3D-QSAR-aided design, synthesis, in vitro and in vivo evaluation of dipeptidyl boronic acid proteasome inhibitors and mechanism studies

Article abstract of DOI:10.1016/j.bmc.2016.04.025

Proteasome had been clinically validated as an effective target for the treatment of cancers. Up to now, many structurally diverse proteasome inhibitors were discovered. And two of them were launched to treat multiple myeloma (MM) and mantle cell lymphoma (MCL). Based on our previous biological results of dipeptidyl boronic acid proteasome inhibitors, robust 3D-QSAR models were developed and structure-activity relationship (SAR) was summarized. Several structurally novel compounds were designed based on the theoretical models and finally synthesized. Biological results showed that compound 12e was as active as the standard bortezomib in enzymatic and cellular activities. In vivo pharmacokinetic profiles suggested compound 12e showed a long half-life, which indicated that it could be administered intravenously. Cell cycle analysis indicated that compound 12e inhibited cell cycle progression at the G2M stage.

Full text of DOI:10.1016/j.bmc.2016.04.025

Accepted Manuscript
3D-QSAR-aided design, synthesis, in vitro and in vivo evaluation of dipeptidyl
boronic acid proteasome inhibitors and mechanism studies
Meng Lei, Huayun Feng, Cheng Wang, Hailing Li, Jingmiao Shi, Jia Wang,
Zhaogang Liu, Shanshan Chen, Shihe Hu, Yongqiang Zhu
PII:
S0968-0896(16)30259-0
DOI:
http://dx.doi.org/10.1016/j.bmc.2016.04.025
BMC 12940
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
15 March 2016
9 April 2016
12 April 2016
Please cite this article as: Lei, M., Feng, H., Wang, C., Li, H., Shi, J., Wang, J., Liu, Z., Chen, S., Hu, S., Zhu, Y.,
3D-QSAR-aided design, synthesis, in vitro and in vivo evaluation of dipeptidyl boronic acid proteasome inhibitors
and mechanism studies, Bioorganic & Medicinal Chemistry (2016), doi: http://dx.doi.org/10.1016/j.bmc.
2016.04.025
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.

3D-QSAR-aided design, synthesis, in vitro and in vivo evaluation of
dipeptidyl boronic acid proteasome inhibitors and mechanism studies
Meng Lei a, Huayun Feng a, Cheng Wang a, Hailing Li b, Jingmiao Shi c, Jia Wang c,
Zhaogang Liu c, Shanshan Chen c, Shihe Hu c, Yongqiang Zhu b,*
a
College of Science, Nanjing Forestry University, No. 159 Longpan Road, Nanjing
210037, PR China
b
College of Life Science, Nanjing Normal University, No. 1 Wenyuan Road, Nanjing
210037, PR China
c
Jiangsu Chia Tai Fenghai Pharmaceutical Co. Ltd., No. 9 Weidi Road, Nanjing
210046, PR China
Abstract
Proteasome had been clinically validated as an effective target for the treatment of
cancers. Up to now, many structurally diverse proteasome inhibitors were discovered.
And two of them were launched to treat multiple myeloma (MM) and mantle cell
lymphoma (MCL). Based on our previous biological results of dipeptidyl boronic acid
proteasome inhibitors, robust 3D-QSAR models were developed and
structure-activity relationship (SAR) was summarized. Several structurally novel
compounds were designed based on the theoretical models and finally synthesized.
Biological results showed that compound 12e was as active as the standard
bortezomib in enzymatic and cellular activities. In vivo pharmacokinetic profiles
suggested compound 12e showed a long half-life, which indicated that it could be
administered intravenously. Cell cycle analysis indicated that compound 12e inhibited
cell cycle progression at the G2M stage.
Keywords: 3D-QSAR; Proteasome inhibitor; Pharmacokinetic; Cell cycle
* Corresponding author. Tel.:+86 25 85891591.
1

E-mail address: zhyqscu@hotmail.com (Y.Q. Zhu)
1. Introduction
The ubiquitin-proteasome pathway (UPP) plays a pivotal role in cellular protein
turnover, protein quality control, antigen processing, signal transduction, cell cycle
regulation and cell differentiation.1-3 It can also regulate the degradation of
intracellular proteins with extreme specificity as to target, time and space. Due to
these important discoveries, three scientists won the 2004 Nobel Prize for chemistry.
Recent studies proved that this pathway played a central role in the development of a
number of human major diseases such as inflammation, aging, cancer,
neurodegenerative diseases, diabetes, etc.4-8 The clinical introduction of dipeptidyl
boronic acid bortezomib (Velcade®, PS-341, Fig. 1) and tetrapeptidyl epoxyketone
carfilzomib (Kyprolis®, PR-171, Fig. 1) for the treatment of multiple myeloma and
mantle cell lymphoma9-14 provided the “proof of concept” that targeting the UPP was
a viable route for the treatment of human cancers.15,16
Figure 1. Structures of approved and currently developed proteasome inhibitors.
2

Besides the bortezomib and carfilzomib, many structurally diverse proteasome
inhibitors had been discovered and developed (Fig. 1) based on this interesting target,
including synthetic and natural chemical compounds.17 Peptide aldehydes (such as
MG132, Fig. 1) were the first proteasome inhibitors to be developed. However, due to
their fast dissociation and rapid oxidation into inactive acid by cells, none of these
kinds of inhibitors was developed into drugs.18 Both NPI-0052 (marizomib, developed
by Nereus Pharmaceuticals, Inc.) and lactacystin were derivatives of natural products.
It is noteworthy to point out that NPI-0052 is currently in clinical trials for the
treatment of patients with hematologic and solid tumor malignancies.19-21 The high
selectivity and low dissociation rates of boronic acids have placed these chemical
compounds into the focus of medical research and drug development. This led to the
development of the above mentioned bortezomib and structurally related peptidyl
boronic acid CEP-18770, which is also under clinical evaluation.22,23 Synthetic
peptidyl inhibitors are complemented by microbial natural product peptides of the
α’β’-epoxyketone family, including epoxomicin,24,25 which served as a model for the
synthetic inhibitor carfilzomib, and it was recently approved by FDA for the treatment
of multiple myeloma.26
Previously, a series of structurally novel dipeptidyl boronic acid proteasome
inhibitors were synthesized and biologically evaluated by our group.27 It turned out
that several compounds were as active as the marketed bortezomib. To expand the
structure diversity and quantitatively summarizing the structure-activity relationship
(SAR) of the synthesized inhibitors, in this manuscript we firstly carried out
3D-QSAR studies with the methods of CoMFA (comparative molecule field analysis)
and CoMSIA (comparative molecule similarity indices analysis) based on our own
biological results. Then the SAR was summarized and several structurally diverse
proteasome inhibitors were designed and synthesized. The enzymatic and cellular
activities of these inhibitors were evaluated and in vivo pharmacokinetic parameters
were obtained. Finally the mechanism of action was investigated for these
compounds.
3

2. Results and discussion
2.1. Alignment of molecules
The structures of the proteasome inhibitors investigated in the calculations were
shown in Table 1.
Table 1
Structures of compounds 1-39, bortezomib and their inhibition of human 20S proteasome
IC50
a
Compd.
P1
P2
P3
P4
pIC50
-0.28
(nM)
1.92
1
4.85
3.21
-0.68
-0.51
2
3
4.18
1.12
5.17
-0.62
-0.049
-0.71
4
5
6
4

