Design and synthesis of potent dual inhibitors of JAK2 and HDAC based on fusing the pharmacophores of XL019 and vorinostat

Article abstract of DOI:10.1016/j.ejmech.2018.09.024

Specifically blocking more than one oncogenic pathway simultaneously in a cancer cell with a combination of different drugs is the mainstay of the majority of cancer treatments. Being able to do this via two targeted pathways without inducing side effects through a general mechanism, such as chemotherapy, could bring benefit to patients. In this work we describe a new dual inhibitor of the JAK-STAT and HDAC pathways through designing and developing two types of molecule based on the JAK2 selective inhibitor XL019 and the pan-HDAC inhibitor, vorinostat. Both series of compounds had examples with low nanomolar JAK2 and HDAC1/6 inhibition. In some cases good HDAC1 selectivity was achieved while retaining HDAC6 activity. The observed potency is explained through molecular docking studies of all three enzymes. One example, 69c had 16–25 fold selectivity against the three other JAK-family proteins JAK1, JAK3 and TYK2. A number of compounds had sub-micromolar potencies against a panel of 4 solid tumor cell lines and 4 hematological cell lines with the most potent compound, 45h, having a cellular IC₅₀ of 70 nM against the multiple myeloma cell line KMS-12-BM. Evidence of both JAK and HDAC pathway inhibition is presented in Hela cells showing that both pathways are modulated. Evidence of apoptosis with two compounds in 4 sold tumor cell lines is also presented.

Full text of DOI:10.1016/j.ejmech.2018.09.024

Accepted Manuscript
Design and synthesis of potent dual inhibitors of JAK2 and HDAC based on fusing the
pharmacophores of XL019 and vorinostat
Yu-yi Chu-Farseeva, Nurulhuda Mustafa, Anders Poulsen, Eng Chong Tan, Jeffrey
J.Y. Yen, Wee Joo Chng, Brian W. Dymock
PII:
S0223-5234(18)30796-7
10.1016/j.ejmech.2018.09.024
EJMECH 10735
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 11 July 2018
Revised Date: 5 September 2018
Accepted Date: 9 September 2018
Please cite this article as: Y.-y. Chu-Farseeva, N. Mustafa, A. Poulsen, E.C. Tan, J.J.Y. Yen, W.J.
Chng, B.W. Dymock, Design and synthesis of potent dual inhibitors of JAK2 and HDAC based on fusing
the pharmacophores of XL019 and vorinostat, European Journal of Medicinal Chemistry (2018), doi:
10.1016/j.ejmech.2018.09.024.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Design and Synthesis of Potent Dual Inhibitors of JAK2 and HDAC Based on
Fusing the Pharmacophores of XL019 and Vorinostat
a
b
c
d
d
Yu-yi Chu-Farseeva, Nurulhuda Mustafa, Anders Poulsen, Eng Chong Tan, Jeffrey J. Y. Yen, Wee
b,e,f
a
Joo Chng, Brian W. Dymock, #
5
a
Department of Pharmacy, National University of Singapore, Singapore ; Department of Medicine,
Yong Loo Lin School of Medicine, National University of Singapore, 1E Kent Ridge Road, NUHS
b
Tower, Block Level 10, Singapore 119228 , Experimental Therapeutics Centre, 31 Biopolis Way, 03-
c
d
0
1 Nanos, Singapore 138669 , Institute of Biomedical Sciences, Academia Sinica, Taiwan , Cancer
e
1
0 Science Institute, Singapore, National University of Singapore, Singapore 117599 , National
f
University Cancer Institute of Singapore, National University Health System Singapore 119074 .
Running Head: Synthesis and SAR of Novel Dual inhibitors of JAK2/HDAC
5 #Address correspondence to Brian W. Dymock, bwdnus@gmail.com.
1
Supporting information for this article includes additional synthesis of compounds 26, 30-32, 37-39,
S1-S5, 63b-f; selectivity indices for inhibition of cell proliferation; additional JAK-STAT Western
1
13
blots; and H and C NMR spectra for all key compounds.

