Structure based drug design of Pim-1 kinase followed by pharmacophore guided synthesis of quinolone-based inhibitors

Article abstract of DOI:10.1016/j.bmc.2017.07.036

Over expression of Human phosphatidyl inositol mannoside kinases isoform 1 (Pim-1 kinase) has been reported in several leukemia and solid tumors. Our continuous interest to reveal the secrecies of the mysterious Pim-1 kinase binding pocket has led us to e

Full text of DOI:10.1016/j.bmc.2017.07.036

Accepted Manuscript
Structure based drug design of Pim-1 kinase followed by pharmacophore guided
synthesis of Quinolone-based inhibitors
Lubna Swellmeen, Rand Shahin, Yusuf Al-Hiari, Amani Alamiri, Alaa Hasan,
Omar Shaheen
PII:
S0968-0896(17)31071-4
DOI:
http://dx.doi.org/10.1016/j.bmc.2017.07.036
BMC 13873
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
17 May 2017
18 July 2017
19 July 2017
Please cite this article as: Swellmeen, L., Shahin, R., Al-Hiari, Y., Alamiri, A., Hasan, A., Shaheen, O., Structure
based drug design of Pim-1 kinase followed by pharmacophore guided synthesis of Quinolone-based inhibitors,
Bioorganic & Medicinal Chemistry (2017), doi: http://dx.doi.org/10.1016/j.bmc.2017.07.036
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.

Structure based drug design of Pim-1 kinase
followed by pharmacophore guided synthesis
of Quinolone-based inhibitors
a
LubnaSwellmeen , Rand Shahin , Yusuf Al-Hiari , Amani Alamiri , Alaa Hasan , Omar
b
c
b
a
d
Shaheen
a
Department of Pharmaceutical Sciences, Faculty of Pharmacy, Zarqa University,
Azzarqa, Jordan.
Email: lswellmeen@zu.edu.jo
b
*Corresponding Author,
Department of Pharmaceutical Chemistry, Faculty of Pharmaceutical sciences, Hashemite
University,
Az-zarqa, Jordan.
Telephone: 00962 799799044.
Email: r.shahin@hu.edu.jo
c
Department of Pharmaceutical sciences, Faculty of Pharmacy, University of Jordan,
Amman, Jordan.
hiary@ju.edu.jo
d
Department of Pharmacology, Faculty of Medicine, University of Jordan,
Amman, Jordan.
Telephone: 00962 5548511.
Email: oshaheen@ju.edu.jo
1

Abstract
Over expression of Human Phosphatidyl inositol mannoside kinases isoform 1 (Pim-1 kinase)
has been reported in several leukemia and solid tumors. Our continuous interest to reveal the secrecies
of the mysterious Pim-1 kinase binding pocket has led us to employ a structure based drug design
procedure based on receptor-ligand pharmacophore generation protocol implemented in Discovery
Studio 4.5 (DS 4.5). Subsequently, we collected 104 crystal structures of Pim-1 kinase from the Protein
Data Bank (PDB) and used them to generate pharmacophores based on the anticipated co- crystallized
ligand-Pim 1 kinase receptor interactions. All selected pharmacophoric features were enumerated and
only those that had corresponding valuable receptor-ligand interactions were retained. This was
followed by modeling all pharmacophore combinations and scoring them according to their Receiver
Operating Characteristic (ROC) curve analysis parameters as well as a DS.4.5 built-in Genetic
Function Algorithm (GFA) validating model. Accordingly, 111 pharmacophores resulted with
acceptable ROC performances; 1XWS_2_04, 2BIK_2_06, and 1XWS_2_06 (ROC AUC value of:
0.770, 0.743 and 0.741 respectively) were the best pharmacophores. These pharmacophores were
employed to guide the synthesis of new series of 7-[(2-Carboxyethyl) amino]-1-substituted-6-fluoro-8-
nitro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid and their reduced 8-amino derivatives. The
synthesized compounds were later evaluated for their Pim-1 kinase inhibitory potencies. Of which the
most potent illustrated an IC50 value of 0.29 µM against Pim-1 kinase.
Key words:
Pim 1 kinase, Discovery Studio, Binding pocket, Pharmacophores, Receptor, Quinolone.
2

Introduction:
Human phosphatidyl inositol mannoside kinases (Pim kinases) are a family of serine/threonine
kinases composed of three members; Pim-1, Pim-2 and Pim-3. Pim kinases proto oncogenes regulate
several signaling pathways which are directly related to the development of numerous malignancies.
The Pim-1 kinase isoform (a 313-amino acid kinase) has been specifically reported in the last two
decades for its direct role in tumor genesis process in various hematological malignancies, such as
1-3
lymphoma and acute myeloid leukemia. Similarly, over expression of Pim-1 has been linked to
2-7
several solid tumors such as urothelial and prostate carcinomas. Furthermore, developing effective
Pim-1 inhibitors is also important to overcome the Pim kinase promoted chemo-resistance of tumor
4, 8
cells through hypoxia-induced chemotherapy resistance. Per se, the Pim-1 is considered as a very
attractive target for pharmacological inhibition in cancer therapy.
The crystal structure of Pim-1 kinase adopts a two-lobe kinase fold; the N-terminal and the C-
terminal. The two lobes are combined with the hinge region amino acids residues, comprising,
Glutamine 121, Arginine 122, Proline 123, Glutamine 124 and Valine 126, this hinge region also
extends to the Aspartate 128 and Aspartate 131 residues. Along the hinge region resides is the deep
Adenosine triphosphate (ATP) binding pocket which encompasses, the phosphate binding loop lying
on the top of the ATP binding pocket like a lid (P-loop: residues 46-54), the catalytic loop (C-loop:
1-2
residues 166-170), and the activation loop (A-loop: residues 191-202). Designing new selective
kinase inhibitors has been always an issue due to the high degree of conservation in the protein amino
acid sequence among different kinases, but the Pim-1 kinase binding pocket is known to have some
distinctive features that can be exploited, for example; Pim-1 kinase binding pocket is considered
atypical in many aspects; this is due to the presence of the Proline 123 moiety in the hinge region
7
, 9
where adenine usually binds via two hydrogen bonds. As well, the hydrophobic region I (BR-I) in
Pim-1 kinase is distinctively different from other kinases'. So, revealing more possible binding
3

interactions inside the binding pocket of Pim-1 kinase will open the road towards discovering more
selective and specific new inhibitors against this target.
We believe that illuminating the secrecies of the mysterious Pim-1 kinase binding pocket and
understanding the Pim-1 versus ligand interaction models is of great importance for determining the
potential activities of novel Pim-1 kinase inhibitors and designing new ones. In our previously
7
published research project, we designed a QSAR guided ligand-based study for Pim-1 kinase enzyme,
and we were able to reveal some of the steric and electronic features that represent the optimal
interactions between Pim-1 kinase inhibitors and its binding pocket from the resulted QSAR successful
equation. Our interest to reveal even more ambiguous points inside the binding pocket of pim-1 kinase
has encouraged us to continue our investigation through another structure based study. Subsequently,
our aim of this work is to extract and interpret Pim-1 kinase co-crystallized ligands from the Protein
Data Bank (PDB) focusing on X-ray structure complexes, which have a satisfactorily good resolution.
A PDB complex does not only contain experimental data but also a reasonable amount of interpretation
1
0
that can be done by a researcher. In most cases, structure determination efforts focus on the
macromolecule (Pim-1 kinase protein, for example), while in our work ligand structures were
elucidated then the potential interactions between the ligand and the Pim-1 kinase binding pocket were
clarified and explained.
Furthermore, the most successful pharmacophores that resulted from modelling the ligand-
receptor interactions were engaged as guiding maps to design new 7-[(2-Carboxyethyl)amino]-1-alkyl-
6-fluoro-8-nitro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid derivatives as novel inhibitors against
Pim-1 kinase enzyme. The present study is aiming to explore the key structural requirements for Pim-1
kinase inhibition utilizing combination of the above modeling methods for the design of novel,
selective Pim-1 kinase inhibitors.
4

