2-Substituted-2-amino-6-boronohexanoic acids as arginase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2013.02.024

Substitution at the alpha center of the known human arginase inhibitor 2-amino-6-boronohexanoic acid (ABH) is acceptable in the active site pockets of both human arginase I and arginase II. In particular, substituents with a tertiary amine linked via a two carbon chain show improved inhibitory potency for both enzyme isoforms. This potency improvement can be rationalized by X-ray crystallography, which shows a water-mediated contact between the basic nitrogen and the carboxylic acid side chain of Asp200, which is situated at the mouth of the active site pocket of arginase II (Asp181 in arginase I). We believe that this is the first literature report of compounds with improved arginase inhibitory activity, relative to ABH, and represents a promising starting point for further optimization of in vitro potency and the identification of better tool molecules for in vivo investigations of the potential pathophysiological roles of arginases.

Full text of DOI:10.1016/j.bmcl.2013.02.024

Bioorganic & Medicinal Chemistry Letters 23 (2013) 2027–2030
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
2-Substituted-2-amino-6-boronohexanoic acids as arginase
inhibitors
a
a
a
a
Adam Golebiowski a, , R. Paul Beckett , Michael Van Zandt , Min Koo Ji , Darren Whitehouse ,
Todd R. Ryder a, Erik Jagdmann a, Monica Andreoli a, Adam Mazur b, Manyian Padmanilayam b,
Alexandra Cousido-Siah c, Andre Mitschler c, Francesc X. Ruiz c, Alberto Podjarny c, Hagen Schroeter d
⇑
a Institutes for Pharmaceutical Discovery, 23 Business Park Drive, Branford, CT, United States
b Girindus America Inc, 8560 Reading Rd, Cincinnati, OH 45215, United States
c Department of Integrative Biology, IGBMC, CNRS, INSERM, Université de Strasbourg, 1 rue Laurent Fries, Illkirch, France
d Mars Incorporated, 6885 Elm Street, McLean, VA 22101, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Substitution at the alpha center of the known human arginase inhibitor 2-amino-6-boronohexanoic acid
(ABH) is acceptable in the active site pockets of both human arginase I and arginase II. In particular, sub-
stituents with a tertiary amine linked via a two carbon chain show improved inhibitory potency for both
enzyme isoforms. This potency improvement can be rationalized by X-ray crystallography, which shows
a water-mediated contact between the basic nitrogen and the carboxylic acid side chain of Asp200, which
is situated at the mouth of the active site pocket of arginase II (Asp181 in arginase I). We believe that this
is the first literature report of compounds with improved arginase inhibitory activity, relative to ABH, and
represents a promising starting point for further optimization of in vitro potency and the identification of
better tool molecules for in vivo investigations of the potential pathophysiological roles of arginases.
Ó 2013 Elsevier Ltd. All rights reserved.
Received 22 December 2012
Revised 26 January 2013
Accepted 1 February 2013
Available online 13 February 2013
Keywords:
Arginase
Inhibitor
Metalloenzyme
Boronic acid
ABH
Arginase is a binuclear manganese metalloenzyme that cata-
lyzes hydrolysis of -arginine to urea and -ornithine. Two isoforms
Recently, Ilies et al. have reported the synthesis and character-
ization of the quaternary amino acids 2-amino-6-borono-2-
methyl-hexanoic acid (MABH) and 2-amino-6-borono-2-(difluoro-
methyl)-hexanoic acid (FABH), both of which showed decreased
inhibitory potency, relative to ABH, against human Arg I.6
L
L
are known, arginase I (Arg I) and arginase II (Arg II), which have dif-
ferent tissue distributions and biological functions. Arg I is ex-
pressed primarily in the liver, whereas Arg II is expressed in
several extra-hepatic tissues, including the brain, spinal cord, kid-
ney, small intestine, and mammary gland. The two isoforms are en-
coded by different genes, and have differing pI values,
immunological reactivity, and subcellular locations, but share sim-
ilar enzymatic reactivities.1
Human arginases are implicated in wide variety of diseases,
including atherosclerosis, pulmonary hypertension, systemic
hypertension, erectile dysfunction, asthma, wound healing, multi-
ple sclerosis,2 as well as several tropical diseases such as malaria
and leishmaniasis.3 Development of potent and selective arginase
inhibitors would help to elucidate the pathophysio-logical roles
that these enzymes play and potentially lead to novel therapies.
Several compounds have already been reported, for example,
2(S)-amino-6-boronohexanoic acid (ABH),4 which is the arginase
inhibitor most widely used and reported in the literature.5
NH2
H3C NH2
HO2C
F2HC NH2
HO2C
OH
OH
OH
HO2C
B
B
B
OH
OH
OH
ABH
MABH
FABH
racemic - our data:
Arg1 1.4 uM (n=25)
optically active ref 4:
Arg1 0.8 uM;
Independently, we have developed a similar approach and pres-
ent here the first steps in our research effort that, in contrast, re-
sulted in significant improvements in arginase inhibitory potency
compared to ABH.7
Our analysis of the binding mode of ABH to arginase I (PDB:
1D3V) revealed that while the whole molecule is tightly embedded
in the active site pocket, its alpha-hydrogen projects out towards
various surface residues of the enzyme and towards bulk solvent.
Therefore, we reasoned that addition of appropriate functionality
⇑
Corresponding author. Tel.: +1 203 315 5947; fax: +1 203 315 4002.
E-mail address: golebiowski.a@gmail.com (A. Golebiowski).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.02.024