ACCEPTED MANUSCRIPT
2
0 ABSTRACT
Specifically blocking more than one oncogenic pathway simultaneously in a cancer cell with a
combination of different drugs is the mainstay of the majority of cancer treatments. Being able to do
this via two targeted pathways without inducing side effects through a general mechanism, such as
chemotherapy, could bring benefit to patients. In this work we describe new dual inhibitor of the JAK-
5 STAT and HDAC pathways through designing and developing two types of molecule based on the
JAK2 selective inhibitor XL019 and the pan-HDAC inhibitor, vorinostat. Both series of compounds
had examples with low nanomolar JAK2 and HDAC1/6 inhibition. In some cases good HDAC1
selectivity was achieved while retaining HDAC6 activity. The observed potency is explained through
molecular docking studies of all three enzymes. One example, 69c had 16-25 fold selectivity against
0 the three other JAK-family proteins JAK1, JAK3 and TYK2. A number of compounds had sub-
micromolar potencies against a panel of 4 solid tumor cell lines and 4 hematological cell lines with the
most potent compound, 45h, having a cellular IC of 70 nM against the multiple myeloma cell line
2
3
3
50
KMS-12-BM. Evidence of both JAK and HDAC pathway inhibition is presented in Hela cells showing
that both pathways are modulated. Evidence of apoptosis with two compounds in 4 sold tumor cell
5 lines is also presented.
KEYWORDS: multicomponent ligand, JAK2 inhibitor, HDAC inhibitor, JAK/HDAC dual inhibitor
INTRODUCTION
Target therapies have achieved significant gains in the fight against cancer, however, they are still a
0 long way from providing generally curative treatments for the majority of cancers. Targeted agents can
4
1
be used with each other or with traditional chemo- or radiotherapy. However more research is
required to develop new combinations of targeted agents by finding pathways which can be inhibited

ACCEPTED MANUSCRIPT
2
together and lead to better and more durable responses in patients. Targeting the JAK-STAT pathway
3
with JAK kinase inhibitors, such as ruxolitinib, has provided value for patients with
4
4
5
5
6
5 myeloproliferative neoplasms, conditions which can lead to leukemia. Likewise, inhibitors of the
5
histone deacetylase (HDAC) pathway have found value in T-cell lymphomas. In an exciting new way
forward, we propose to combine the targeting of JAK kinases and HDACs with a single, dual inhibitor.
Clinical trials of ruxolitinib with panobinostat support the notion that blockade of these pathways may
6
be useful in treating solid tumors or hematological malignancies. Recently we have reported new dual
0 JAK-HDAC inhibitors, such as 1 and 2 (Figure 1), based on the JAK inhibitor templates pacritinib and
7
, 8
ruxolitinib, respectively.
In this latest study we report the discovery of dual JAK-HDAC agents
based on XL019 (3). Key discoveries with this series are the very potent JAK and HDAC1/6 activities
which have led to very potent cellular activites. Compound 3 is a potent and selective JAK2 inhibitor
with IC50 of 2.2 nM, exhibiting >50-fold selectivity over JAK1, JAK3 and TYK2. It has been studied
9
5 in a phase 1 clinical trial but was forced to terminate due to CNS side effects. In our study, CNS
penetration is not expected due to the higher molecular weight and polarity of the compounds.
Importantly, compound 3 has inherent selectivity for JAK2 over the other JAK family enzymes, unlike
ruxolitinib which has a JAK1/2 profile, possibly a factor in its safety profile where patients with low
1
0
platelet counts cannot be treated. In this report we demonstrate the design and synthesis of novel
0 JAK-HDAC dual inhibitors based on 3 and the marketed HDAC inhibitor vorinostat (4).

ACCEPTED MANUSCRIPT
OH
HN
O
N
H
O
N
O
HN
N N
O
O
N
N
O
N
NH
O
N
N
OH
H
N
N
H
N
N
H
YLB343B (2)
EY3238 (1)
JAK2 IC50 = 1.4 nM
HDAC6 IC50 = 2.1 nM
JAK1 IC50 = 11 nM
HDAC1 IC50 = 6 nM
HDAC6 IC50 = 1 nM
XL019 (3)
JAK2 IC50 = 5.7 nM
JAK3 IC50 = 368.1 nM
7
8
Figure 1 Structures of JAK-HDAC dual inhibitors EY3238 (1), YLB243B (2) and the JAK2
9
selective XL019 (3).
6
5 RESULTS AND DISCUSSION
In silico analysis and design of compounds
Designing a dual JAK-HDAC inhibitor by merging of the pharmacophores of 3 and 4 may be achieved
with two designs (Figure 2). Both designs employ the core aniline substituted biarylpyrimidine as the
JAK2 binding template. Both designs also take advantage of solvent channels in JAK2 where the
0 hydroxamate bearing side chain, necessary for binding in the deep HDAC pocket, can be attached. By
varying this chain length control over HDAC potency is expected to be achieved. In Design A the
hydroxamate side chain is attached to the aniline at either the meta or para position with variations
possible in the R group attached to the acetamide. In Design B, the hydroxamate side chain is attached
to the acetamide, taking advantage of the solvent channel in JAK2, and the R group is again variable
5 attached to the aniline phenyl ring. A range of compounds based on Designs A and B were then
prepared varying the length of the hydroxamate bearing side chain and with different R groups.
7
7