2
. Results
Receptor-ligand
pharmacophore
generation
technique
models
drug–receptor
interactions using information derived from co-crystallized ligand contact points with their
11
receptors. HYPOGEN detects a 3D collection of five to six chemical features that reflect
the hot contact points between the co-crystallized ligand and the binding pocket in which it
was originally entrenched, which delivers a relative configuration for each co-crystallized
ligand consistent with their binding modes inside the receptor site. A total of 104 X-ray Pim-
1
kinase crystal structure complexes were used in this study. 111 training Hypogens
pharmacophores) were generated, of which only pharmacophores showed acceptable
results upon validation with Receiver Operating Characteristic (ROC) analysis as well as a
DS.4.5 built-in Genetic Function Algorithm (GFA) validating model. The successful
(
7
7
pharmacophores were then engaged as guiding maps to design new quinolone-based series of
Pim-1 kinase inhibitors.
2.1 Literature survey and data mining
The ChEMBL data base was surveyed to collect as many X-ray crystal structure complexes
for Pim-1 kinase as possible. Pim-1 kinase target (CHEMBL2147) revealed 104 crystal structures in
the Protein Data Bank (http://www.rcsb.org/) please refer to SM-1 in supplementary material for the
names of all PDB crystal structures used. Altogether, the 104 X-ray crystal structure complexes were
collected, cleaned, and then engaged in consequent modelling processes.
2.2 Generating receptor-ligand pharmacophores
After employing the receptor-ligand pharmacophore generation procedure, only 111
pharmacophores resulted in a ROC-AUC value above 0.5. Table 1 shows the PDB code of the Pim-1
kinase crystals that displayed accepted results in addition to their resolution values and the structure of
their native co-crystallized ligands. Moreover, Table 1 shows a description of the enumerated ligand
features during the first step of the receptor-ligand pharmacophore generation procedure, the features
5

that matched the receptor-ligand interactions, and the number of pharmacophores generated from each
crystal structure during the receptor-ligand pharmacophore generation step. Additionally, Table 2
shows a description of the 111 pharmacophores generated in addition to their validation results in the
ROC analysis procedure, i.e. (Number of true positives, Number of true negatives, Number of true
positives, Number of true negatives, Sensitivity and Specificity).
2.3 Assessing the Validity of the resulted pharmacophores
In each pharmacophore generation run the resulting pharmacophores were directly graded
according to their ROC performance and their GFA model results
2.3.1 Receiver Operating Characteristic (ROC) Curve
ROC analysis aims towards analyzing the ability of a particular pharmacophore to sort a total
list of already known actives and in-actives into their correct classes. The ROC performance is usually
illustrated by the area under the curve (AUC) of the corresponding ROC along with other parameters
such as: No. of true positives, No. of true negatives, No. of false positives, No. of false negatives,
1
2-14
Sensitivity and specificity.
Table 2 shows the ROC analysis results of our selected passed pharmacophores. Figures 1A
and 1B show the ROC curve of the pharmacophores with best ROC results (i.e.: Hypo 1XWS-2-4
(AUC value of 77.0 %), Hypo 2BIK-2-6 (AUC value of 74.3 %), Hypo 1XWS-2-6 (AUC value of 74.1
%), Hypo 3DCV-2-3 (AUC value of 71.1 %), Hypo 3BGZ-2-3 (AUC value of 71.2 %), Hypo 3DCV
(AUC value of 71.10 %), and Hypo 2BIK-2-7 (AUC value of 70.0 %)).
The fair results of ROC curves are attributed to the fact that the AUC calculation method
during the ROC analysis process is dependent on the method used for curve fitting. ROC-
AUC does not account for occurrence or different misclassification costs coming from false-
15
negative and false-positive analyses;
meaning that using the pharmacophore mapping
procedure in the early steps of ROC analysis to identify the actives from the in-actives does
not in some cases give the right results. For example, in this case; the pharmacophore
mapping procedure in DS 4.5 doesn’t take into account the Pi-anion interactions (Anion-π
interactions are defined as favorable non-covalent contacts between electron-rich anions and
6

electron-deficient aromatic systems (π-acid)). This means that some truly active Pim-1
kinase inhibitors will be falsely un-mapped in the very early steps of the ROC analysis and
so, all the later results will be in some way undervalued since some truly active pim-1 kinase
inhibitors were excluded during the early steps of ROC analysis procedure. For this reason,
this explanation encouraged us to accept the fairly good ROC analysis results and to use the
DS.4.5 built in GFA model as an additional method to support our pharmacophore validation
procedure.
2.3.2 DS.4.5 built in GFA model designed to score the resulted pharmacophores for the selectivity
depending on the features and the 3D locations
The GFA model has an excellent prediction power for pharmacophoric selectivity, each
pharmacophore resulted from the receptor-ligand pharmacophore generation step is validated according
to the selectivity score based on a predictive model that uses simple descriptors derived from the type
and relative location of the pharmacophore features to predict the number of hits from a diverse
16
database, (please refer to the experimental part: 4.1.5.2 : DS.4.5 built-in GFA validation model).
As we mentioned previously we used both the AUC-ROC curve criteria and the GFA
selectivity score to choose the best pharmacophores that can specifically describe the inhibitory binding
interactions inside the binding pocket of Pim-1 kinase. Table 3 displays the success criteria of the
superior pharmacophoric hypotheses in terms of both GFA selectivity and ROC-AUC validation
procedures. The bolded pharmacophores are the ones that attained the best validation criteria among all
the resulted 111 pharmacophores. Table 4 shows the pharmacophoric features and corresponding
weights, tolerances and 3D coordinates of successful selected pharmacophores: (Hypo 2BIK-2-6 (GFA
selectivity value of 9.07), Hypo 1XWS-2-6 (GFA selectivity value of 9.07), Hypo 1XWS-2-4 (GFA
selectivity value of 9.98), Hypo 3DCV-2-3 (GFA selectivity value of 9.50), Hypo 3BGZ-2-3(GFA
selectivity value of 5.95), Hypo 2BIK-2-7 (GFA selectivity value of 8.59) and Hypo 3DCV-2-10 (GFA
selectivity value of 7.98). And finally, Figures 2 and 3 summarize all pharmacophore extraction
processes from each successful crystal ligand, and show the structural features of every successful
pharmacophore.
7

2.4 Pharmacophore-Guided Synthesis of Novel Pim-1 kinase inhibitors
By comparing the features of pharmacophores, Hypo 1XWS_2_04, Hypo 1XWS_2_06,
and Hypo 2BIK_2_06 (ROC AUC value of: 0.770, 0.743 and 0.741 respectively) (Figures:
C, 2F, and 2I) with the other less superior pharmacophores (Figures: 2L, 3C, 3F, and 3I),
2
we realized that the positively charged pharmacophoric feature that appeared seldom in the
most superior pharmacophores is of a special value during the Pim-1 kinase inhibitory
interaction inside the binding pocket of Pim-1 kinase. And by further analysis of the Pim-1
kinase binding pocket we concluded that the positively charged feature corresponds to the
interaction between the positively charged ligand and the amino acids Asp128 and Asp131
in the hinge region extension part of Pim-1 kinase binding pocket. This conclusion has
prompted us to synthesize an electron deficient Quinolone core that is able to confirm anion-
π interactions with the matching Aspartic acid moieties (Asp128 and Asp131) (Please refer
to Figure 4 which shows detailed perspective about the important interactions inside this
10
binding pocket of Pim-1 kinase).
Anion-π interactions are defined as favorable non-covalent contacts between electron-rich
anions and electron-deficient aromatic systems (π-acid). Designing an electron deficient core
for Pim-1 kinase inhibitors will confirm positive-negative interactions in the newly designed
molecules without the need to steer the positive charge for
a
definite location and
1
7-19
coordinate.
Furthermore, we integrated a carboxylic acid moiety at position number 3 of
the Quinolone core to enhance the hydrogen bonding interaction in this region in order to
assure the inhibitory effect of the ligands.
Additionally, the promising parameters of pharmacophores (Hypo 1XWS-2-4, Hypo
1
XWS-2-6, Hypo 2BIK-2-6, Hypo 2BIK-2-7, Hypo 3BGZ-2-3, Hypo 3DCV-2-3, and Hypo
DCV-2-10) (Figures 2C, 2F, 2I, 2L, 3C, 3F, and 3I respectively) regarding their ROC and
3
GFA performances encouraged us to engage them as templates to design novel Pim-1 kinase
inhibitors which led us to envisage 6-fluoro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid
based analogues as potential Pim-1 kinase inhibitors (Please refer to Figure 5 which shows
8