1.92
5.24
6.70
-0.28
-0.72
-0.83
7
8
9
1.26
-0.10
10
39.0
14.2
17.2
4.26
7.63
6.12
-1.59
-1.15
-1.24
-0.63
-0.88
-0.79
11
12
13
14
15
16
CH3, CH3
CH3, CH2CH3
5

3.67
-0.56
17
1.46
3.37
-0.16
-0.53
18
19
1.16
1.15
6.96
0.76
1.34
2.46
1.16
-0.064
-0.061
-0.84
0.12
20
21
22
23
24
25
26
CH3, CH3
-0.13
-0.39
-0.064
8.46
2.34
-0.93
-0.37
27
28
CH3
CH3
6

2.73
-0.44
29
CH3
16.4
114
-1.21
-2.06
-1.69
-2.08
30
31
32
33
49.5
120.2
471
41.2
3.34
-2.67
-1.61
-0.52
34
35
36
4.37
1.60
-0.64
-0.21
37
38
H, H
H, H
1.20
1.9
-0.08
-0.29
39
H, H
H, H
bortezomib
7

a pIC50 = -log IC50.
All molecular modeling was performed using SYBYL 7.3 software package (Tripos
Inc., St. Louis, MO) running on Red Hat Enterprise linux workstation. For the
absence of empirical potential energy functions for the boron atom in default settings
in SYBYL 7.3, we adopted the same strategy reported in the literatures28,29 and
changed the boron atom ‘B’ in bortezomib to carbon atom ‘C.3’ and colored it with
‘Magenta’. The initial conformation of the template molecule bortezomib was
extracted from the crystal structure of bortezomib-20S proteasome complex (PDB
2F16).30 And after structure correction and minimization, partial atomic charges were
calculated using the Gasteiger-Hückel method.31,32 The rest molecules were built
based on the optimized bortezomib after structural modification and minimization.
The 3D-QSAR analyses were carried out using the CoMFA and CoMSIA methods.33,
34
Based on the rule of structural diversity and wide range of biological data,
compounds 3, 13, 29, 31, 35, and 37 were selected as test set and the rest 34
molecules as the training set. The structural diversity of the aligned molecules was
shown in Fig. 2.
Figure 2. Alignment of the inhibitors in the training set. Hydrogens are not shown.
2.2. Statistic results of CoMFA and CoMSIA
The statistic results obtained from standard CoMFA and CoMSIA models are
8

summarized in Table 2. The CoMFA models were constructed from steric and
electrostatic descriptor fields and the CoMSIA models were built by varying the steric,
electrostatic, hydrophobic, hydrogen-bond donor and acceptor descriptor fields. It is
generally considered that the built models were reliable for prediction if the q2 and r2
were greater than 0.5 and 0.7. Table 2 showed both CoMFA and CoMSIA models
gave the cross-validated correlation coefficients q2 of 0.707 and 0.704, respectively,
and the correlation coefficients r2 of 0.917 and 0.893, respectively, suggesting the
models generated by the two methods in this paper could be used for the prediction
and design of novel proteasome inhibitors. At the same time, the statistic data in
CoMFA models indicated that the steric and electrostatic fields accounted for 0.647
and 0.353, respectively, which reflected that the former was quite crucial in
explaining the variations in inhibitory potency of these molecules. This theoretically
result was consistent with the early discussed SAR of the synthesized compounds in
this manuscript. For example, placement of a less steric valine at P2 position
(compound 11) resulted in an IC50 value 39.0 nM (Table 1), while placement of a
much bulkier leucine at the same position (compound 12, 14.2 nM) led to more than
2-fold increase in potency. The CoMSIA models extensively calculated five field
contributions, including steric (0.212), electrostatic (0.268), hydrophobic (0.331),
hydrogen-bond donor (0.096) and acceptor (0.093). And the hydrophobic field made
the greatest contribution to the biological activity, which coincided with our
previously reported results35 and the characteristics of the active domain.36
Table 2
Statistical data for QSAR method with CoMFA and CoMSIA
Field contributions (%)
Methods
n
q2
SPRESS
r2
F
s
S
E
H
D
A
CoMFA
4
4
0.707
0.704
0.441
0.433
0.917 90.7 0.235 0.647 0.353
69.2 0.260
0.212 0.268 0.331 0.096 0.093
CoMSIA
0.893
n, optimal number of components; q2, squared cross-validated coefficient; SPRESS, root mean
predictive error sum of squares; r2, squared conventional coefficient; F, F-test value; s, standard
error of estimate; S, E, H, D, A denote the steric, electrostatic, hydrophobic, hydrogen-bond donor
9

and acceptor, respectively.
2.3. Predicted power of CoMFA and CoMSIA models
Table 3 summarized the actual and predicted biological data of the inhibitors in
the test set obtained by the CoMFA and CoMSIA methods. Six inhibitors, 3, 13, 29,
31, 35, and 37 that were not included in the training set were used as a test set to
verify the predictive power of the built CoMFA and CoMSIA models. By comparison
of the experimentally observed and theoretically predicted pIC50 and their residual
values of dipeptidyl boronic acid proteasome inhibitors, both calculated models
performed well in the prediction of the activities of the test inhibitors. In almost all the
cases, the predicted values were close to the observed pIC50 values, deviating by less
than 0.5 logarithm units in binding affinities to the 20S proteasome. All results
including statistical data from this predictive validation analysis further confirmed the
robustness of the obtained models in terms of their predictive abilities. There was no
significant difference between the CoMFA and CoMSIA models for this predictive
validation test.
Table 3
CoMFA and CoMSIA actual and predicted activities for test set molecules
pIC50 (calcd)b
Residuals
CoMSIA
pIC50 (obsd)a
Compd.
CoMFA
CoMSIA
-0.59
CoMFA
-0.13
-0.51
-1.24
-0.44
-2.06
-1.61
0.72
-0.38
-1.21
-0.75
-1.89
-1.40
0.61
0.08
0.08
0.32
-0.18
-0.41
0.05
3
-1.32
-0.76
-1.88
-1.20
0.67
-0.03
0.31
13
29
31
35
37
-0.17
-0.21
0.11
a obsd, observed values; b calcd, calculated values.
Fig.s 3 and 4 showed the CoMFA and CoMSIA contour maps, respectively and for
convenient visualization, the drug bortezomib was selected as the reference molecule
in all maps.
10

2.4. CoMFA contour maps analysis
CoMFA method calculates the steric and electrostatic fields. Fig. 3 showed the
combination of the CoMFA steric and electrostatic fields. Green and yellow
polyhedra indicated regions where bulky groups would improve or decrease the
activity, respectively. Red and blue polyhedra indicated regions where negatively
charged groups were favorable or unfavorable. At P1 position, the area around the
aromatic pyrazine ring of bortezomib was identified with yellow and red polyhedra,
which indicated that large volume and positive moieties at this position were not well
accepted. This was in concordance with the inactive compounds 30-34 (IC50 range
16.4-471 nM) in Table 1, molecules that were all substituted by positive and bulky
Boc groups. On the contrary, while the negative and moderate volume
5,6,7,8-tetrahedronaphthyl and 2-naphthyl groups were replaced at this position, such
as compounds 21-26 (IC50 range 0.76-6.96 nM), the inhibition capacities against the
20S proteasome were greatly improved. At P2 position, the green and red polyhedron
in the contour map near the aromatic phenyl ring of bortezomib reflected that bulky
and negatively charged groups were well abided to increase the potency and this
completely coincided with our published 3D-QSAR results calculated with Millenium
Co. Ltd. data set. 35 This was also confirmed by the comparison between compounds
11 (39.0 nM) and 15 (7.63 nM), which were substituted with small, positive methyl
and steric, negative 2-naphthyl groups, respectively, however, compound 15 was
5-fold potent more than 11. At P3 position, the CoMFA model showing a green
isopleth near the isopropyl group of the reference molecule bortezomib suggested that
bulky groups could be added to increase the potency. This was confirmed with the
remarkable difference of the activity between the presence of an isopropyl moiety in
compound 28 (2.34 nM) compared with an n-propyl substituent in compound 27 (8.46
nM). Furthermore, at the same place, the favorable negative red isopleths beside the
isopropyl group indicated that a negative substituent on the aromatic ring at this
position was preferred instead of a positive one or even without substituents on the
aromatic ring. Both compounds 3 (3.21 nM) and 10 (1.26 nM) harboring negative
fluorine atoms on the para-phenyl ring at P3 position were more potent than those
11