ACCEPTED MANUSCRIPT
O
N
H
N
HN
O
linker optimisation to
search for best balance
of JAK and HDAC activity
JAK2
N
O
N
kinase
domain
critical
binding
elements
O
N
H
N
HN
R
OH
merge key
pharmacophore
elements
HN
N
N
H
O
O
maintain key Zn-
binding hydroxamate
OH
NH
XL019 (3)
JAK2 inhibitor
R
tuning of
key HDAC
Zn-binding
N
N
H
physicochemical
properties
maintian critical kinase hinge binding
components in aminopyrimidine
N
N
H
O
head group
O
Design A
Design B
N
H
HN
OH
Vorinostat (4)
pan-HDAC inhibitor
Designed Multiple Ligands (DML)
based on XL019 and Vorinostat
Figure 2 Design options for construction of a dual JAK-HDAC inhibitor from XL019 (3) and
vorinostat (4). Designs A and B indicate attachment of the HDAC binding hydroxamate chain at two
0 different positions of the core JAK binding biaryl-aminopyrimidine. Physical properties could be tuned
by varying the ‘R’ group.
8
Chemistry
Simplified analogues of 3 were prepared to assess the ‘base’ levels of JAK2 activity in the scaffold
(
Figure 3). Replacement of the basic pyrrolidine of 3 with a methyl, giving an acetamide substitution,
8
5 enabled isolation of reaction products without problems of protonation of the basic centre. Three
substitutions of the pyrimidine: aniline (6), benzylamine (7) and phenethylamine (8) were prepared
from 5 in hot butanol catalysed by para-toluene sulfonic acid (PTSA). Chloropyrimidine 5 was
9
prepared by a palladium coupling between dichloropyrimidine and the boronate ester of acetanilide.

ACCEPTED MANUSCRIPT
O
N
O
HN
HN
a
N
R
N
Cl
N
N
H
5
6 (R=Ph)
7
8
(R=Bn)
(R=(CH ) Ph)
2
2
Reagents and Conditions: (a) aniline, benzylamine
or phenethylamine, PTSA, nBuOH, 100°C, 2h.
9
0 Figure 3 Synthesis of compounds 6-8.
To enable exploration of amide-linked Design A (Figure 2) specific aniline building blocks had to be
prepared with variation of chain length (Figure 4). Mono-Boc protection of diaminobenzene (meta- (9)
and para-(17)) was achieved employing excess diamine to minimize unwanted bis-protection while
ensuring practical isolation of 10 and 18. Amide formation via acid chlorides of 11 (4 methylene
5 chain) and 12 (6 methylene chain) gave protected esters 13/14 and 19/20. Deprotection using
concentrated sulfuric acid furnished the desired anilines 15/16 and 21/22 in overall good yield.
9

ACCEPTED MANUSCRIPT
Figure 4 Synthesis of meta and para substituted aniline building blocks 21-22.
In addition to the amide linked series, we wanted to explore a flexible ether linked side chain (Figure
00 5). Hence meta- (23) or para-nitrophenols (33) were alkylated with bromo-esters 24-26 in hot
acetonitrile. The resulting nitro-ethers 27-29 and 34-36 were reduced to the desired anilines 30-32 and
1
3
7-39 using iron and ammonium chloride in aqueous methanol.
Figure 5 Synthesis of meta and para substituted ether building blocks 30-32 and 37-39.

ACCEPTED MANUSCRIPT
1
1
1
05 Ester intermediates 42a-j were formed with a nucleophilic substitution reaction between biaryl 5 and
anilines (Figures 4 and 5), under similar conditions to 6-8 (Figure 3). A mixture of the desired methyl
and the transesterified butyl ester were obtained in moderate yield, however this was inconsequential
since ester intermediates were saponified using lithium hydroxide in THF/water to yield the desired
acids 43a-j. The crude product was directly used for the next step without further purification.
10 Formation of THP-protected hydroxamates 44a-j was achieved in a HATU-mediated amide formation
with THP-protected hydroxylamine in DMF overnight. Final THP deprotection was accomplished
under mild conditions with in situ HCl generated from acetyl chloride in cold MeOH/DCM to afford
4
5a-j. Target compounds which precipitated were washed with DCM or MeOH. Preparative reverse
phase HPLC purification was necessary for target compounds which did not satisfactorily precipitate
15 from the reaction mixture.
Figure 6 Synthesis of amide and ether linked JAK-HDAC dual inhibitors 45a-j.