the major synthons used to synthesize new inhibitors against Pim-1 kinase). Also, refer to
Figure which shows designed compound 39-B (8-amino-7-((2-carboxypropyl)amino)-6-
fluoro-1-(4-fluorophenyl)-4-oxo-1,4-dihydroquinoline-3-carboxylic acid) mapped against:
6
Hypo 2BIK-2-6, Hypo 1XWS-2-6, Hypo 3DCV-2-3, Hypo 3BGZ-2-3, Hypo 2BIK-2-7, and
Hypo 3DCV-2-10. The 6-fluoro, 4-oxo, 3-carboxylic substitutions are considered electron
withdrawing deactivating groups that can confer the Quinolone core even more electronically
deficient and able to attract the nearby negatively charged Aspartic acid via an ionic Anion-π
interactions.
Accordingly, we can say that the 6-fluoro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid
synthons (Figure 5) combine two advantages: (I) Simplistic synthesis method, and (II)
approachability to wide-ranging functionalization required to accomplish the feature-rich
nature of the successful pharmacophores. (Hypo 1XWS-2-4, Hypo 1XWS-2-6, Hypo 2BIK-
2-6, Hypo 2BIK-2-7, Hypo 3BGZ-2-3, Hypo 3DCV-2-3, and Hypo 3DCV-2-10) (Figures
2C, 2F, 2I, 2L, 3C, 3F, and 3I respectively).
Consequently, we prepared 6 Quinolone-based analogues (Table 5), Synthesis proceeded
via nucleophilic aromatic replacements of chloro substituents on synthons A, B, C, and D
Figure 5) by β-alanine, DL- Aspartic acid, or isomethyl β-alanine, as outlined in Schemes 1
(
and 2. The synthesized compounds were tested their bioactivities. Table 5 shows the
synthesized hit molecules with their docking scoring function values against Pim-1 kinase
(PDB code: 3BGZ, resolution 2.4 Å).
2.5 Comparison of Quinolone-based analogues with the active site of Pim-1 kinase
Quinolone-based analogues were further emphasized through their analogy within the binding
pocket of Pim-1 kinase, i.e., upon analyzing the features of compound 40-B (Table 5) and how it docks
within Pim-1 kinase (PDB code: 1XWS, resolution 1.8 Å), as in figure 7.
Particularly, The Para substituted halogen phenyl group at nitrogen number 1 of the
Quinolone backbone (Please see compounds 38, 39-B, and 40-B in table 5); undergoes clear
Van der Waals interactions in the binding pocket of Pim-1 kinase mainly with the amino acid
residue Leucine 44 (Leu44), Please See figure 7C which illustrates the hydrophobic
9

interactions between compound 40-B and the binding pocket of Pim-1 kinase). On the other
side, the 7-ethylamino side chain was also added to the Quinolone back bone and this also
matches more hydrophobic interactions in the successful pharmacophores and the binding
pocket of Pim- kinase (please refer to figure 7B to see the fitting between the ethyl amino
group in compound 40-B and the hydrophobic feature in pharmacophore Hypo 1XWS-2-6).
Then also, (refer to figure 7C and 7F to see the hydrophobic interaction between the docked
compound 40-B and the amino acid Valine 126 (Val 126).
Mapping the Quinolone core of the newly synthesized ligands against PosIon feature in
Hypo 1XWS_2_04, Hypo 1XWS_2_06, and Hypo 2BIK_2_06 (Figure 7B) corresponds to
electrostatic attraction interactions tying the Quinolone electron deficient core with the
carboxylic acid of Asp131 (Figure 7A, 7D, 7E, and 7F). Similarly, mapping the carboxylic
acid at position number 3 of the ligand's Quinolone core against hydrogen bond acceptor
feature in Hypo 1XWS_2_06 (figure 7B) corresponds to conventional hydrogen bond
interactions involving this hydrogen bond acceptor group with the side chain of Asp121
(figures 7A, 7D, 7E, and 7F). Also, 7-ethylamino and 8-amino groups in compounds 39-B
and 40-B satisfy the hydrogen bond donor functionality in Hypo 3DCV-2-3, Hypo 2BIK-2-7
and Hypo 3DCV-2-10 (please refer to figures 6C, 6E, and 6F). Meanwhile, the oxy group on
the quinolone back bone satisfy the hydrogen bond acceptor functionality in Hypo 2BIK-2-
6, Hypo 1XWS-2-6, Hypo 3BGZ-2-3, and Hypo 2BIK-2-7 (Please refer to figures 6A, 6B,
6D, and 6 E).
2.6 In vitro Evaluation for the synthesized compounds
The synthesized pure compounds were evaluated in vitro against human recombinant
Pim-1 kinase assay kit (Cyclex®, Japan). These compounds were further assessed to
determine their IC50 values. Our six compounds exhibited measurable anti-Pim-1 kinase
bioactivities therefore they were evaluated to determine their IC50 values. Figure L under
supplementary material illustrates the dose/response plots of the active hits.
Table 5 shows the active hits (35-40) and their measured in-vitro activities. Figure 7
shows the most active hit 40-B (IC50 = 0.2950 µM) docked into the binding pocket of Pim-1
1
0

kinase (PDB code: 1XWS, resolution 1.8 Ǻ) and by referring to this figure it is very clear
that the 6-fluoro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid backbone confers very
important contacts with Asp128 and Asp131 through the Anion-π interactions and this result
matches very well our first assumptions. In addition to that the Val 126 which is part of the
1
hinge region binds via Fluorine hydrogen bonding with the Florine atom at position number
6
on the Quinolone core. All our synthesized compounds share the same Quinolone core
please refer to table 6 for the Alignment and similarity analysis between our synthesized
compounds.
Unexpectedly, upon performing the docking experiment on compound 39-B (IC50
.335 µM) into the binding pocket of Pim-1 kinase crystal structure 1XWS no results were
achieved although the structure of compound 39-B and compound 40-B are very similar;
compound 39-B is 8-amino-7-((2-carboxyethyl)amino)-1-(4-flouroophenyl)-6-fluoro-4-oxo-
,4-dihydroquinoline-3-carboxylic acid (Table 5). Meanwhile, compound 40-B is 8-amino-7-
(2-carboxypropyl)amino)-6-fluoro-1-(4-clorophenyl)-4-oxo-1,4-dihydroquinoline-3-
=
3
1
(
carboxylic acid ( Table 5, Figure 7). Docking compound 39-B into the crystal structure of
Pim-1 kinase (PDB code: 3DCV, resolution 2.7 Ǻ) (Figure 8) revealed very different binding
mode for this compound. Actually, it seems that compound 39-B flips upside down inside
the binding pocket of Pim-1 kinase (Please see figure 8A which shows the binding mode of
compound 39-B inside the Pim-1 kinase binding pocket).
The Florine atom located at the para position of the phenyl group on the 6-fluoro-4-oxo-
1,4-dihydroquinoline backbone in compound 39-B (Table 5) acts as
a hydrogen bond
2
0
acceptor, this fact has been published and approved by Murray-Rust et al in 1983.
The
Florine directed favorable interaction positions compound 39-B inside the binding pocket of
Pim- kinase so that the Asp131 is away from the Quinolone core. Accordingly, the Asp131
will have two kinds of interactions with the Florine-phenyl group: (i) Halogen hydrogen
bonding interaction with Florine atom. (ii) Pi-Anion interaction with the electron deficient
phenyl group. (Please refer to figures 8A - 8 F, that show 2D and 3D illustration of these
interactions).
1
1

Our previous analysis that focuses on the different binding modes of compounds 39-B
and 40-B inside the binding pocket of Pim-1 kinase explains the difference in their in vitro
inhibitory activity against Pim-1 kinase, the IC50 value for compound 39-B is 3.335 µM
meanwhile, the IC50 value for compound 40-B is 0.2950 µM. Moreover, compound 38
(
Table 5) which also holds the Florine-phenyl substituent on its structure exhibits similar in
vitro activity to compound 39-B (IC50 value of 3.387 µM, Table 5). Compounds 35 and 36
Table 5) exhibits inferior anti-pim-1 kinase activity and this can be explained by the short
(
side chains that are attached to the nitrogen number 1 on the Quinolone core of these
compounds. Finally, compound 37 exhibits good anti-Pim-1 activity (IC50 value of 2.124
µM, Table 5) although it holds a short side chain on the nitrogen number 1, and this can be
explained by the internal lactonization on the ethyl amine side chain that might happen to
this compound under physiological pH which might lead to new binding mode for this
compound inside the binding pocket of Pim-1 kinase.
3.0 Conclusion
Our structure based receptor-ligand drug design of Pim-1 kinase proteins revealed
unanticipated mode of recognition in the peri-hinge area of Pim-1 kinase. Unlike other kinases, Pim-1
kinases possess a hinge region which creates a unique binding pocket for ATP. Crystal structures of the
1
Pim-1 kinase has been reported by more than a few labs. And we have exploited the binding mode of
these crystal structures and their co-crystalized ligands to anticipate and design new Pim-1 kinase
inhibitors. In our previous ligand based study we showed that electron rich polycyclic structure is
needed to interact with Ile185, Val 52, Lys 67 (Figure 4). And now our structure based drug design
revealed that electron deficient aromatic system is important to anchor pim-1 kinase inhibitors at this
site of Asp128 and Asp131 through Pi-anion interactions (Figure 4). 111 pharmacophores resulted with
acceptable ROC performance; 1XWS_2_04, 2BIK_2_06 and 1XWS_2_06 (ROC AUC value of:
0.770, 0.743 and 0.741 respectively) were the best pharmacophores. These pharmacophores were
employed to guide the synthesis of new series of 7-[(2-Carboxyethyl) amino]-1-substituted-6-fluoro-8-
nitro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid and their reduced 8-amino derivatives which were
later evaluated for their Pim-1 kinase inhibitory potencies. Of which the most potent illustrated an IC50
value of 0.29 µM against Pim-1 kinase.
Acknowledgments
1
2