substituted with methyl groups or without any substituents, such as 2 (4.85 nM), 8
(5.24 nM), and 9 (6.70 nM). At P4 position, the favorable positive blue isopleths
surrounding the two hydroxyls of boronic acid of bortezomib revealed that positively
charged groups linked with these two hydroxyls, such as pinane, were beneficial to
high activity. Compound 3 having positive GH charge at P4 position showed a higher
activity (3.21 nM), whereas the ‘naked’ boronic acid at the same position of
compound 37 slightly decreased the potency (4.37 nM).
Figure 3. Std*coeff contour map of final CoMFA analysis with 2Å grid spacing in combination
with bortezomib. Green contours (80% contribution) refer to sterically favored regions; Yellow
contours (20% contribution) refer to sterically disfavored regions; Blue contours (70%
contribution) refer to regions where positively charged substituents are favored; red contours (30%
contribution) indicate regions where negatively charged substituents are favored.
2.5. CoMSIA contour maps analysis
CoMSIA was an extension of the CoMFA method and additionally calculated
hydrophobic, hydrogen-bond donor and acceptor descriptor fields in addition to steric
and electrostatic fields. Fig. 4 provided graphical representations of the CoMSIA
models. Since the steric contour map of CoMSIA well agreed with that of CoMFA, so
it was not shown and discussed here again. Fig. 4a showed the electrostatic contour
maps of CoMSIA, which was different from those of CoMFA in some extent. At P1
and P2 positions, favorable positive blue contours around the pyrazine ring and
methylene of bortezomib respectively showed that the positive hydrogen atoms on
these positions were accepted. At P3 position, isopropyl group fell into a large blue
12

isopleths suggesting that positively charged substituents instead of negatively charged
ones were required for the high activity. This was confirmed by the comparison of the
different activity of compounds 1 (1.92 nM), 2 (4.85 nM), and 3 (3.21 nM). Among
these three compounds, the highest potent 1 was substituted by a positive isopropyl at
P3 position while moderate 2 and 3 were occupied by negatively substituted phenyl
rings. Blue contours below the two hydroxyls at P4 position indicated that positive
substituents such as pinane were beneficial for high activity, which had been
discussed in the CoMFA electrostatic field. The hydrophobic contour map is displayed
in Fig. 4b. White and yellow contours highlighted areas where hydrophilic and
hydrophobic properties were preferred, respectively. Table 2 showed that hydrophobic
field made the largest contribution to the CoMSIA QSAR models, which suggested
that among the descriptors considered, the hydrophobicity of inhibitors was the most
important factor affecting the potency. This was consistent with the hydrophobic
properties of the amino acid residues in the β5 subunit of 20S proteasome.30 Within 5
Å around bortezomib, ALA20, MET45, ALA46, ALA49 and ALA50 were all
hydrophobic amino acid residues, which could effectively interact with inhibitors
through hydrophobic interaction.36 Substituents at P1, P2 and P3 positions fell into
three yellow contours, respectively, indicating that hydrophobic groups were well
tolerated at these positions. On the contrary, hydrophilic groups were disfavored for
the activity. Compounds 30 (5.24 nM) and 31 (5.24 nM) bearing hydrophilic phenol
hydroxyl groups on the para-phenyl ring significantly reduced the potency. This
conclusion paralleled with our previous SAR discussion and published results.27
White polyhedra encompassing the two hydroxyl groups reflected the hydrophilic
oxygen atoms were well permitted. Fig.s 4c and 4d present the hydrogen-bond donor
and acceptor contour maps, respectively, which well agreed with our previously
reported results.35 The presence of two cyan contours in Fig. 4c suggested that
hydrogen bond donor groups on the inhibitors such as NH groups of the carbonyls
were the favored feature for the high activity. Once the hydrogen atom of the NH
moiety was substituted, the potency was decreased in a large extent due to the
breakage of the hydrogen bonds between the inhibitors and the residues of 20S
13

proteasome.36 Also the two red isopleths surrounding the two oxygen atoms of the
carbonyls in Fig. 4d indicated the importance of such hydrogen bond acceptors in the
inhibitors. In the crystal structure of bortezomib-20S proteasome, formation of two
3.2 Å and 3.1 Å bond-length H-bonds with NH groups of Ala49 and Thr21 residues of
20S proteasome was observed.30 It was noteworthy to point out that a red contour
appeared below the aromatic pyrazine ring of bortezomib, revealing that the nitrogen
atom on the ring was a hydrogen bond acceptor.
Figure 4. Std*coeff contour maps of final CoMSIA analysis with 2Å grid spacing in combination
with bortezomib. (a) Electrostatic contour maps. Blue contours (80% contribution) refer to regions
where negatively charged substituents are disfavored; red contours (20% contribution) indicate
regions where negatively charged substituents are favored. (b) Hydrophobic contour map. White
contours (20% contribution) refer to regions where hydrophilic substituents are favored; yellow
contours (80% contribution) indicate regions where hydrophobic substituents are favored. (c)
Hydrogen-bond donor contour map. Cyan contours indicate regions where hydrogen bond
14

acceptor groups on the receptor increase activity. (d) Hydrogen-bond acceptor contour map. Red
contours refer to areas where hydrogen bond donors on the receptor promote the affinity.
The newly built 3D-QSAR models with our own experimental data were different
from 3D-QSAR models previously constructed by us in some extents, but they were
more reliable and powerful for SAR discussion and the design of structurally novel
compounds.
2.6. SAR summarization
Based on 3D-QSAR analysis, some structural requirements were summarized in
Fig. 5, which can be referred to rational design of novel and potent molecules with
favorable biological activities.
Figure 5. SAR summarization of dipeptidyl boronic acid proteasome inhibitors.
Bortezomib covalently binds to hydroxyl group of Thr1 residue of β5 subunit,
which is the most important binding site in its interaction to keep the biological
activities. Moreover, some key hydrogen interactions are formed between bortezomib
and Thr1, Thr21, Gly47 and Ala49. An aromatic group was beneficial to the binding
at phenyl ring of bortezomib. According to the detailed contour analyses of the
CoMSIA model, we observed magenta and yellow contours around the pyrazine ring
of bortezomib. Thus, we summarized three important regions for the binding mode
15