ACCEPTED MANUSCRIPT
Improvement in JAK2 potency has been shown to be possible by adding small hydrophobic groups
20 at the 5’ position of the pyrimidine. Acetylation of para-bromoaniline (46) gave 47 followed by
palladium coupling gave boronate ester 48. Coupling of 48 with dichloropyrimidine furnished biaryl
chloropyrimidine 49 which was then coupled in a third palladium catalyzed reaction followed by
saponification, THP-protected hydroxamate formation and final deprotection to give three test
compounds, 50-52, exploring meta/para substitution and chain length (Figure 7 and Table 3). Model
25 compound 49b was also prepared via aniline 63c (see Figure 9).
1
1
Figure 7 Synthesis of 5’-methylpyrimidine intermediate 49.
To explore the compatibility of a basic side chain with the hydroxamate we chose to install a
methylpiperazine appended to a methylene, to increase flexibility, attached to the aromatic amide
30 (Figure 8). Starting from intermediate 53 and 54 the acetamide was cleaved in 4N HCl to reveal
anilines 55 and 56. Amide formation with methylpiperazine acetic acid was achieved with 1-
1
[
(
bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium
3-oxid
hexafluorophosphate
HATU) as coupling agent to give 57 and 58. THP-protected hydroxamates 59 and 60 were then
formed from THP protected hydroxylamine and deprotected as described previously to afford target
135 compounds 61 and 62.

ACCEPTED MANUSCRIPT
Figure 8 Synthesis of basic methylpiperazine hydroxamates 61 and 62.
To prepare analogues with the hydroxamate chain appended to the upper aryl group of the structure
40 chloropyrimidine intermediate 5 underwent Buchwald coupling with aniline (63a) and a range of
meta- and para- substituted anilines 63b-f (Figure 9). The acetamides 64a-f of the resulting
anilinopyrimidines were cleaved to give anilines 65a,c-f then amides formed with methyloctanoic acid
affording esters 66a,c-f. Saponification of the esters formed 67a,c-f and was followed by hydroxamate
formation (68a,c-f) via the THP protected hydroxamate and deprotection as described above to give
45 target compounds 69a,c-f.
1
1

ACCEPTED MANUSCRIPT
Figure 9 Synthesis of hydroxamates 69a, c-f linked through the aromatic amide.
Appending a methyl group to the 5’ position of the pyrimidine of 69b,f was achieved from 5’-
methylpyrimidine intermediate 49. Initial hydrolysis of 49 with HCl/dioxane to give anilines 70
50 followed by amide formation to install the methylene chain (71), hydrolysis to acids (72), capping with
THP-protected hydroxylamine and final deprotection with either acetyl chloride in methanol (74a) or
HCl/dioxane (74b).
1
O
O
O
H
NH2
OR
N
HN
HN
HN
OR
O
X
O
X
a
b
d
X
X
N
N
N
N
N
N
N
N
N
N
H
N
N
N
N
N
N
H
H
H
4
9b (X=O)
70a (X=O)
70b (X=NMe)
71a (X=O, R=Me)
71b (X=NMe, R=Me)
73a (X=O, R=pyran)
73b (X=NMe, R=pyran)
49c (X=NMe)
c
e
72a (X=O, R=H)
74a (X=O, R=H)
72b (X=NMe, R=H)
74b (X=NMe, R=H)
(
a) 4N HCl/Dioxane, MeOH, 50°C, 18h; (b) Methyl-1,8-dioctanoic acid, HATU, Et N, DMF, RT, 18h; (c) LiOH, THF/H O (3:2), RT, 18ht; (d)
3
2
THPONH2, HATU, Et3N, DMF,RT, 18h; (e) AcCl, MeOH/ DCM (1:10), RT, 18h (74a) or 4N HCl/Dioxane, DCM, RT, 18h (74b)

ACCEPTED MANUSCRIPT
Figure 10 Synthesis of 5’-methyl substituted hydroxamates 74a,b.
55 In Vitro Enzyme Assay Assessments
1
Synthesised compounds were tested in single concentration or dose response assays against JAK2,
HDAC1 and HDAC6. A biochemical assay measuring the rate of ATPase activity of JAKs was used
for all kinase data and a fluorescence assay measuring the rate of HDAC cleavage of an acetylated
7
peptide substrate was used for quantification of HDAC inhibition.
1
60 Model JAK2 compounds 6-8 were tested in single concentration assays to ascertain the best template
for connection to the aminopyrimdine (Table 1). Not surprisingly, the aniline 6 gave over 96%
inhibition of JAK2 activity at 10 µM (IC of 27.4 nM, see Table 2). However benzyl 7 and phenethyl
5
0
8
had only weak inhibition of the kinase. Based on this result further studies were carried out with 6 as
the core template substituting various 6-membered rings to the meta- and para- positions of the aryl
165 ring. These compounds were prepared using a Pd-coupling protocol as intermediates for dual inhibitors
(
see Figure 9). Meta-substituted morpholine 64b and piperidine 64d both had JAK2 IC values in the
50
1
0-20 nM range. Furthermore, para-substituted piperidine 64e and methylpiperazine 64f had IC50
values less than 10 nM. Hence both meta- and para-substituted aryls were selected for linking to
hydroxamates.
Table 1. JAK2 Kinase Inhibition for model compounds
170