This project was sponsored by the Deanship of Scientific Research in Zarqa University
and the Deanship of Scientific Research in Hashemite University. The authors thank the
Faculty of pharmaceutical sciences in Hashemite University for their generous help and
support.
4
4
4
. Experimental
.1 Molecular modeling
.1.1 Computational software
The following software were applied:
•
CS ChemDraw Ultra 6.0, Cambridge Soft Corp. (http:// www.cambridgesoft.Com), USA.2D
Structure drawing was performed employing.
•
Discovery Studio 4.5 (DS 4.5) Standalone Applications, including Catalyst for analysis of
pharmacophore data, CHARM for analyzing molecular simulation, CNX for 3D-structure
determination and analysis of macromolecules, MODLER for analyzing protein structures, and
LigandFit for molecular modeling and docking Biovia ® (www.3ds.com),USA.
Accelrys Enterprise Platform Server (AEP) (www.3ds.com), USA
•
•
. The crystal structures of Pim-1 kinases were obtained from the protein data bank
(http://www.rcsb.org/)
4.1.2 Literature survey and Pim 1 crystal structures search
Our first goal in this project was to gather as much information as possible about Pim-1 kinase
as a drug design target and to satisfy this goal we explored the Chemical European Bioinformatics
Database (ChEMBL, https://www.ebi.ac.uk/chembl/) for Pim-1 kinase. The information in (ChEMBL)
revealed the target CHEMBL2147 which is a single protein Serine/threonine Pim-1 kinase derived
from Homo sapiens. CHEMBL2147 showed 104 crystal structures of Pim-1 kinase in the Protein data
bank (http://www.rcsb.org/) and all of these crystal structures were explored in next drug design steps
(Please refer to SM-1 under supplementary material for the PDB codes of Pim-1 kinase crystal
structures used in the modelling step).
1
3

4.1.3 Preparing proteins for modeling
All Pim-1 kinase crystal structures that were downloaded from PDB were prepared for
16
modeling using the protocol "Prepare protein" implemented in DS 4.5 protein preparation step
includes removal of alternate conformations, removal of hetero-atoms from a protein structure, adding
hydrogens, correcting missing or incorrectly specified residues, detecting valency violations and
correcting them by changing the bond type or changing the formal charge.
4.1.4 Generating receptor-ligand pharmacophores
Selective pharmacophoric models representing ligand-receptor interaction are derived from
co-crystalized molecules and their surrounding amino acids were generated using the protocol
16
"
Receptor-ligand Pharmacophore Generation". The procedure is briefed as in five steps:
First step: Enumeration of pharmacophoric features on the original crystalized ligand:
The original crystalized ligand is scanned within its binding pocket structure for distinctive
pharmacophoric features. The pharmacophoric features that have been considered in this step were:
Hydrogen Bond Donor/Acceptor (HBD/HBA), Hydrophobic (HYD), Positive/Negative Ionizable
16
(PI/NI), and Ring Aromatic (RA).
Second step: Identifying features in the crystalized protein that match the receptor-ligand interactions:
The pharmacophoric features that match the receptor-ligand interactions are identified by tying
Hydrogen Bond Donors/Acceptors in the crystalized ligand with their Bond Donors/Acceptors pairs in
the surrounding amino acids within a distance of 3.0 Å. Unmatched hydrogen bonds in the already
1
1
scanned ligand in step 1 will be removed. On the other hand, hydrophobic features on the scanned
11
ligand within 5.5 of the centroid of a hydrophobic residue ( Alanine (Ala), Cysteine (Cys), Isoleucine
Ile), Leucine (Leu), Methionine (Met), Phenylalanine (Phe), Valine (Val) will be retained. As for
(
11
Positive/Negative Ionizable groups: If the charge interaction is within 5.6 Å , the feature is retained.
All others are removed. Finally, Ring Aromatic: Retained if the protein has an aromatic ring within 2.5
1
6
Å from the projection point of the RA on the ligand.
Third step: Building the of pharmacophore models
1
4

During this step, the pharmacophores are assembled. To assure the selectivity of the
pharmacophores the number of pharmacophoric features will be specified to range from four to six. If
the total of matching features for a protein complex is much greater than six, then there may be too
many candidates. The basic goal of building a pharmacophore model is to search for multiple ligands
that can bind to the same receptor. Therefore, the model should be sensitive enough to find potential
novel ligands that can bind to the protein. However, it needs to be selective so that the false positive
16
candidates are avoided.
Fourth step: Adding exclusion volumes or shape constraints
Exclusion volumes are added to the pharmacophoric models by default. Addition of the
exclusion volumes is based on inserting the exclusion volumes in the pharmacophoric map where ever
there are a residue atoms 4.1 to 5.0 Å away from the ligand. The excluded volumes are placed at the
centroid of the atoms and the size is proportional to the number of atoms.
Fifth step: Ranking the pharmacophores and retaining the most selective ones.
Screening a small diverse 3D database, CapDiverse, which is distributed with DS 4.5 is used
as a direct method for ranking the selectivity of the resulted pharmacophore. The greater the hits
captured, the less selective the pharmacophore.
4.1.5 Assessing the Validity of the resulted pharmacophores
During the modeling process two DS 4.5 built-in optional validation methods were employed
in order to sort the resulted pharmacophores according to their selectivity, specificity and predictivity.
These two methods are: The Receiver Operating Characteristic (ROC) Curve Analysis and the DS.4.5
built in Gentetic Function Algorithm (GFA) model designed to score the resulted pharmacophores for
the selectivity depending on the features and the 3D locations. The validation score is based on a
predictive model that uses simple descriptors derived from the type and relative location of the
pharmacophore features to predict the number of hits from a diverse database
5.1.5.1 Receiver Operating Characteristic (ROC) Curve
ROC curve is used to indicate how often a model correctly identifies true positives and true
negatives. This procedure is based upon challenging the resulted successful pharmacophores to
1
5

selectively identify diverse Pim-1 kinase active inhibitors from a big pool of large actives and similar
decoys using the ROC examination protocol implemented in DS 4.5.
ROC analysis procedure delivers some success criteria for evaluation: (i) Area Under the Curve
(AUC) of the corresponding ROC curve, (ii) Number of true positives (TP) (iii) Number of true
negatives (TN) (iv) Number of false positive (FP) (v) Number of false negative (FN) (vi) specificity
11, 14, 21
(Sp., true negative rate), (vii) sensitivity (Se., true positive rate)
.
Our test set is composed of 1030decoy compounds and 25 previously known pim-1 kinase inhibitors
12, 22-27
(Please see Table A under supplementary material for the structures of Pim-1 kinase inhibitors
utilized as actives in decoy list). This set was assembled using the protocol "Find Similar Molecules" in
1
3
DS 4.5, i.e. the 25 Pim-1 active were used to find similar decoy compounds from the zinc library the
following 2D descriptors were applied during finding similar molecules step: molecular weight,
number of hydrogen bond donors and acceptors, number of rotatable bonds, log p, number of rings
count and others. lastly, both the active diverse inhibitors and the similar zinc decoy compounds were
combined together to form the decoy testing set.
Later on, the testing set was screened by each resulted pharmacophore throughemploying the
validation option in Receptor-Ligand Pharmacophore Generation protocol implemented in DS 4.5.
The ROC curve analysis describes the sensitivity (Se or true positive rate, Eq. (1) as a function of
12-13
(1-Sp). Sp is defined as specificity or true negative rate (Eq. (2)).
Numberof SelectedActives
TP
Se=
=
TotalNumberof Actives TP+ FN
……………………...……. (1)
Number of Discarded Inactives
Total Number of Inactives
TN
Sp =
=
……………….
……… (2)
TN + FP
The success of a certain pharmacophore in selecting the active compounds and discarding the
1
3-14, 27
inactive ones can be judged from the Area under the ROC curve (AUC).
In an ideal ROC curve
an AUC value of 1 is attained; however, random distributions cause an AUC value of 0.5. Virtual
screening that performs better than a random discrimination of actives and decoys retrieve an AUC
value between 0.5 and 1, however an AUC value lower than 0.5 signifies the unfavorable case of a
1
6