between residues of β5 subunit and bortezomib (1) the covalent site where a covalent
bond is formed between bortezomib and Thr1 residue; (2) the Aromatic ring favored
region of the phenyl moiety of bortezomib; (3) and an acceptor and hydrophobic
favored region of pyrazinyl group. Based on the theoretically models, five compounds,
(Table 4), were designed and synthesized.
12a-12e
2.7. Chemistry
Scheme 1 summarized the synthesis of the intermediates 7a-7c, which were
prepared according to literatures.37,38 Firstly phthaloyl group 1 reacted with ethyl
chloroformate to form amide 2, which was amonified by L-alanine at basic conditions
to give intermediate 3 (yield 79.3%). The carboxyl acid of alanine derivative 3 then
coupled with directing group 8-aminoquinoline to obtain amide 4 with the yield of
71.9%. Different aromatic groups were introduced in the terminal carbon of alanine
via palladium-catalyzed reaction to afford protected amino acids 5a-5c (yields
72.1-75.5%). The directing groups 8-aminoquinoline of 5a-5c were removed by the
treatment of BF3·Et2O in MeOH to afford the corresponding amino acid methyl esters
6a-6c with the amines protected by phthaloyl group (yields 75.2-80.1%). Finally the
protecting phthaloyl groups were removed by ethylenediamine to afford unnatural
amino acid methyl esters 7a-7c (yields 48.6-63.7%).
The synthesis of the target compounds 12a-12e was illustrated in Scheme 2.
Non-natural amino acid methyl esters 7a-7c were coupled with various acids
P1COOH in the presence of 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide
hydrochloride (EDCI) and 1-hydroxybenzotriazole (HOBt) to afford the
corresponding peptide esters 8a-8e, which were saponified and acidified to give
amino acids 9a-9e in 72.3-85.2% yields. Acid groups of 9a-9e were reacted with the
commercialy available amino boronate trifluoroacetate 10 using the common
peptide synthesis method to afford the dipeptidyl boronic acid esters 11a-11e in
68.6-87.6% yields. Finally after transesterification of 11a-11e with isobutylboronic
acid, target compounds 12a-12e were obtained in moderate to high yields
(48.6-63.7%). The structures of the target compounds were displayed in Table 4.
16

Scheme 1. General synthesis of unnatural amino acid methyl esters 7a-7c. Reagents and
conditions: (i) ClCOOEt, TEA, DMF, rt; (ii) (1) L-alanine, Na2CO3, H2O, rt; (iii) (1) SOCl2,
CH2Cl2, 42℃; (2) 8-Aminoquinoline, DIPEA, CH2Cl2, -20℃ to rt; (iv) Ar-I, Pd(OAc)2, AgBF4,
t-BuOH, 85℃; (v) BF3·Et2O, MeOH, 100℃; (vi) Ethylenediamine, MeOH, 70℃.
Scheme 2. General synthesis of dipeptidyl boronic acids 12a-12e. Reagents and conditions: (i)
P1COOH, EDCI, HOBt, DIPEA, CH2Cl2, -10℃ to rt; (ii) (1) LiOH·H2O, MeOH, H2O, rt; (2) 1N
HCl, ethyl acetate, rt; (iii) EDCI, HOBt, DIPEA, CH2Cl2, -10℃ to rt; (iv) 1 N HCl, MeOH,
hexane, rt.
2.8. Biology
2.8.1. 20S Human proteasome inhibitory activities
The capacities of target compounds to inhibit the CT-L activity of 20S human
proteasome were assayed using appropriate fluorogenic substrates. The marketed
bortezomib was used as standard. The IC50 values of dipeptide boronic acids 12a-12e
and bortezomib were showed in Table 4. The biological results showed that addition
17

of methyl group at pyrazinyl ring of bortezomib was not beneficial to the activity (12a,
26.24 nM). To prevent the metabolism of phenyl group at P2 position of bortezomib, a
trifluoromethyl substituent (12b, 42.61 nM) was employed at para-position of phenyl
group, however, the activity was reduced. According to the 3D-QSAR model,
hydrophobic substituents at P1 position would be beneficial to the activity. So
2,5-dichlorophenyl group (12c, 8.32 nM) was used to replace the pyrazinyl of
compound 12b, and the activity was indeed increased, which was consistent with the
theoretical models. However, introduction of additional trifluoromethyl group at P2
positionto compound 12c led to more than 3 fold reduction of proteasome inhibitory
activity (12d, 25.02 nM). But the activity was increased greatly when two
trifluoromethyl groups on phenyl ring were substituted by two methoxy ones (12e,
4.60 nM). So based on the 3D-QSAR theoretical models, two dipeptidyl boronic acid
proteasome inhibitors 12c and 12e with IC50 less than 10 nM were screened and
selected to investigate the cellular activities.
Table 4
Structures of compounds 12a–12e and proteasome inhibitory activities
Compd.
P1
P2
IC50(nM)
7.05
Bortezomib
26.24
42.61
12a
12b
8.32
12c
18

25.02
4.60
12d
12e
2.8.2. Cellular activities of compounds 12c and 12e
The marketed drug bortezomib was approved for the treatment of multiple
myeloma. So in this manuscript, we just evaluated the cytotoxic potentials of
compounds
and
against three multi myeloma cell lines, RPMI8226, ARH-77
12e
12c
and U266B1. The results were displayed in Table 5. It showed that compound
12e
against
was nearly as active as the standard bortezomib and more active than
12c
three cancer cell lines. So based on the enzymatic and cellular activities, compound
could be considered as lead compound for in vivo investigation.
12e
Table 5
The cellular activities of compounds 12c and 12e against MM Cell Lines
Compd.
12c
RPMI8226
233.7
ARH-77
98.15
14.58
9.57
U266B1
100.4
37.78
18.66
12e
bortezomib
11.2
11.63
2.8.3. In vivo pharmacokinetic studies
Compound 12e was further investigated by profiling iv and ig pharmacokinetics in
male SD rats. The pharmacokinetics parameters were listed in Table 6. It illustrated
that the elimination phase half-lifes of lead compound 12e were 7.24 h for iv and 0.2
h for ig, respectively, which indicated that this lead compound could be administered
intravenously. However the exposure needed to be improved.
19

Table 6
Single dose iv and ig pharmacokinetic profiles of 12e in SD rats
Dose
Cmax
AUClast
AUC0-inf MRTlast
Compd.
iv
T1/2(h) Tmax(h)
Vz_obs Cl_obs
(mg/kg)
(ng/ml) (h*ng/ml) (h*ng/ml)
(h)
1
7.24
0.2
9.36
0.03
7.83
245
47.9
407
83.0
408
8.44
-
-
12e
ig
0.2
6.9
351
1258
a ig , intragastric administration; b iv, intravenous injection.
2.8.4. Research of cell cycle
We carried out mechanism study to conclude how compound 12e stopped the
progression of cancer cells. And the cell cycle distributions of U266 cells treated with
compound 12e and DMSO were shown in Fig. 6, and it indicated that compound 12e
inhibited cell cycle progression at G2M stage.
Figure 6. Cell cycle profiles of U226 cells treated with DMSO (a), 10 nM of 12e (b).
3. Conclusion
Proteasome was an important target for novel drug discovery and design. In this
manuscript, powerful 3D-QSAR models were developed based on our own
experimental data and were used to design new proteasome inhibitors. Five
compounds were synthesized and evaluated to inhibit the human 20S proteasome
activity. Two potent compounds 12c and 12e were screened against three MM cancer
cell lines. Of the two inhibitors, 12e was active enough to be a lead compound. The in
vivo pharmacokinetics parameters showed that this lead compound can be
20