ACCEPTED MANUSCRIPT
JAK2 IC50 (nM) or
Compound
R
%
Inhibition @ 10 µM
9
3
(XL019)
-
2.2
6
96.42%
3.85%
6.22%
7
8
6
6
4b
4d
19.6
13.8
6
4e
2.9
7.9
6
4f

ACCEPTED MANUSCRIPT
Testing of amide and ether linked hydroxamates was carried out in duplicate 10-point dose
responses against JAK2, HDAC1 and HDAC6 (Table 2). Surprisingly, the meta-substituted 4
methylene 45a had an IC50 of 2.6 nM for JAK2 and 4.6 nM for HDAC6 with selectivity of nearly
100 fold over HDAC1. Extending the number of methylenes in the linker to 6 (45b) resulted in a
boost of HDAC1 activity by over an order of magnitude and of HDAC6 potency by more than 5
fold with only a small loss of JAK2 activity. This level of potency was not replicated with para-
derivative 45c, bearing 4 methylenes, which had reduced potency against all targets. However
extending this compound to 6 methylenes (45d) resulted in sub-nanomolar potency against both
JAK2 (0.9 nM) and HDAC6 (0.1 nM) with over 7,000 fold selectivity for HDAC6 over HDAC1.
For confirmation the methyl/butyl ester 42i was tested against all targets and confirmed to show
negligible HDAC activity while retaining moderate JAK2 activity. A similar trend continued into
the ether series with meta-substituted 45e/45f having low nanomolar activity for JAK2 and
HDAC6 with moderate HDAC1 activity. A further extension to 7 methylenes (45g) did not
substantially affect the profile. Para-substituted analogues 45h-j exhibited a similar overall profile
to the meta-substituted analogues with the most selective having the longer linker.
175
180
185
Table 2. JAK2 and HDAC Inhibition for Compounds 45a-j
JAK2 IC50
nM)
HDAC1 IC50 HDAC6 IC50
(nM) (nM)
Compound
X
(

ACCEPTED MANUSCRIPT
2
9
4
(Vorinostat)
-
-
306
20
6
-
27.4 ± 1.2
45a
2.6
384
4.6
4
5b
9.2 ± 1.5
32.4 ± 4.2
0.81 ± 0.05
H
N
4
5c
58.3 ± 12.8
3350 ± 2774
774 ± 30
117 ± 138
O
4
5d
0.9 ± 0.4 nM
0.10 ± 0.05
4
5e
3.4 ± 3.1
3.6 ± 0.1
17 ± 1.5
637 ± 214
83 ± 29
2.4 ± 0.4
1.8 ± 1.2
6.7 ± 5.7
4
5f
45g
1043 ± 95
4
5h
33 ± 12
239
89 ± 31
>10,000
1.3 ± 0.6
3150
4
2i
Ester

ACCEPTED MANUSCRIPT
4
5i
23 ± 6.4
145 ± 8
3.0 ± 0.7
14 ± 3.5
45j
46 ± 12
1701 ± 1227
JAK2 inhibitors with a biaryl pyrimidine have been reported to enjoy an increase in potency when a
1
1, 12
methyl group is installed at the 5’position.
This is due to hydrophobic interactions in a small
1
3
1
90 pocket at the base of the ATP site. We prepared one reference compound to confirm that potency was
present in our template (49b). This benefit does not transfer to ethers 52 and 50 (Table 3), which have
a very similar profile to their corresponding non-methyl counterparts (see compounds 45h and 45i,
Table 2). On the other hand, 51 has gained just over 5 fold JAK2 activity, compared to 45g, while
increasing selectivity over HDAC1 with maintenance of potency against HDAC6.
195
Table 3. JAK2 and HDAC Inhibition for Compounds 49b, 50-52
Compound
X
JAK2 IC50
HDAC1 IC50
HDAC6 IC50

ACCEPTED MANUSCRIPT
(nM)
(nM)
(nM)
4
9b
28.7
-
-
-
5
0
3.9 ± 2.7
155 ± 75
6.1 ± 0.7
9.7 ± 6.3
51
2.93 ± 1.21
25.6 ± 4.2
4581 ± 2367
52
260 ± 116
3.5 ± 1.7
Compound 3 possesses a basic pyrrolidine appended to its aromatic amide. This pyrrolidine is
somewhat rigid and could restrict optimal binding to either JAK2 or HDAC. Hence to explore whether
basicity in our series was compatible with a hydroxamate side chain we prepared two analogues
00 bearing a flexible methylpiperazine side chain, 61 and 62. Generally, this change reduced potency
against all targets but particularly against JAK2 with a 30 fold loss of activity for 61 (compared with
2
4
5f) but only a 4 fold loss for the para-linked derivative 62 (compared with 45i). However HDAC1
and HDAC6 activities were not strongly affected.
Table 4. JAK2 and HDAC Inhibition for Compounds 61 and 62