virtual screening method that has a higher probability to assign the best scores to decoys than to
7
, 13-14
actives.
5.1.5.2 DS.4.5 built-in GFA validation model
DS.4.5 built-in Genetic Function Algorithm(GFA) model was used to score the resulted
pharmacophores for the selectivity depending on the features and the 3D locations. Obviously, the total
number of computed pharmacophore models can be quite large since the total number of matching
features is already large. The ROC method for ranking the resulted pharmacophores in is not
satisfactory when used alone. Therefore, a GFA model is used as a rule-based scoring function for the
1
6
selectivity depending on the features and the 3D locations. .This GFA model is employed within the
receptor-Ligand Pharmacophore Generation protocol when using the Database option of the Advanced
|
Selectivity Scoring parameter.
The GFA model for predicting the selectivity of a pharmacophore was built from a training set
of 1544 pharmacophore models, each with two to eight features. This set is used for searching a
database of 5384 diverse, druglike molecules. The database consists of 3000 drug-like compounds
randomly selected from the Bio info database (http://bioinfo-pharma.u-strasbg.fr/bioinfo) and 2390
selected from the CAP database. The drug-like criteria used were: molecular weight of 150-550 D, 0-
15 rotatable bonds, 0-10 HBAs, 0-5 HBDs, fewer than three charged atoms, and Polar surface area less
than 150 Ų.
The logarithmic values of the number of database search hits are used as the targets. The
following log values are used for zero hits based on the number of features: Two to five features:
log(0.3), Six features: log(0.1), Seven Features: log(0.03), and Eight features: log(0.01). The number of
total features in the pharmacophore models and the feature-feature distance bin values are used as the
descriptors for training the GFA model. The GFA model has excellent prediction for selectivity.
4.2 Pharmacophore-Guided Synthesis of Novel Pim-1 kinase inhibitors
1
7

Our argument that pharmacophores: (Hypo 2BIK-2-6, Hypo 1XWS-2-6, Hypo
DCV-2-3, Hypo 3BGZ-2-3, Hypo 2BIK-2-7 and Hypo 3DCV-2-1) are of promising
3
parameters regarding their ROC and GFA performances, encouraged us to engage them as
templates for designing novel Pim-1 kinase inhibitors. The successful pharmacophores
(Tables 2 & 3, Figures 2C, 2F, 2I, 2L, 3C, 3F, and 3I) were used as guiding maps to design
new compounds with proposed good affinity towards the Pim-1 kinase binding site. Figure 6
show our designed compound 39-B mapped against the successful pharmacophores (Hypo
2
BIK-2-6, Hypo 1XWS-2-6, Hypo 3DCV-2-3, Hypo 3BGZ-2-3, Hypo 2BIK-2-7 and Hypo
DCV-2-10). Besides, since these pharmacophores were originally extracted from the Pim-1
3
kinase binding pocket by the "Receptor-ligand Pharmacophore Generation" method, we can
assume that our newly designed compounds would exert acceptable affinity against the Pim-
1
kinase binding pocket.
Furthermore, the fairly good ROC performance of pharmacophores Hypo 1XWS-2-
and Hypo 2BIK-2-6, (AUC values of 0.770 and 0.743 respectively) encouraged us to
6
further investigate their structures (Figures 2C and 2I respectively) which exhibit three
hydrophobic wings and extra positively ionizable group that is only found in these two
distinctive pharmacophores. This prompted us to envisage quinolone-based analogues as
potential Pim-1 kinase inhibitors. Electron withdrawing groups such as (halogen, nitro and
oxo groups) were added to the quinolone back bone to confer it more electron deficient and
able to match the positively ionized group in our superior pharmacophores through Anion-π
interactions. Accordingly, our newly designed 7-[(2-Carboxyethyl) amino]-1-substituted-6-
fluoro-8-nitro-4-oxo-1,4-dihydro-quinoline-3-carboxylic acid derivatives and their reduced
8
-amino analogues would have the accessibility to diverse functionalization necessary to
satisfy the feature-rich nature of the successful pharmacophores as shown in figure 6 which
shows designed compound 39-B (8-amino-7-((2-carboxypropyl) amino)-6-fluoro-1-(4-
fluorophenyl)-4-oxo-1,4-dihydroquinoline-3-carboxylic acid) mapped against: (A) Hypo
BIK-2-6 (B) Hypo 1XWS-2-6 (C) Hypo 3DCV-2-3 (D) Hypo 3BGZ-2-3 (E) Hypo 2BIK-
-7 (F) Hypo 3DCV-2-10.
2
2
1
8

4.3 Docking experiments confirming the pharmacophore guided synthesis of
Quinolone-based derivatives
Figure 6 shows the preliminary possibilities for the structural mapping between our
designed compounds and the successful pharmacophores resulted from the Receptor-ligand
pharmacophore generation step. The pharmacophore mapping was not enough to represent
the affinity of our compounds towards the binding pocket of Pim-1 kinase enzyme since we
based our drug design process on the anion-π interactions between the electron deficient
quinolone-based back bone and the positively ionized group of the superior pharmacophores:
Hypo 1XWS-2-6 and Hypo 2BIK-2-6, but this interaction can't be mapped through simple
ligand pharmacophore mapping procedure implemented in DS 4.5.
So, in order to inspect the fitting of our designed quinolone-based derivatives (Table
5) into the binding pocket of Pim 1 kinase and to visualize the proposed anion-π interactions
between the quinolone core and the amino acids located at the binding pocket of Pim-1
kinase, we docked our newly designed molecules using the ligandfit docking algorithm
implemented in the DS 4.5 into the binding pocket of the successful Pim-1 kinase crystal
structures, namely: (PDB code: 1XWS, resolution 1.8 Ǻ),(PDB code: 2BIK, resolution 1.8
Ǻ), (PDB code: 3BGZ, resolution 2.4 Ǻ), and (PDB code: 3DCV, resolution 2.7 Ǻ).
16, 28
The Ligandfit docking experiments consist of two major steps.
First: Specifying
the region of the receptor to be used as the binding site and Second: Docking ligands to the
specified site which itself consists of the following: site partition using the conformational
ligand/site shape matching to select ligand conformations that are similar to the shape of site.
This is followed by positioning the selected ligand conformation into the binding site and
rigid body energy minimization of the candidate ligand pose/conformation using the
DockScore energy function. Finally, pose-saving algorithm comparing the candidate pose
1
6, 28
with the other stored candidates to ensure that redundant (i.e., similar) poses are rejected.
Accordingly, we prepared
6
Quinolone-based analogues and tested their
bioactivities against Pim 1 kinase. Table 5 shows the prepared compounds, their predicted
docking scores.
1
9