administered intravenously. However, the structure needed to be optimized to improve
the exposure. Finally mechanism studies indicated 12e stopped the cell cycle
progression at G2M stage.
4. Experimental section
4.1. Computational details
4.1.1. Data set and biological activity
A dataset comprising 40 dipeptidyl boronic acid 20S proteasome inhibitors
previously synthesized by us was used in the present study. All compounds and
biological data were taken from our previously published paper.27 The in vitro affinity
values were reported as IC50 in the same biological assay conditions. Finally 40
compounds with IC50 values ranging from 0.76 to 476 nM were used in the study,
including bortezomib (shown in Table 1). For the QSAR analyses, the IC50 values
were expressed in negative logarithmic units, pIC50 (-log IC50).
4.1.2. Molecular modeling
3D-QSAR calculations were performed using SYBYL 7.3 software package
(Tripos Inc., St. Louis, MO) running on Red Hat Enterprise linux workstation.39
The template molecule bortezomib was extracted from the crystal structure of
bortezomib-20S proteasome complex (PDB 2F16).30 And after structure correction
and minimization, partial atomic charges were calculated using the Gasteiger-Hückel
(GH) method.31 And the rest of the molecules were created from it using the ‘sketch
molecule’ function in SYBYL. A constrained minimization followed by full
minimization was carried out on these molecules in order to prevent the
conformations moving to false regions. Tripos force field with a distance-dependent
dielectric and the Powell conjugate-gradient algorithm with a convergence criterion of
0.05 kcal/mol were used. Partial atomic charges were also calculated using the GH
method.
4.1.3. Alignment of the molecules
We used the ‘DATABASE ALIGNMENT’ option in SYBYL to align our
compounds with bortezomib as the target structure. The common 13 atoms labeled
with numbers in Table 1 were selected as the reference atoms and the same rule
21

described in our published paper was used.35 The final models and all the discussions
were built based on 40 molecules.
4.1.4. CoMFA calculations
All the molecules were placed in a 3D regular lattice (2 Å spacing) extending at
least 2 Å beyond the volumes of all investigated molecules on all axes. The van der
Waals potentials and Coulombic term representing the steric and electrostatic fields
were calculated using standard Tripos force field for CoMFA. A Csp3 atom with a
formal charge of +1 and a van der Waals radius of 1.52 Å served as probe atom to
generate steric (Lennard-Jones 6-12 potential) and electrostatic (Coulombic potential)
field energies, which were obtained by summing the individual interaction energies
between each atom of the molecule and the probe atom at every grid point. The cutoff
values for both steric and electrostatic fields were set to 30.00 kcal/mol with a
distance-dependent dielectric constant.
4.1.5. CoMSIA calculations
The aligned molecules were placed in a 3D lattice with regular grid points
separated by 2 Å similar to that of CoMFA studies. CoMSIA calculated also steric and
electrostatic fields, and in addition hydrophobic, hydrogen-bond donor and acceptor
fields and it used Gaussian equations for field calculation that did not require cutoff
values. The analysis was carried out using a common probe atom with 1 Å radius,
+1.0 charge, and hydrophobic and hydrogen-bond property values of +1. The
attenuation factor α, which determined the steepness of the Gaussian function, was
assigned a default value of 0.3.
4.2. Chemistry
Unless otherwise stated, commercial reagents were used directly without any
purification, and all solvents were dried by standard methods before using. Reactions
were monitored by thin-layer chromatography on precoated silica gel GF254 plates
1
and the spots were tested under 254 nm UV light. H NMR and 13C NMR spectra
were recorded at room temperature on a Bruker ARX400 spectrometers using TMS as
an internal standard. Mass spectra were recorded on Thermo Fisher LC-MS
instruments in electrospray positive and negative ionization modes. High-resolution
22

mass spectra were obtained using a LTQ Orbitrap instrument using an electrospray
source (ESI).
4.2.1. N-ethoxycarbonylphthalimide (2)
To a solution of phthalimide (7.36 g, 50.00 mmol) dissolved in anhydrous DMF
(25 mL) was added TEA (9 mL, 65.00 mmol). Then temperature of the reaction
system was cooled to 0℃ and ethyl chloroformate (5.7 mL, 60.00 mmol) was added
dropwise. The mixture stirred at room temperature for 2 h and poured into ice water,
filtered. The soild was washed with cold water and dried to obtain 8.67 g (79.1% yield)
as white solid. mp 81.4-83.6℃. 1H NMR (CDCl3, 400 MHz) δ 1.44 (s, 3H, CH3), 4.48
(q, J = 7.1 Hz, 2H, CH2), 7.80-7.85 (m, 2H, Ar-H), 7.93-7.99 (m, 2H, Ar-H). MS
(ESI): m/z 220.1 [M+H]+.
4.2.2. (S)-2-phthalimidopropionic acid (3)
To a stirred solution of 2 in H2O (100 mL) were added L-alanine (8.9 g, 100.00
mmol) and Na2CO3 (10.6 g, 100.00 mmol). After 1.5 h, the aqueous solution was
slowly acidified with aqueous HCl (1N) until pH = 1-2, and filtered to obtain 17.37 g
1
(79.3% yield) as white solid. mp 145.8-146.6℃. H NMR (CDCl3, 400 MHz) δ 1.71
(s, 3H, CH3), 5.02 (q, J = 7.4 Hz, 1H, CH), 7.69-7.75 (m, 2H, Ar-H), 7.82-7.88 (m,
2H, Ar-H). MS (ESI): m/z 218.2 [M-H]-.
4.2.3. (S)-2-(1,3-dioxoisoindolin-2-yl)-N-(quinolin-8-yl)propanamide (4)
To a solution of 3 (17.37 g, 79.25 mmol) in anhydrous CH2Cl2 (80 mL) was
added thionyl chloride (29 mL, 396.25 mmol) and the resulting solution was refluxed
for 6 h. The solvent was evaporated to give yellow oil. DIPEA (20.5 g, 158.5 mmol)
and 8-aminoquinoline (11.4 g, 79.25 mmol) was dissolved in anhydrous CH2Cl2 (103
mL) and the obtained yellow oil dissolved in CH2Cl2 (30 mL) was added dropwise at
-20℃ for 1 h and then allowed to react at room temperature overnight. After
evaporation of the solvent and purification by column chromatography, 18.7g (71.9%
1
yield) of an orange solid was obtained. mp 180.0-181.9℃. H NMR (CDCl3, 400
MHz) δ 1.98 (s, 3H, CH3), 5.27 (q, J = 7.5 Hz, 1H, CH), 7.42 (dd, J1 = 4.2 Hz, J2 =
8.3 Hz, 1H, Ar-H), 7.51 (s, 1H, Ar-H), 7.53 (d, J = 9.0, 1H, pyrazine-H), 7.65-7.85 (m,
2H, Ar-H), 7.90 (dd, J1 = 3.6, J2 = 7.1 Hz, 2H, Ar-H), 8.15 (d, J = 8.3 Hz, 1H,
23