ACCEPTED MANUSCRIPT
JAK2 IC50
HDAC1 IC50
HDAC6 IC50
(nM)
Compound
X
(nM)
(nM)
6
1
2
109 ± 4.7
348 ± 43
6.8 ± 0.9
6.1 ± 0.7
6
95.5 ± 7.39
142 ± 89
2
05 Having explored connection of the hydroxamate via the pyrimidine ring we next turned to
investigation of Design B (Figure 2) with the hydroxamate being connected to the phenyl ring of the
biaryl system. A linker of 6 methylenes, generally optimal for HDAC potency, was maintained with
variation of the pyrimidine substituent (Table 5). With a simple aniline (69a), very good potency was
achieved for both HDACs 1 and 6, however the JAK2 IC50 dropped to 41 nM. Despite the drop in
10 JAK2 potency, this template is clearly worthy of exploration as a potency dual JAK/HDAC inhibitor.
Exploring para-substituted analogues with 3 rings known to be potent against JAK2 in the XL019
series led to interesting results. Morpholine 69c had low nanomolar activity with IC s of 3.1 nM and
2
5
0
1
.2 nM for JAK2 and HDAC6, respectively, whereas potency against HDAC1 was 56 nM. This data
indicates that 69c is one of the most potent HDAC1 inhibitors of the series along with 45b. Piperidine

ACCEPTED MANUSCRIPT
215 69d generally loses potency, especially against HDAC1, but piperazine 69f generally loses potency
with similar IC s to 69c. Meta-substituted 69d was more potent than its para-substituted analogue
5
0
with a similar selectivity profile.

ACCEPTED MANUSCRIPT
Table 5. In vitro enzyme inhibition activities (IC50) of target compounds (69a,c-f)
JAK2 IC50
HDAC1 IC50
(nM)
HDAC6 IC50
(nM)
Compound
R
(nM)
6
9a
41.3
3.8
1.44
6
9c
3.1 ± 2.4
56 ± 23
1.2 ± 0.1
6
9d
6.6 ± 0.3
18 ± 0.1
179 ± 27
934 ± 9.6
9.0 ± 0.7
23 ± 4.6
6
9e
6
9f
4.6 ± 0.4
77 ± 15
3.1 ± 0.3

ACCEPTED MANUSCRIPT
2
20 Methylation of the 5’-position of the pyrimidine (74a, Table 6) retained potency against JAK2 and
HDAC6 while potency against HDAC1 was reduced about 5 fold (compare 74a with 69c). With
methylpiperazine derivative 74b the inhibitory profile did not change upon methylation (compare 74b
with 69f).
Table 6. Enzyme inhibition activities (IC ) of compounds with 5’methyl pyrimidine substitution
5
0
#
R
JAK2 IC (nM)
50 50
HDAC1 IC (nM) HDAC6 IC (nM)
5
0
7
4a
1.8 ± 0.1
289 ± 15
3.7 ± 1.1
7
4b
4.3 ± 0.3
139 ± 23
2.1 ± 1.2
225

ACCEPTED MANUSCRIPT
Inhibition of JAK Family Kinases
Compound 69c, with strong JAK2 potency, was selected for assessment against the JAK3 family kinases
comprising JAK1, JAK3 and TYK2. 69c had an IC of 52.1 Nm against JAK1, a selectivity of 16.8 fold. For
5
0
230 JAK3 and TYK2 the values were very similar, around 80 nM, with selectivities of 25-26 fold (Table 7).
Table 7. Enzyme inhibition activities (IC50) and fold selectivities of compound 69c
against the JAK family kinases
JAK1
JAK3
TYK2
IC (nM)
52.1
80.1
79.4
5
0
Selectivity over JAK2
16.8
25.8
25.6

ACCEPTED MANUSCRIPT
Docking Studies Explain the Observed Potency
When docked into JAK2 69c forms 2 hydrogen bonds with the hinge residue Leu932 (Figure 11A) utilising both the carbonyl and NH.
35 Furthermore, the phenyl and elongated side chain is buried in a pocket formed from the glycine rich loop (Figure 11B) with the
hydroxamate extending out into solvent. In HDAC6, the hydroxamate of 69c chelates the Zn at the base of the substrate pocket with the
extended elongated chain exiting the binding site at the amide linkage (Figure 11C/D). The hinge-binding motif interacts on the surface
of HDAC6 with residues Arg673, Phe679 and Thr678. Taken together, modelling studies indicate that 69c can bind to both JAK2 and
HDAC6 explaining the observed potency.
2
A
B