4.4 Synthesis of Quinolone derivatives
General Synthetic procedures
All chemicals, reagents and solvents were of analytical/ synthetic grade were
purchased from Sigma-Aldrich and Acros. Nuclear magnetic resonance spectra (NMR) were
recorded on Bruker, Avance DPX-300 spectrometer. Infra-red (IR) spectra were recorded
using Shimadzu 8400F FT-IR spectrophotometer (KBr discs). Melting points (MP) were
determined in open capillaries on a Stuart scientific electro-thermal melting point apparatus,
and are uncorrected. High-resolution mass spectra (HRMS) were measured in positive or
negative ion mode using electrospray ion trap (ESI) technique by collision induced
dissociation on a Bruker APEX-4 (7 Tesla) instrument. Microanalyses were performed using
EuroVector Elemental Analyser, model (EA3000 A), Jordan University. Mobile phase
mixtures for TLC were: System (1): Chloroform: methanol: formic acid (CHCl3: MeOH:
FA) (94: 5: 1); System (2): CHCl3: MeOH: FA (90: 10: 1); System (3):Hexane: Ethyl acetate
(50:50); System (4): System (1): system (3) (50: 50).
The following abbreviations are used: singlet (s), doublet (d), triplet (t), doublet of
doublet (dd), doublet of doublet of doublet(ddd), doublet of triplet (dt), multiplet (m), broad
signal (brs); High resolution mass spectra were determined on a high-resolution Micro Q-Tof
apparatus (CRMP, Université Blaise Pascal, Clermont–Ferrand, France) or on a Waters Q-
Tof
2 apparatus (CRMPO, Université de Rennes, France). Chromatographic purifications
were performed by column chromatography using 40–63 ml silica gel. Reactions were
monitored by TLC using fluorescent silica gel plates (60 F254 from Merck).
5.4.1 Synthesis of (7-Chloro-1-Alkyl-6-fluoro-8-nitro-4-oxo-1,4-dihydroquinoline-3-carboxylic
acid) Analogues (35-40-B, Schemes 1 & 2):
Synthons A, B, C, and D Figure 5, were used as starting materials during the synthesis process. These
29
30
synthons were previously reported in literature by (Al-Hiari et al., 2008) , (Al-Hiari et al., 2014) ,
31
and (Al-Hiari et al., 2015) (Please refer to SM-2: Synthesis of Synthons A, B, C, and D and scheme
A under supplementary material).The synthesis of novel C-7-substituted derivatives of 1-alkyl-6-
fluoro-8-nitro-4-oxo-1,4-dihydroquinoline-3-carboxylic acid was proceeded via nucleophilic aromatic
2
0

replacements of chloro substituents on Synthons A, B, C, and D (Figure 5) by β-alanine, DL- Aspartic
acid, or isomethyl β-alanine as outlined in Scheme 1.
7-[(2-Carboxyethyl)amino]-1-cyclopropyl-6-fluoro-8-nitro-4-oxo-1,4-
dihydro-quinoline-3-carboxylic acid (compound 35, Table 5, scheme 1)
A stirred mixture of β-alanine (1.1 g, 12 mmol), synthon A (Figure 5) (1.0 g, 3 mmol) and
sodium hydrogen carbonate (1.5 g, 18 mmol) in 50 % aqueous ethanol (140 mL) was heated at 70-80
°
C for 6 days under reflux conditions. The mixture slowly developed alight yellow color that changed
into bright yellow, then into clear orange solution. The progress of the reaction was monitored by TLC,
and was completed within 6 days. The mixture was extracted with dichloromethane (2 x 50 mL). The
aqueous layer was cooled, its pH adjusted to 6-7 by addition of 3.5N HCl and extracted with CH
50mL). Further acidification of the leftover aqueous layer to pH = 1-2 gave the title compound as
yellowish solid which was collected by filtration, washed with cold water (2 x 10 mL), dried and
2 2
Cl
(
recrystallized from a mixture of chloroform and ethanol (1:1,v/v). 1H- NMR (300MHz, DMSO- d6 ):
1
H- NMR (300MHz, DMSO- d6 ): δ0.96 ( 2m, 4H, H
2
-2'/H
2
2
-3'), 2.61 (d, J= 5.70 Hz, 2H, CH -COOH),
3
3
.69 (m, 1H, H
-
1′), 4.30 (m, 2H, CH
2
-NH), 7.20 (d, J = 9.3 Hz, 1H, NH-CH
-CO H), 14.52 (br s, 1H, C(3)-CO
H ), 48.41 (d, JC-F=14.1 Hz,
-NH), 109.50 (C-3), 114.73 (d, JC-F = 22.57 Hz, C-5 ), 116.74 (d, JC-F =7.05 Hz, C-4a), 128.68
2
), 8.06 (d, JH-F = 13.8 Hz,
1
3
1H, H-5), 8.76 (s, 1H, H-2), 12.40 (br s, 1H, CH
2
2
2
H); C-NMR
3
(300MHz, DMSO, d6 ): δ 10.13 (C-2′/ C-3′), 40.55 (C-1′), 34.82(CH-CO
2
2
3
CH
2
2
1
(
C-8a), 135.673 (C-8), 138.71 ( d, JC-F= 13.8 Hz, C-7), 150.55 (d, JC-F= 249 Hz, C-6), 152.15 (C-
2
),165.4 (C(3)-CO
2
H), 175.49 (CH
2
-CO
2
H), 176.16 (C-4); HRMS (ESI, -ve): m/z [M-H]- C16
3
H13FN O
7
378.07430 requires: 378.07265.
7
-[(2-carboxyethyl)amino]-1-ethyl-6-fluoro-8-nitro-4-oxo-1,4-
2
9
dihydroquinoline-3-carboxylic acid ( compound 36, Table 5, Scheme 1)
2
1

A stirred mixture of β-alanine (2.0 g, 22.5mmol), synthon B (Figure 5) ( 2.0 g, 6.3 mmol) and sodium
hydrogen carbonate (3.0 g, 35.6 mmol) in 50 % aqueous ethanol (200 mL) was heated at 65-70 °C for
7-8 days under reflux conditions. The mixture slowly developed yellow color. The progress of the
reaction was monitored by TLC, and was completed within 8 days. The mixture was extracted with
dichloromethane (2 x 50 mL). The aqueous layer was cooled, its pH adjusted to 6-7 by addition of 3.5N
HCl and re-extracted with CH
gave the title compound as yellowish solid which was collected by filtration, washed with cold water
2 x 10 mL), dried and recrystallized from a mixture of chloroform and ethanol (1:1, v/v); 1H- NMR
300MHz, DMSO-d6 ): δ 1.06 (t, J=6.9 Hz, 3H, CH CH ),2.57 (t, J = 6.5 Hz, 2H, CH -CO H), 3.70 (br
), 4.1(q,J=7.2Hz, 2H, CH CH ), 7.96 (d, JH-F = 11.7 Hz, 1H, H-5), 7.65 (br
s, 1H, NH), 8.58 (s,1H, H-2); 13C-NMR (300MHz, DMSO- d6): δ 16.03 (CH3), 29.43 (CH -CO H),
2.89
CH -NH), 50.51 (CH
2 2
Cl (50mL). Further acidification of the leftover aqueous layer to pH= 1-
2
(
(
2
3
2
2
3
t, J = 6.5 Hz, 2H, NH-CH
2
2
3
2
2
4
2
3
(
2
2
), 108.70 (C-3), 114.7 (d, JC-F = 22.9 Hz, C-5), 116.5 (d, J C-F =7.2 Hz, C-4a),
3
2
1
1
28.3 (d, J C-F = 5.5 Hz, C-8), 135.7 (C-8a), 138.8 (d, J C-F = 14.3 Hz, C-7), 148.33 (C-2), 150.4 (d, J
4
C-F = 248 Hz, C-6), 166.2 (C(3)-CO H), 169.4 (CH
2
2
CO
2
H),176.0 (d, J C-F = 2.6 Hz, C-4); HRMS (ESI,
+
ve): m/z [M++ Na] 390.07 C15H14FN O Na requires: 390.07144.
3 7
2-[(3-carboxy-1-ethyl-6-fluoro-8-nitro-4-oxo-1,4-dihydroquinolin-7-yl)
amino]succinic acid (compound 37, Table 5, Scheme 1)
A stirred mixture of DL- Aspartic acid (2.5 g, 24.0 mmol), synthon B (Figure 5) (2.0 g, 6.35 mmol) and
sodium hydrogen carbonate (4.5 g, 53.6 mmol) in 50 % aqueous ethanol (200 mL) was heated at 70-80
°
C for 6 days under reflux conditions. The mixture slowly developed bright yellow, then into clear
orange solution. The progress of the reaction was monitored by TLC, and was completed within 8 days.
The mixture was worked up as described for compound 36. The titled compound was collected as
1
bright yellow solid; H- NMR (300MHz, DMSO- d6 ): δ 1.27 (t, J= 7.7, 3H, CH
2
CH
3
), 2.95 (d, d, d, J
=
5.1, 14.7, 10.2 Hz , 2H, CH
2
-CO
2
H), 4.10 (m, 2H, CH CH ), 4.85 (d, d, J= 5.0, 13.5 Hz, 1H, CH-
2 3
3
NH), 7.47 (d, J = 8.7 Hz, 1H, NH), 8.14 (d, JH-F = 13.6 Hz, 1H, H-5), 8.90 (s, 1H, H-2), 12.51-13.32
13
(
2m, 2H, C-1''-COOH and C(3)-COOH), 14.72 (brs, 1H, C-2''-COOH); C-NMR (300MHz, DMSO-
d6 ):δ 15.58 (CH ), 37.12 (CH -CO H), 51.88 (CH CH ), 54.99 (d, J = 12.07 Hz, CH-NH), 109.8 (C-
3
2
2
2
3
2
2