pyrazine-H), 8.69 (d, J = 4.2 Hz, 1H, Ar-H), 8.73 (dd, J1 = 4.7, J2 = 8.9 Hz, 1H,
pyrazine-H), 10.33 (s, 1H, CONH). MS (ESI): m/z 346.0 [M+H]+.
4.2.4. (S)-2-(1,3-dioxoisoindolin-2-yl)-N-(quinolin-8-yl)-3-(4-(trifluoromethyl)
phenyl)propanamide (5a)
To a solution of 4 (5.2 g, 15 mmol) in t-BuOH (105 mL) was added 4-
iodobenzotrifluoride (6.12 g, 22.5 mmol), Pd(OAc)2 (337 mg, 1.5 mmol) and AgBF4
(3.65 g, 18.75 mmol). The resulting solution was stirred at 85℃ for 24 h. After
cooling to room temperature, the reaction was diluted with dichloromethane (100 mL)
and triethylamine (6 mL) was added to the mixture. The mixture was maintained for 6
hours and then filtered through a pad of Celite. After evaporation of the solvent and
purification by column chromatography, 5.29g (72.1% yield) of a white solid was
obtained. mp 124.0-125.5℃. 1H NMR (CDCl3, 400 MHz) δ 3.77-3.95 (m, 2H, CH2) ,
5.47 (dd, J1 = 6.9 Hz, J2 = 9.7 Hz, 1H, CH), 7.39 (dd, J1 = 4.3 Hz, J2 = 8.3 Hz, 1H,
Ar-H), 7.42 (d, J = 8.1 Hz, 2H, Ar-H), 7.49 (d, J = 8.2 Hz, 2H, Ar-H), 7.51 (s, 1H,
Ar-H), 7.53 (t, J = 5.5 Hz, 1H, pyrazine-H), 7.68-7.78 (m, 2H, Ar-H), 7.78-7.91 (m,
2H, Ar-H), 8.12 (dd, J1 = 7.1 Hz, J2 = 14.1 Hz, 1H, pyrazine-H), 8.58 (dd, J1 = 1.5 Hz,
J2 = 4.2 Hz, 1H, Ar-H), 8.67-8.79 (m, 1H, pyrazine-H), 10.28 (s, 1H, CONH). MS
(ESI): m/z 487.1 [M-H]-.
Compounds 5b-5c were prepared from intermediate 4 and 1-lodo-3,5-bis
(trifluoromethyl)benzene according to the same procedure as that of 5a.
4.2.5. (S)-3-(3,5-bis(trifluoromethyl)phenyl)-2-(1,3-dioxoisoindolin-2-yl)-N-
(quinolin-8-yl)propanamide (5b)
1
White solid (4.95 g, 74.0%), mp 121.1-123.8℃. H NMR (CDCl3, 400 MHz) δ
3.87 (dd, J1 = 10.7 Hz, J2 = 14.0 Hz, 1H, CH2), 3.96 (dd, J1 = 5.9 Hz, J2 = 14.3 Hz,
1H, CH2), 5.41 (dd, J1 = 6.0 Hz, J2 = 10.4 Hz, 1H, CH), 7.39 (dd, J1 = 4.2 Hz, J2 = 8.2
Hz, 1H, Ar-H), 7.47-7.59 (m, 2H, Ar-H), 7.69 (s, 1H, pyrazine-H), 7.73 (s, 2H, Ar-H),
7.76 (dd, J1 = 3.2 Hz, J2 = 5.3 Hz, 2H, Ar-H), 7.86 (dd, J1 = 3.1 Hz, J2 = 5.1 Hz, 2H,
Ar-H), 8.13 (d, J = 8.2 Hz, 1H, pyrazine-H), 8.57 (d, J = 3.8 Hz, 1H, Ar-H), 8.72 (dd,
J1 = 2.7 Hz, J2 = 5.8 Hz, 1H, pyrazine-H), 10.27 (s, 1H, CONH). MS (ESI): m/z 558.2
[M+H]+.
24

4.2.6. (S)-3-(2,4-dimethoxyphenyl)-2-(1,3-dioxoisoindolin-2-yl)-N-(quinolin-8-yl)
propanamide (5c)
1
White solid (4.36 g, 75.5%), mp 72.7-74.2℃. H NMR (CDCl3, 400 MHz) δ
3.69 (dd, J1 = 10.4 Hz, J2 = 13.8 Hz, 1H, CH2), 3.74 (s, 3H, CH3), 3.79 (dd, J1 = 5.1
Hz, J2 = 13.8 Hz, 1H, CH2), 3.86 (s, 3H, CH3), 5.57 (dd, J1 = 5.1 Hz, J2 = 10.3 Hz, 1H,
Ar-H), 6.28 (dd, J1 = 2.3 Hz, J2 = 8.2 Hz, 1H, Ar-H), 6.44 (d, J = 2.1 Hz, 1H, Ar-H),
7.03 (d, J = 8.2 Hz, 1H, Ar-H), 7.43 (dd, J1 = 4.2 Hz, J2 = 8.3 Hz, 1H, pyrazine-H),
7.49-7.58 (m, 2H, Ar-H), 7.72 (dd, J1 = 3.1 Hz, J2 = 5.4 Hz, 2H, Ar-H), 7.84 (dd, J1 =
3.5 Hz, J2 = 7.1 Hz, 2H, Ar-H), 8.16 (d, J = 8.2 Hz, 1H, pyrazine-H), 8.70 (d, J = 4.2
Hz, 1H, Ar-H), 8.78 (dd, J1 = 2.3 Hz, J2 = 6.5 Hz, 1H, pyrazine-H), 10.37 (s, 1H,
CONH). MS (ESI): m/z 480.0 [M-H]-.
4.2.7. (S)-methyl-2-(1,3-dioxoisoindolin-2-yl)-3-(4-(trifluoromethyl)phenyl)
propanoate (6a)
To a 48 mL sealed bottle was added 5a (300 mg, 0.613 mmol), BF3·Et2O (0.8
mL, 6.13 mmol) and anhydrous methanol (14 mL). The mixture was stirred at 100℃
for 25 hours. After cooling to room temperature, Et3N (1.3 mL, 9.195 mmol) was
added dropwise. After evaporation of the solvent and dissolved in CH2Cl2 (12 mL),
the solution was washed with 10% hydrochloric acid, 5% NaHCO3 and brine, dried
over anhydrous Na2SO4. After evaporation and purification by column
chromatography, 185 mg (80.1% yield) of yellow oil was obtained. 1H NMR (CDCl3,
400 MHz) δ 3.55-3.71 (m, 2H, CH2), 3.78 (s, 3H, CH3), 5.18 (dd, J1 = 5.8 Hz, J2 =
10.7 Hz, 1H, CH), 7.30 (d, J = 8.0 Hz, 2H, Ar-H), 7.46 (d, J = 8.0 Hz, 2H, Ar-H),
7.67-7.75 (m, 2H, Ar-H), 7.79 (dd, J1 = 3.6 Hz, J2 = 7.1 Hz, 2H, Ar-H). MS (ESI) :
m/z 378.3 [M+H]+.
Compounds 6b-6c were prepared from corresponding starting materials using the
similar method as 6a.
4.2.8. (S)-methyl-3-(3,5-bis(trifluoromethyl)phenyl)-2-(1,3-dioxoisoindolin-2-yl)
propanoate (6b)
1
Yellow oil (1.28 g, 79.8%). H NMR (CDCl3, 400 MHz) δ 3.57-3.68 (m, 1H,
CH2), 3.73 (dd, J1 = 5.4 Hz, J2 = 14.5 Hz, 1H, CH2), 3.79 (s, 3H, CH3), 5.14 (dd, J1 =
25