2028
A. Golebiowski et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2027–2030
CO2Et
CO2Et
at this position might create new interactions and increase potency
activity.
Most compounds were prepared by successive alkylations of
the glycine equivalent 2, using methodology developed by O’Don-
nell8 (Scheme 1) or a variation thereof, that is, by alkylation of ben-
zophenone protected amino acids (e.g., 1d or 1j).
b
CO2Et
CO2Et
a
BocHN
CO2Et
BocHN
BocHN
CO2Et
CO2H
BocHN
9
10
11
c
CO2Et
O
B
d, e
Alkylation with the commercially available bromide 3 allowed
installation of the protected n-butyl boronic acid fragment in one
easy step. The second alkylation (4 ? 5) required more aggressive
reaction conditions or the use of more active alkylating agents,
which limited the choice of potential side chains. The alkylated
products 5 were deprotected under acidic conditions to provide
the desired final products 1b–1e as racemates. Alternatively, alkyl-
ation with allyl bromide allowed us to prepare compound 6. Ozon-
olysis of 6 followed by reductive amination and deprotection
provided the desired products 1m–1p. Final compounds obtained
after acidic deprotection were very polar and were purified using
reverse phase HPLC and lyophilized.
Compounds were tested for inhibition of human Arg I and Arg II
using a colorimetric assay based upon the published method.9,10
The inhibition of arginases by boronic acid inhibitors is strongly
pH dependent.11 Although the pH optimum for arginase is 9.0–
9.5,12 we decided to perform our assay at pH 7.4 which is more
physiologically relevant. Our observed inhibitory potency for ABH
for Arg I is very similar to the value reported in the literature
(IC50 = 800 nM; optically active,4 vs 1450 nM for racemic com-
pound/our data).
2,3-Diaminopropionic acid derivatives 1k and 1l were prepared
starting from diethyl (Boc-amino)malonate (Scheme 2). Although
the key step (the partial hydrolysis of the diester 10 to mono acid
11) was achieved in non-selective way, it does provided a potential
entry into enantioselective synthesis (e.g., via chemoenzymatic
desymmetrization).13 Acidic hydrolysis of intermediate 13 pro-
vided compound 1f.
The initial compounds with alkyl and benzyl substituents at the
alpha-position (1b–1e) had similar or lower potencies to the
unsubstituted compound ABH (1a), but these findings confirmed
that there is enough room to accommodate a larger alpha-substitu-
ent. Analogs with one to three carbon atom aliphatic chain-linked
alcohols and ethers were synthesized (e.g., 1f–1i) and tested in the
hope of identifying new interactions at the opening of the active
site pocket of Arg I, for example, Thr136, Ser137, Asp181 Asp183
(Ser155, Ser156, Asp200, Asp202, respectively in Arg II). Most of
these compounds in this class showed little or no improvement
in potency compared to ABH. An X-ray crystal structure of com-
OH
O
O
NHBoc
OAc
13
12
f, g, h
O
B
OH
B
CO2Et
CO2H
NH2
i
HO
NHBoc
R2
N
R2
N
R1
14
1k-1l
R1
Scheme 2. Synthesis of inhibitors 1k and 1l. Reagents and conditions: (a) 4-
bromobut-1-ene, NaH, DMF, 90 °C (90%); (b) KOH, EtOH, 0 °C to rt, 19 h, (82%); (c)
ClCO2Et, Et3N, then NaBH4 À40 °C, 1 h, (78%); (d) Ac2O, DMAP 1 equiv, DCM, rt, 16 h,
(80%); (e) pinacol borane, chloro(1,5-cyclooctadiene)iridium(I) dimer, 1,2-bis(di-
phenyl-phosphino)ethane, DCM, rt, 10 h, (54%); (f) K2CO3, EtOH, rt, 2 h, (78%); (g)
(COCl)2, DMSO, Et3N, DCM, À78 °C, 30 min, (57%); (h) R2NH, 1,2-DCE, NaBH(OAc)3,
AcOH, 60 °C, 2 h (30–50% yield); (i) 6 N HCl, 95 °C, 16 h (50–80% yield).
pound 1h bound to Arg II measured at 2.2 Å resolution on a labo-
ratory X-ray source,14,15 revealed a contact distance of around
2.9 Å between the terminal hydroxyl group and a single oxygen
atom of Asp202 (Fig. 1). The corresponding structure of 1h bound
to Arg I (2.0 Å resolution) revealed an almost identical binding
mode, with the terminal hydroxyl group oriented toward Thr136
at distance 3.2 Å. In both Arg I and Arg II complexes, all other en-
zyme inhibitor interactions remained essentially unchanged com-
pared to the respective ABH–enzyme complexes (data not shown).
Compounds 1f–1i are representative examples of alpha-linked
C1–C3 chains with non-basic hydrogen bond donor or acceptor
groups. Several other aliphatic chain linked ureas, amides and sul-
fonamides were synthesized and tested with similar results (IC50
potencies in 1–5 lM range). Based on the improved potency seen
for the histidine analog 1j, compared to alpha-benzyl analog 1e,
efforts were focused on increasing the interaction of the
Ph CO2R
O
B
a
Ph CO2R
+
Br
2
O
Ph
b
N
B
O
Ph
N
O
3
4
CO2H
R1
Ph CO2R
g
OH
c
O
H2
N
B
Ph
N
B
O
R1
OH
1b-1e
5
CO2H
Ph CO2R
N
OH
d, e, f, g
H2
N
B
O
Ph
B
OH
O
N
1m-1p
6
R
R
Scheme 1. Synthesis of inhibitors 1b–1e, 1g–1j and 1m–1p. Reagents and
conditions: (a) LiHMDS, THF, À78 °C to +50 °C (64%); (b) LiHMDS, R1CH2I, THF,
À78 °C to rt; (c) LiHMDS, allyl bromide 3 equiv, THF, À78 °C to rt, 16 h (59%); (d)
Et2O–1 N HCl, rt then Boc2O, EtOAc–NaHCO3 sat (80%); (e) O3, DCM, À78 °C then
Ph3P À78 °C to rt (93%); (f) R2NH, 1,2-DCE, NaBH(OAc)3, rt, 16 h (70–90% yield); (g)
6 N HCl, 95 °C, 16 h (50–90% yield).
Figure 1. X-ray structure of compound 1h in human Arg II.