2
3
3
3
1
1
4
), 115.12 (d, JC-F = 22.8 Hz, C-5), 118.47 (d, JC-F = 7.12 Hz, C-4a), 129.35 (d, JC-F = 5.17 Hz, C-8),
2
1
33.45 (C-8a), 137.99 (d, JC-F = 14.17 Hz, C-7), 150.67 (d, J C-F = 248.7 Hz, C-6), 151.08 (C-2),
4
65.54 (C(3)-CO
2
H), 172.55 (C-1''-CO
2
H), 172.62 (C-2''-CO
2
H), 175.45 (d, JC-F = 2.47 Hz, C-
);HRMS (ESI, +ve): m/z [M+] 412.07 C16
H
3 9
15FN O
requires: 412.07875.
7
-[(2-carboxyethyl)amino]-6-fluoro-1-(4-fluorophenyl)-8-nitro-4-oxo-
,4-dihydroquinoline-3-carboxylic acid (compound 38, Table 5, Scheme 1)
1
A stirred mixture of β-alanine (2.1 g, 23.5 mmol), synthon D (Figure 5) (2.30 g, 6 mmol) and sodium
hydrogen carbonate (3 g, 35.8 mmol) in 50 % aqueous ethanol (140 mL) was heated at 70-80 °C for 6
days under reflux conditions. The mixture slowly developed alight yellow color that changed into
bright yellow, then into clear orange solution. The progress of the reaction was monitored by TLC, and
was completed within 6 days. The mixture was extracted with dichloromethane (2 x 50 mL). The
aqueous layer was cooled, its pH adjusted to 6-7 by addition of 3.5N HCl and re-extracted with CH
50mL). Further acidification of the leftover aqueous layer to pH = 1-2 gave the title compound as
yellowish solid which was collected by filtration, washed with cold water (2 x 10 mL), dried and
recrystallized from a mixture of chloroform and ethanol (1:1,v/v). Yield 2.0 g (76.6%); R value in
system (3) = 0.2; mp= 231.6–233 °C; 1H- NMR(300MHz, DMSO, d6 ): δ 2.55 ( m, 2H, CH -COOH ),
.63 ( br m, 2H, CH -NH), 7.08(br t, 1H, NH-CH ), 7.36 (d, d, J= 8.5, 8.5 Hz, 2H, H-3'/ H-5'), 7.68 (d,
d, d, J= 4.8 Hz,3.7 Hz, 4.2 Hz, 2H, H-2'/ H-6'), 8.13 (d, JH-F = 13.76 Hz, 1H, H-5), 8.47 (s, 1H, H-2),
2 2
Cl
(
f
2
3
2
2
3
1
2.42 (br s, 1H, C(3)-CO
2
H), 14.67 (br s, 1H, CH
2
-CO
2
H); 13C-NMR (300MHz,DMSO, d6 ): δ 35.19
2
(
CH -CO H ), 41.73 (CH
2
2
2
-NH), 109.39(C-3), 114.15 (d, J C-F =23.25 Hz, C-5), 116.87 (C-8a), 116.89
2
3
(
d, J C-F = 23.24 Hz, C-3'/ C-5'), 128.42 (C-4a),128.86 (d, J C-F = 9.0 Hz, C-2'/ C-6'), 133.85 (C-8),
2
1
1
38.16( C-1'), 138.41 ( d, JC-F =14.6 Hz, C-7), 150.76 (d, J C-F = 248 Hz, C-6), 152.01 (C-2), 162.26
1
(
d, JC-F = 242 Hz,C-4' ), 165.33 (C(3)-CO
2
2 2
H), 173.09 (CH CO H), 175.5 (C-4); HRMS (ESI, -ve): m/z
[M-H]-432.06433C19
12
H F
2
3
N O
7
requires 432.06488.
2
3

7
-[(2-carboxypropyl)amino]- 1-(4-fluorophenyl)-6-fluoro-8-nitro-4-
oxo-1,4-dihydroquinoline-3-carboxylic acid (compound 39-A, Scheme 1)
A stirred mixture of isomethyl β-alanine (1.27 g, 12 mmol), synthon D (Figure 5) (1 g, 2.6 mmol)
and sodium hydrogen carbonate (1.5 g, 18 mmol) in 50 % aqueous ethanol (140 mL)
was heated at 70-80 °C under reflux conditions. The mixture slowly developed a light
yellow color that changed into bright yellow, then into clear orange solution. The
progress of the reaction was monitored by TLC, and was completed within 8 days. The
mixture was worked up as described for compound 35. The titled compound was
collected as bright yellow solid; 1H- NMR (300MHz, DMSO-d6 ): δ 1.02 (d, J = 6.76 Hz, CH
m, 1H, CH-COOH ), 3.25 (m, 2H, CH -NH), 7.18 (br t, J = 2.43 Hz, 1H, NH-CH ),
.36 (d, d, J=8.1, 8.3 Hz, 2H, H-3'/ H-5'), 7.67 (m, 2H, H-2'/ H-6'), 8.13 (d,
Hz, 1H, H-5), 8.49 (s, 1H, H-2), 12.31 (br s, 1H, (C(3)-CO H), 15.25 (br s, 1H,
CH CO H); 13C-NMR (300MHz, DMSO- d6): δ 14.94 (CH ), 40.81 (CH-CO H ),
8.24 (CH -NH), 109.39 (C-3), 114.18 (d, C-F = 23.85 Hz, C-5), 115.10 (C-8a),
16.89 (d, C-F = 23.32 Hz, C-3'/ C-5'), 119.94 (C-4a), 128.90 (d,
3
), 2.72
(
2
2
7
3
JH-F = 14.1
2
2
2
3
2
4
1
2
2
J
2
J
3
J
C-F = 9.45 Hz, C-2'/
2
1
C-6'), 133.89 (C-8), 138.50 (d, JC-F = 17.8 Hz, C-7), 145.05 ( C-1'), 151.25 (d, JC-F
=
2
55 Hz, C-6), 152.01 (C-2), 162.50 (d,
75.08 (CH-CO H), 175.12 (C-4).
1 2
JC-F = 225 Hz, C-4' ), 165.50 (C(3)-CO H),
1
2
8
-amino-7-[(2-carboxypropyl)amino]-6-fluoro-1-(4-fluorophenyl)-4-
oxo-1,4-dihydroquinoline-3-carboxylic acid (39-B, Scheme 2)
2
4