5.4 Hz, J2 = 10.6 Hz, 1H, CH), 7.63 (s, 2H, Ar-H), 7.67 (s, 1H, Ar-H), 7.70-7.77 (m,
2H, Ar-H), 7.81 (d, J = 3.2 Hz, 2H, Ar-H). MS (ESI): m/z 446.1 [M+H]+.
4.2.9. (S)-methyl-3-(2,4-dimethoxyphenyl)-2-(1,3-dioxoisoindolin-2-yl)propanoate
(6c)
Yellow oil (1.15 g, 75.2%). 1H NMR (CDCl3, 400 MHz) δ 3.36 (dd, J1 = 11.4 Hz,
J2 = 13.9 Hz, 1H, CH2), 3.55 (dd, J1 = 6.3 Hz, J2 = 12.5 Hz, 1H, CH2), 3.69 (s, 3H,
CH3), 3.72 (s, 3H, CH3), 3.77 (s, 3H, CH3), 5.30-5.37 (m, 1H, CH), 6.22 (dd, J1 = 2.4
Hz, J2 = 8.2 Hz, 1H, Ar-H), 6.33 (d, J = 2.3 Hz, 1H, Ar-H), 6.89 (d, J = 8.2 Hz, 1H,
Ar-H), 7.65-7.71 (m, 2H, Ar-H), 7.74-7.80 (m, 2H, Ar-H). MS (ESI): m/z 370.4
[M+H]+.
4.2.10. (S)-methyl-2-amino-3-(4-(trifluoromethyl)phenyl)propanoate (7a)
To a solution of 6a (745 mg, 1.975 mmol) in anhydrous methanol (19 mL) was
added ethylenediamine (297 mg, 4.938 mmol) and the resulting solution was refluxed
for 9 h. The insoluble material was filtered off and the filtrate was evaporated to give
yellow oil 311 mg (63.7%). 1H NMR (DMSO, 400 MHz) δ 2.84 (dd, J1 = 7.7 Hz, J2 =
13.3 Hz, 1H, CH2), 2.95 (dd, J1 = 9.5 Hz, J2 = 19.0 Hz, 1H, CH2), 3.59 (s, 3H, CH3),
3.61 (d, J = 6.9 Hz, 1H, CH), 7.40 (t, J = 11.9 Hz, 2H, Ar-H), 7.59 (t, J = 21.6 Hz, 2H,
Ar-H). MS (ESI): m/z 248.1 [M+H]+.
Compounds 7b-7c were prepared from corresponding starting materials using the
similar method as 7a .
4.2.11. (S)-methyl-2-amino-3-(3,5-bis(trifluoromethyl)phenyl)propanoate (7b)
Yellow oil (522 mg, 75.8%). 1H NMR (CDCl3, 400 MHz) δ 2.99 (dd, J1 = 7.9 Hz,
J2 = 13.6 Hz, 1H, CH2), 3.18 (dd, J1 = 4.9 Hz, J2 = 13.7 Hz, 1H, CH2), 3.72 (s, 3H,
CH3), 3.73-3.78 (m, 1H, CH), 7.68 (s, 2H, Ar-H), 7.76 (s, 1H, Ar-H). MS (ESI): m/z
316.2 [M+H]+.
4.2.12. (S)-methyl-2-amino-3-(2,4-dimethoxyphenyl)propanoate (7c)
Yellow oil (351 mg, 48.6%). 1H NMR (CDCl3, 400 MHz) δ 2.71 (dd, J1 = 8.1 Hz,
J2 = 13.4 Hz, 1H, CH2), 2.98 (dd, J1 = 5.5 Hz, J2 = 13.4 Hz, 1H, CH2), 3.63 (s, 3H,
CH3), 3.69 (dd, J1 = 5.5 Hz, J2 = 8.1 Hz, 1H, CH), 3.73 (s, 6H, CH3), 6.36 (dd, J1 =
26

2.4 Hz, J2 = 8.1 Hz, 2H, Ar-H), 6.93-6.98 (m, 1H, Ar-H). MS (ESI): m/z 240.1
[M+H]+.
4.2.13. (S)-2-(5-methylpyrazine-2-carboxamido)-3-phenylpropanoic acid (9a)
To a cooled solution (-10℃) of 5-methylpyrazine-2-carboxylic acid (97 mg, 0.71
mmol) dissolved in anhydrous CH2Cl2 (15 mL) was added HOBt (144 mg, 1.07
mmol). After 10 min, EDCI (204 mg, 1.07 mmol) was added. Finally 7a (176 mg,
0.71 mmol) and DIPEA (229 mg, 1.78 mmol) was added. The mixture stirred at -10℃
for 1h and at room temperature for 15 h. The mixture was washed with 10%
hydrochloric acid, 5% NaHCO3, and brine, dried over anhydrous Na2SO4. After
filtration and evaporation, the obtained crude product was directly used in the next
reaction.
The prepared methyl ester was dissolved in methanol (22 mL) and was added
LiOH·H2O (148 mg, 3.55 mmol) and H2O (6.6 mL).The mixture was stirred for 2h at
room temperature and the methanol was evaporated under vacuo. The reaction system
was acidified with 1 N HCl until the pH value was 2. The product was extracted with
ethyl acetate and dried over anhydrous Na2SO4. After filtration and evaporation, 191
1
mg (85.2%) of yellow solid was obtained. mp 208.6-209.2℃. H NMR (DMSO, 400
MHz) δ 2.43-2.51 (s, 3H, CH3), 3.21-3.23 (m, 2H, CH2), 4.71-4.77 (m, 1H, CH),
7.18-7.24 (m, 5H, Ar-H), 8.62 (s, 1H, pyrazine-H), 8.77 (d, J = 8.2 Hz, 1H, CONH),
9.00 (s, 1H, pyrazine-H). MS (ESI): m/z 316.4 [M-H]-.
Compounds 9b-9c were prepared from corresponding starting materials using the
similar method as 9a .
4.2.14. (S)-2-(pyrazine-2-carboxamido)-3-(4-(trifluoromethyl)phenyl)propanoic
acid (9b)
Yellow soild (279 mg, 73.4%). mp 176.8-178.9℃. 1H NMR (DMSO, 400 MHz)
δ 3.31 (s, 1H, CH2), 3.33-3.47 (m, 1H, CH2), 4.78 (dd, J1 = 5.4 Hz, J2 = 8.5 Hz, 1H,
CH), 7.48 (d, J = 8.1 Hz, 2H, Ar-H), 7.61 (d, J = 8.2 Hz, 2H, Ar-H), 8.70-8.77 (m, 1H,
pyrazine-H), 8.88 (d, J = 2.5 Hz, 1H, pyrazine-H), 9.04 (d, J = 8.3 Hz, 1H, CONH),
9.13 (d, J = 1.3 Hz, 1H, pyrazine-H), 13.13 (s, 1H, COOH). MS (ESI): m/z 338.2
[M-H]-.
27

