20 mM HEPES, PH 7.3, 50 mM KCl, 5 mM MgCl , 20 mM Na MoO , 0.01% NP40, 2
2
2
4
N
mM DTT, 0.1 mg/mL BGG). The fluorescent geldanamycin (GMꢀFITC) and purified Hsp90α or
N
Grp94 were diluted into the needed concentration in assay buffer. The experiments were conducted in
3
84 well black flatꢀbottomed polystyrene plates (Corning #3575). For each assay, equal volumes of the
N
N
diluted compounds (20 ꢁL), Hsp90α or Grp94 (30 nM final) and GMꢀFITC solution (10 nM final)
were added into the plate wells orderly and yielded a final volume of 60 ꢁL. Plates were covered and
rocked for 30 min at room temperature in the dark and then the FP values were detected using a
SpectraMax multimode microplate reader (Molecular Devices) with excitation and emission wavelength
at 485 and 535 nm, respectively. For each assay, the FP values of blank controls (GMꢀFITC only) were
recorded as P , the FP values of negative controls (GMꢀFITC and protein) were recorded as P and
min
max
the FP values of test wells (compounds, GMꢀFITC and protein) were recorded as P . The inhibition
test
rate of the compounds at each concentration point was calculated using the equation as follows:
inhibition rate (%) = [1 − (Ptest − P ) / (P − P )] × 100% and the IC values were calculated using
min
max
min
50
Graphpad Prim 6.0 software.
.2. In vitro antiꢀproliferative assay.
1
Antiꢀproliferative activities of the compounds against different cancer lines were determined using
ACS Paragon Plus Environment