To a stirred solution of compound 39-A (0.5 g, 1.1 mmol) and potassium carbonate (0.96
g, 7 mmol) in water (20 mL) was added dropwise an aqueous solution of sodium dithionite (2.0 g, 11.5
mmol) in water (15 mL). The reaction mixture was further stirred at rt for 2 hrs. Thereafter, the pH of
the solution was adjusted to about 2. The precipitated product was filtered, washed with water, air-dried
and recrystallized from acetone and ethanol (1:1, v/v) to furnish faint yellow crystals. 1H- NMR
(
300MHz, DMSO- d6 ): δ 1.08(d, J = 6.77 Hz, 3H, CH
NH), 4.19 ( brs, 2H, NH ), 5.12 ( br s, 1H, NH-CH), 7.39-7.52 (m, 3H: d, 1H, JH-F = 8.15 Hz, H-5
and 2H, H-3'/ H-5'), 7.69 (m, 2H, H-2'/ H-6'), 8.45 (s,1H, H-2), 12.29 (br s, 1H, C(3)-CO H), 15.15 (br
H ), 48.65 H-NH
3 2
), 2.27 (m, 1H, CH-COOH), 3.10 (m, 2H, CH -
3
2
2
s, 1H, CH
2
-CO
2
H); 13C-NMR (300MHz, DMSO- d6 ): δ 15.36 (CH3),40.24 (CH-CO
2
2
3
)
, 100.69 (d, JC-F = 23.78 Hz, C-5 ), 106.37 (C-3),117.06 (d, JC-F = 23.1 Hz, C-3'/C-5'), 121.09 ( C-4a),
2
2
1
26.45 (C-8a), 128.74 (d, JC-F =8.85 Hz, C-2'/C-6'), 130.24 (d, JC-F = 16.5 Hz, C-7), 131.21 (C-8),
1
1
1
39.55 (C-1'), 150.84(d, JC-F = 255 Hz, C-6), 151.36 (C-2), 162.16 (d, JC-F = 234 Hz, C-4'), 166.31
H), 176.85 (CH CO H), 177.69 (C-4),HRMS (ESI, +ve): m/z [M+] 418.11 C20
requires: 418.11885.
(C(3)-CO
2
2
2
18 2 3 5
H F N O
7
-(2-Carboxy-ethylamino)-1-(4-chloro-phenyl)-6-fluoro-8-nitro-4-oxo-1,4-
dihydro-quinoline-3-carboxylic acid (40-A, Scheme 1)
A stirred mixture of β-alanine (2.1 g, 23.5 mmol), Synthon C (2.0 g, 4.4mmol) and
sodium hydrogen carbonate (3 g, 35.8 mmol) in 50 % aqueous ethanol (140 mL) was
heated at 70-80 °C for 6 days under reflux conditions. The mixture slowly developed a
light yellow color that changed into deep yellow. The progress of the reaction was monitored by TLC,
and was completed within 6 days. The mixture was worked up as described for compound 35.
1
7
5
1
1
1
H- NMR (300 MHz, DMSO- d6 ): δ 2.28 ( d, J =8.1, 2H, CH
.51(br t, J =5.7Hz, 1H, NH-CH ), 7.51-7.54 (m, 2H, H-3', H-5'), 7.61-7.67 (m, 2H, H-2', H-6'), 8.2 (H-
), 8.62 (s, 1H, H-2); 13C-NMR (300MHz,DMSO- d6 ): δ 35.17 (CH -CO H ), 41.84 (CH -NH),
2
-COOH ), 3.66 (m, 2H, CH
2
-NH),
2
2
2
2
09.60 (C-3), 114.18 (C-5), 116.58 (C-4a), 127.7 (C-3'), 128.03 (C-5'), 128.48 (C-8), 129.10 (C-2'),
1
29.55 (C-6'), 133.79 (C-8), 134.50 (C-4'), 138.72 (C-7), 151.33 (d, JC-F =303.7 Hz, C-6), 145.08 (C-
2 2 2
'), 150.37 (C-2), 165.27 (C(3)-CO H), 173.11 (CH CO H), 177.36 (C-4); HRMS (ESI, -ve): m/z [M-
3 7
H]-448.04 C19H12ClFN O
requires 448.0353
2
5

8
-Amino-7-(2-carboxy-ethylamino)-1-(4-chloro-phenyl)-6-fluoro-4-oxo-1,4-
dihydro-quinoline-3-carboxylic acid (40-B , Table 5, Scheme 2)
To a stirred solution of compound 40-A (0.5 g, 1.19mmol) and potassium carbonate (0.96g, 7 mmol)
in water (20 mL) was added dropwise an aqueous solution of sodium dithionite (2.0 g, 11.5 mmol) in
water (15 mL). The reaction mixture was further stirred at rt for 2 hrs. Thereafter, the pH of the
solution was adjusted to about 2. The precipitated product was filtered, washed with water, air-dried
and recrystallized from acetone and ethanol (1:1, v/v) to furnish faint yellow crystals.; 1H- NMR
(
300MHz, DMSO- d6 ): δ 2.37 (m, 2H, CH
.01 ( d, 1H, J = 7.29 Hz, CH -NH), 7.42 -7.78 (m, 4H, H-3'/ H-5', H-2'/ H-6'), 8.45 (d,1H, JH-F =8.08,
H-5), 8.61 (s, 1H, H-2), 13.6 (br s, 1H, C(3)-CO H), 15.2 (br s, 1H, CH -CO H); 13C-NMR (300MHz,
DMSO- d6 ):δ 35.08 (CH-CO H),41.97 (CH -NH ), 100.12 (C-5), 106.23 (C-3),130.18 (d, C-2'/C-6'),
19.185 (C-4a), 123.50 (C-8a), 128.76 (d, C-3'/C-5'), 126.69 (C-7), 132.62(C-8), 134.699(C-
2 2 2
-COOH), 3.68 (m, 2H, CH -NH), 5.05 ( brs, 2H, NH ),
3
6
2
2
2
2
2
2
1
4
1
'),141.60(C-1'),153.15(d, JC-F = 285.6 Hz, C-6), 150.85 (C-2),166.25 (C(3)-CO
2
H),173.97
(
CH
2
CO H),177.78(C-4);HRMS (ESI, +ve): m/z [M+] 420.07 C19 16ClFN requires: 420.07462.
2
H
3 5
O
4.5 In Vitro Assay
4.5.1 Materials
Pharmaceutical grade reagents and chemical materials were purchased from commercial
vendors. Pim-1 kinase assay kit was purchased from Cyclex® Japan; solvents for bioanalysis
were all purchased from Sigma-Aldrich (USA).
4.5.2 Preparation of synthesized Compounds for In vitro Assay
Powdered synthesized compounds to be tested (3.5 mg to 10 mg) were primarily dissolved in
DMSO to give stock solutions of fixed concentrations. Then, they were diluted to the required
concentrations with deionized water for in vitro assay evaluation.
4.5.3 Spectrophotometric Assay of Pim-1 Kinase Inhibitors
2
6

Pim-1 kinase drug discovery kit Cyclex® (Japan) is composed of ELISA pre-coated 96
well plates with "recombinant p21waf1" antibody which can be selectively phosphorylated
by Pim-1 kinase enzyme.
A
detector anti-phospho-p2waf1T145 polyclonal antibody is
conjugated to horse radish peroxidase antibody PWT-01, and can explicitly detect the
phosphorylated form of "recombinant p21waf1".
Upon performing the assay hits to be tested are first diluted with the kinase buffer.
Afterwards, the hits' solutions are pipetted into pre-coated 96 well plates to yield final
concentrations of 0.1, 1.0, and 10 µM. Then, Pim-1 kinase was added to each well as
aqueous solution 0.01 units in 10 µL, and allowed to phosphorylate the bound substrate in
the presence of ATP natural kinase substrate. The amount of phosphorylated substrate was
quantified by using the detector antibody conjugated to horse radish peroxidase which then
catalyzes the change of chromogenic substrate tetra-methylbenzidine (TMB) from a colorless
solution to a blue solution. The color of the solution is measured using Bio-Tek instruments
(ELx 800 Winooski VT) spectrophotometric plate reader at a wavelength of 650 nm and the
inhibition of Pim-1 kinase is calculated as percent activity of the uninhibited Pim-1 kinase
enzyme control.
2
7

Table 1: Selected Pim-1 kinase crystals with their co-crystallized ligands and matching features
No. of
No. of
pharm .
generated
a
features match the receptor-ligand
interactions:
PDB code
Ligand
Res.
features
in ligand
H
O
N
O
H
N
b
2
BIK
1
.8 Å
16
HBD, 3xHbic, RingArom, PosIon
10
c
BI11306
N
N
b
3
BGZ
O
N
2
.4 Å
13
3xHbic, 2xRingArom, NegIon
7
VX3314
O
H
O
b
HN
N
3
DCV
2
.7 Å
16
HBA, 2xHBD, 2xHbic, RingArom
10
5
5E500
HO
O
HN
NH
O
b
1
XWS
1
.8 Å
16
HBA, HBD, 3xHbic, PosIon
10
BI11001
N
N
a
Corresponds to the resolution of the Pim-1 kinase specified crystal in the PDB.
The PDB code for the crystal structure of Pim-1 kinase and below is the name of its co-crystallized ligand.
b
c
The name of the co-crystalized ligand in the PDB
2
8

Table 1: Selected Pim-1 kinase crystals with their co-crystallized ligands and matching features
No. of
No. of
pharm .
generated
a
features match the receptor-ligand
PDB code
Ligand
Res.
features
in ligand
interactions:
H
N
O
b
1
YHS
2.15 Å
22
HBD, 4xHbic, PosIon
10
N
N
Staurosporine
CH3
O
O
HN
CH3
CH3
O
N
Cl
NH
b
3
7
CXW
2
.1 Å
14
HBD, 4xHbic, PosIon
10
CP501
Cl
N
N
N
O
b
3F2A
OH
1
.9 Å
13
Hbic, 3xRingArom, PosIon, NegIon
7
N
MG314
N
N
N
b
N
3CY3
2.15 Å
15
HBA, HBD, 3xHbic, PosIon
7
JN5501
N
N
N
S
N
a
Corresponds to the resolution of the Pim-1 kinase specified crystal in the PDB.
The PDB code for the crystal structure of Pim-1 kinase and below is the name of its co-crystallized ligand.
b
c
The name of the co-crystalized ligand in the PDB
.
29