4.2.15. (S)-2-(2,5-dichlorobenzamido)-3-(4-(trifluoromethyl)phenyl)propanoic
acid (9c)
Yellow soild (106 mg, 72.3%). mp 185.1-186.9℃. 1H NMR (DMSO, 400 MHz)
δ 3.02 (dd, J1 = 10.6 Hz, J2 = 13.8 Hz, 1H, CH2), 3.30 (dd, J1 = 4.6 Hz, J2 = 13.9 Hz,
1H, CH2), 4.68 (ddd, J1 = 4.7 Hz, J2 = 8.4 Hz, J3 = 10.4 Hz,1H, CH), 7.15 (d, J = 1.8
Hz, 1H, Ar-H), 7.49 (t, J = 4.9 Hz, 2H, Ar-H), 7.52 (d, J = 6.0 Hz, 2H, Ar-H), 7.66 (d,
J = 8.1 Hz, 2H, Ar-H), 8.90 (d, J = 8.2 Hz, 1H, CONH), 13.12 (s, 1H, COOH). MS
(ESI): m/z 403.9 [M-H]-.
4.2.16. (S)-3-(3,5-bis(trifluoromethyl)phenyl)-2-(2,5-dichlorobenzamido)
propanoic acid (9d)
White soild (651 mg, 75.4%). mp 195.4-196.5℃. 1H NMR (DMSO, 400 MHz) δ
2.99-3.20 (m, 1H, CH2), 3.44 (dd, J1 = 4.0 Hz, J2 = 13.8 Hz, 1H, CH2), 4.61-4.95 (m,
1H, CH), 7.14 (s, 1H, Ar-H), 7.39-7.65 (m, 2H, Ar-H), 7.73-8.31 (m, 3H, Ar-H), 8.97
(d, J = 8.4 Hz, 1H, CONH), 13.07 (s, 1H, COOH). MS (ESI): m/z 471.8 [M-H]-.
4.2.17. (S)-2-(2,5-dichlorobenzamido)-3-(2,4-dimethoxyphenyl)propanoic
acid (9e)
White soild (409 mg, 82.6%). mp 183.4-185.5℃. 1H NMR (DMSO, 400 MHz) δ
2.72 (dd, J1 = 11.8 Hz, J2 = 23.6 Hz, 1H, CH2), 3.14 (dd, J1 = 4.7 Hz, J2 = 13.5 Hz,
1H, CH2), 3.73 (s, 3H, CH3), 3.78 (s, 3H, CH3), 4.59 (dd, J1 = 4.9 Hz, J2 = 9.6 Hz, 1H,
CH), 6.43 (dd, J1 = 2.2 Hz, J2 = 8.3 Hz, 1H, Ar-H), 6.53 (d, J = 2.1 Hz, 1H, Ar-H),
7.08 (d, J = 8.3 Hz, 1H, Ar-H), 7.14 (s, 1H, Ar-H), 7.43-7.57 (m, 2H, Ar-H), 8.74 (d,
J = 8.3 Hz, 1H, CONH), 12.83 (s, 1H, COOH). MS (ESI): m/z 395.9 [M-H]-.
4.2.18. 5-methyl-N-((S)-1-(((R)-3-methyl-1-((3aS,4S,6S,7aR)-3a,5,5-trimethylhex
ahydro-4,6-methanobenzo[d][1,3,2]dioxaborol-2-yl)butyl)amino)-1-oxo-3-phenyl
propan-2-yl)pyrazine-2-carboxamide (11a)
To a cooled solution (-10℃) of 9a (250 mg, 0.79 mmol) dissolved in anhydrous
CH2Cl2 (17 mL) was added HOBt (160 mg, 1.19 mmol). After 10 min, EDCI (228 mg,
1.19 mmol) was added. Finally (aR,3aS,4S,6S,7aR)- Hexahydro-3a,8,8-trimethyl-
alpha-(2-methylpropyl)-4,6-methano-1,3,2-benzodioxaborole-2-methanamine-2,2,2-
trifluoroacetate 10 (300 mg, 0.79 mmol) and DIPEA (357 mg, 2.77 mmol) were
28

added. The mixture stirred at -10℃ for 1h and at room temperature for 15 h. The
mixture was washed with 10% hydrochloric acid, 5% NaHCO3, and brine, dried over
anhydrous Na2SO4. After evaporation and purification with chromatography, 351 mg
1
(83.6%) of glassy solid was obtained. H NMR (CDCl3, 400 MHz) δ 0.82 (s, 3H,
CH3), 0.84 (s, 6H, CH3), 1.28 (s, 3H, CH3), 1.34 (d, J = 4.9 Hz, 2H, CH2), 1.39 (s, 3H,
CH3), 1.41-1.46 (m, 1H, CH), 1.87 (d, J = 20.0 Hz, 2H, CH2), 1.96-2.08 (m, 2H, CH2),
2.18 (d, J = 10.7 Hz, 1H, CH), 2.28-2.37 (m, 1H, CH), 2.64 (s, 3H, CH3), 3.10-3.19
(m, 2H, CH2), 3.19-3.27 (m, 1H, CH), 4.30 (d, J = 8.8 Hz, 1H, CH), 4.79 (dd, J1 = 7.0
Hz, J2 = 14.4 Hz, 1H, CH), 5.89 (dd, J1 = 4.6 Hz, J2 = 12.7 Hz, 1H, CONH), 7.22 (d,
J = 4.3 Hz, 1H, Ar-H), 7.28 (d, J = 3.8 Hz, 4H, Ar-H), 8.24-8.35 (m, 1H, CONH),
8.37 (s, 1H, pyrazine-H), 9.20 (s, 1H, pyrazine-H). MS (ESI): m/z 533.3 [M+H]+.
Compounds 11b-11e were prepared from corresponding starting materials using
the similar method as 11a.
4.2.19. N-((S)-1-(((R)-3-methyl-1-((3aS,4S,6S,7aR)-3a,5,5-trimethylhexahydro-
4,6-methanobenzo[d][1,3,2]dioxaborol-2-yl)butyl)amino)-1-oxo-3-(4-(trifluorome
thyl)phenyl)propan-2-yl)pyrazine-2-carboxamide (11b)
Glassy solid (359 mg, 82.8%). 1H NMR (CDCl3, 400 MHz) δ 0.80 (s, 3H, CH3),
0.83 (s, 6H, CH3), 1.17 (dd, J1 = 10.9 Hz, J2 =17.6 Hz, 1H, CH), 1.24 (s, 2H, CH2),
1.27 (s, 3H, CH3), 1.38 (s, 3H, CH3), 1.75 (s, 1H, CH), 1.86 (ddd, J1 = 2.8 Hz, J2 = 5.8
Hz, J3 = 11.1 Hz, 1H, CH2), 1.90-1.96 (m, 1H, CH2), 2.01 (dd, J1 = 7.5 Hz, J2 = 12.7
Hz, 1H, CH), 2.15 (ddd, J1 = 5.2 Hz, J2 = 10.0 Hz, J3 = 10.9 Hz, 1H, CH2), 2.25-2.37
(m, 1H, CH2), 3.15-3.23 (m, 2H, CH2), 3.24-3.27 (m, 1H, CH), 4.27 (ddd, J1 = 1.8 Hz,
J2 = 8.7 Hz, J3 = 15.0 Hz, 1H, CH), 4.77-4.91 (m, 1H, CH), 6.03 (dd, J1 = 5.7 Hz, J2 =
11.5 Hz, 1H, CONH), 7.40 (d, J = 8.0 Hz, 2H, Ar-H), 7.52 (t, J = 7.3 Hz, 2H, Ar-H),
8.40 (dd, J1 = 4.5 Hz, J2 = 8.2 Hz, 1H, CONH), 8.50-8.57 (m, 1H, pyrazine-H), 8.76
(d, J = 2.3 Hz, 1H, pyrazine-H), 9.29-9.36 (m, 1H, pyrazine-H). MS (ESI): m/z 587.4
[M+H]+.
4.2.20. 2,5-dichloro-N-((S)-1-(((R)-3-methyl-1-((3aS,4S,6S,7aR)-3a,5,5-trimethyl
hexahydro-4,6-methanobenzo[d][1,3,2]dioxaborol-2-yl)butyl)amino)-1-oxo-3-(4-(
trifluoromethyl)phenyl)propan-2-yl)benzamide (11c)
29