ACCEPTED M
C
ANUSCRIPT
C
C
D
2
2
40 Figure 11 Compound 69c docked into JAK2 and the second catalytic domain of HDAC6. Compound 69c is shown with green carbon
in stick representation. JAK2 is shown as a cyan ribbon with selected residues in thin stick with grey carbon (A) and with an electrostatic
surface (B). The aminopyrimidine forms 2 hydrogen bonds (yellow dashed lines) to the backbone of the kinase hinge residue Leu932
(NH and C=O). The phenyl ring bearing the long side chain extends under the glycine-rich loop with the hydroxamic acid extending out
of the binding site into solvent. HDAC6 is shown as a cyan ribbon (C) and with an electrostatic surface (D). The hydroxamic acid forms a
45 cluster of hydrogen bonds/ionic interactions to the His611, Tyr782 and the catalytic Zinc. Compound 69c is shown in stick with green
carbon and the catalytic Zinc in magenta CPK representation. The hydroxamic acid coordinates to the catalytic Zinc in a bidentate
manner. The kinase hinge-binding motif binds on the surface of HDAC6 while the morpholine-aniline may interact with a shallow
hydrophobic pocket around Thr678 and Met682.

ACCEPTED MANUSCRIPT
Inhibition of Cell Proliferation
2
50 All dual inhibitors were tested against a panel of 4 solid tumor cell lines selected as representative of
hard-to-treat solid tumors while being known to be sensitive to HDAC inhibitors: colorectal HCT116,
breast MDA-MB-231 and MCF-7, and prostate PC-3 (Table 8). Reported JAK-HDAC dual inhibitors,
1
and 2, compared similarly to the pan-HDAC inhibitor 4, whereas JAK2 inhibiting 3 was less
antiproliferative in these cell lines. Of the new compounds from Design A, HCT116 cells were
55 sensitive to some compounds (most potent 45f had IC50 = 0.66 M) but not others (least active 45c
IC = 65.6 M). Compound 45c was the least active against all cell lines and was also the least active
2
µ
µ
5
0
over the three isolated enzymes as well. It may at first appear that the cellular activity is driven by
HDAC1 activity (45c has the lowest HDAC1 IC50 and 45b is 10 fold less active for HDAC1 then 45a
with a commensurate reduction in cell activity). However, this does not hold for 45d and 45e which
60 both have lower HDAC1 activity but good cell activity. This result could be due to a strong
JAK2/HDAC6 dual potency which these two compounds from different series have in common. This
trend appears to hold for the remainder of the compounds, which have strong JAK2/HDAC6 activity
with moderate HDAC1 inhibition. The exception appears to be 51 which has a very weak HDAC1
potency with commensurately lower cellular activity. It is possible that a balance of activity across the
65 enzyme targets is required for broad cell activity.
2
2
To assess whether general toxicity contributed to the antiproliferative action of the dual inhibitors we
1
4
assessed their effects on two normal cells lines, the mouse cell line TAMH and the cardiomyocyte
1
5
cell line AC10 (see Supporting Information, Table S1). As an example, one of the more potent
compounds, 45d has IC50 values of 19.5 and 12.5 µM for AC10 and TAMH, respectively, was up to 18
70 fold less active against AC10 and nearly 12-fold for TAMH. In general virtually all the data against
AC10 and TAMH cells indicates higher potency towards cancer cells. These levels of selectivity
2

ACCEPTED MANUSCRIPT
suggest that the antiproliferative actions of the compounds is not due a generally toxic mechanism of
action.
For compounds from Design B, 69a, highly potent against HDAC1 and 6 and moderately potent
2
75 against JAK2, was the most potent across all cell lines with sub-micromolar IC s. This compound has
5
0
SI values in AC10 and TAMH cells ranging from 9 fold to over 38 fold indicating that this compound
is not simply toxic (see Supporting Information, Table S1). Other compounds in this series were potent
in the low micromolar range with significantly positive SI values indicating that dual JAK-HDAC
inhibitors have broad cellular efficacy without affecting normal cells in the same concentration range.
280
Table 8. In vitro cell proliferation activities (IC ) of selected compounds in solid tumor cell lines
50
a
IC (µM) ± SD
5
0
Compound
b
c
c
d
HCT-116
MDA-MB-231
1.43 ± 0.53
0.79 ± 0.18
9.67 ± 3.09
1.49 ± 0.11
14.6 ± 3.31
2.48 ± 0.74
32.1 ± 0.69
1.54 ± 0.20
1.24 ± 0.18
1.76 ± 0.08
3.61 ± 0.64
1.02 ± 0.07
MCF-7
PC-3
1
2
2.23 ± 0.63
2.32 ± 1.03
7.34 ± 0.32
2.20 ± 0.73
34.1 ± 6.16
1.15 ± 0.31
65.6 ± 9.92
1.97 ± 0.16
0.79 ± 0.08
0.66 ± 0.11
1.51 ± 0.40
1.15 ± 0.32
1.47 ± 0.37
0.84 ± 0.07
27.2 ± 6.87
0.81 ± 0.01
26.7 ± 9.45
0.32 ± 0.12
43.4 ± 3.88
1.09 ± 0.56
2.55 ± 0.38
2.42 ± 0.43
1.22 ± 0.09
2.01 ± 0.12
1.7 ± 0.3
2.41 ± 0.1
3
(XL019)
(Vorinostat)
5a
5b
5c
5d
5e
5f
5g
5h
14.9 ± 3.12
1.00 ± 0.15
18.6 ± 4.00
0.34 ± 0.04
47.3 ± 0.83
3.84 ± 0.38
1.66 ± 0.14
1.89 ± 0.15
1.64 ± 0.99
1.51 ± 0.12
4
4
4
4
4
4
4
4
4

ACCEPTED MANUSCRIPT
4
5i
0.99 ± 0.04
1.80 ± 0.35
1.19 ± 0.14
18.4 ± 6.71
0.92 ± 0.34
0.77 ± 0.11
1.69 ± 0.19
0.19 ± 0.1
1.05 ± 0.14
1.07 ± 0.4
0.99± 0.6
1.56 ± 0.13
5.33 ± 1.51
1.40 ± 0.16
11.44 ± 1.12
1.06 ± 0.21
0.65 ± 0.07
3.88 ± 0.26
0.27 ± 0.09
0.99 ± 0.16
2.83 ± 0.29
0.91 ± 0.03
1.23 ± 0.23
0.97 ± 0.05
0.81 ± 0.08
5.53 ± 0.45
1.71 ± 0.12
30.8 ± 3.70
2.37 ± 0.48
1.32 ± 0.48
4.84 ± 0.03
0.33 ± 0.09
0.70 ± 0.18
1.89 ± 0.56
1.61 ± 0.080
2.47 ± 0.74
1.30 ± 0.26
1.47 ± 0.30
4.73 ± 1.25
2.76 ± 0.06
32.8 ± 1.37
2.41 ± 0.29
1.15 ± 0.30
4.09 ± 0.34
0.21 ± 0.03
0.64 ± 0.04
5.58 ± 0.34
1.02 ± 0.14
1.74 ± 0.23
45j
5
0
1
2
1
2
5
5
6
6
6
9a
6
9c
9e
6
6
9f
4a
4b
7
2.53 ± 0.7
1.05 ± 0.42
7
1.04 ± 0.08
a
b
c
d
:
10 dose response carried out at least in triplicate; : colorectal cancer; : breast cancer; : prostate
cancer.
285 Both JAK2 and HDAC inhibitors have been found to be effective in hematological cancers and have
1
6, 17
progressed to the later stages of clinical trials.
We therefore selected compounds for testing in a
small panel of hematological cell lines, the erythroleukemia cell line HEL92.1.7 (bearing mutant
V617F
JAK2
), the acute T-cell leukemia cell line Jurkat and two multiple myeloma cell lines KMS-12-
BM and XG-6 (Table 9). These cells were sensitive to reference compounds 1, 2 and 4 in the 0.5-
90 2.0 µM range. The tested compounds had broadly similar activities to the reference compounds but
with some notable exceptions. The multiple myeloma cell line KMS-12-BM was the most sensitive to
the compounds, but not the reference compounds, with IC values as low as 70 and 100 nM for 45h
2
5
0
and 45i, respectively. Interestingly, even 51 was quite potent against KMS-12-BM cells despite being
of low potency against the other cell lines. On the other hand, IL-6 driven XG-6 cells were somewhat

ACCEPTED MANUSCRIPT
295 insensitive to the compounds with IC50 values ranging from 1.07 to 10.2 µM. Surprisingly, meta-
substituted amides 45a and 45b were not potent at all at the top concentration tested of 10 µM whereas
the para-substituted amide 45d had IC50s in the low micromolar range for 3 cell lines tested. We
surmised that this could be due to unfavourable properties of the meta-substituted amide, perhaps poor
permeability. This was not investigated further since good activity was noted with the ether linked
00 series of compounds. HDAC1 potency did not appear to be a driver of activity in all cases, for example
3
4
5d and 45e, with HDAC1 IC s of around 1 µM were quite potent in all cell lines whereas 45g, with a
50
similar level of HDAC1 potency was only active against KMS-12-BM cells. Similar to the pattern of
activity against solid tumor cells, it appears that a good balance of activity between the JAK and
HDAC targets is required for optimal inhibition of proliferation.
3
05 When the IC s against the normal cells AC10 and TAMH were assessed (see Supporting Information,
5
0
Table S2) some very high SI values were apparent for the KMS-12-BM cell line particularly for the
very potent compound 45h. Other SI values for other cell lines were also significantly positive.
Compound 45g was notable due to its lack of sensitivity in AC10 cells but its lower IC value for
5
0
TAMH resulted in less favourable SI values. In general, this group of JAK-HDAC dual inhibitors have
10 good potency in hematological cell lines with many examples of lower toxicity as measured in the
AC10 and TAMH cell lines.
3