A. Golebiowski et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2027–2030
2029
Table 1
SAR for
a,a
-disubstituted ABH analoguesa
R
H2N
OH
B
HO2C
OH
1
Entry
R
Arg I
Arg II
Arg I/Arg II ratio
IC50 (nM) IC50 (nM)
H
1a (ABH)
1450
3430
4100
1920
5070
4600
0.75
0.68
0.89
Me
Et
1b
1c
1d
1e
11,240
10,500
15,500
17,000
0.72
0.62
1f
OH
OH
2250
4280
0.53
0.33
1gb
19,000
57,000
1h
1i
OH
NTc
4500
6440
NA
O
5120
0.80
H
Figure 2. X-ray structure of compound 1o in human Arg II.
N
1j
2280
3720
0.61
N
NH2
1k
1l
2070
3260
1350
5980
1.53
0.55
alpha-substituent with the acidic side chains of Asp200 and/or
Asp202 (Arg II numbering). Although the 2,3-diaminopropionic
acid analog 1k had a similar potency to the simple ethyl analog
1c, we found that potency increases could be achieved by extend-
ing the spacer to two carbon atoms (compounds 1m–1p). Interest-
ingly, the morpholine compound 1m and the related piperazine
and thiomorpholine analogs (not shown) exhibited substantially
N
H
O
1m
1n
5380
520
4020
934
1.34
0.56
N
N
lower activity (IC50s in the 2–4 lM range). Simple ethylene-linked
OH
tertiary amines (represented by compound 1n–1p) demonstrated a
clear improvement (5- to 10-fold) in potency compared to com-
pounds lacking the amine and provided a promising starting point
for further optimization of in vitro activity. This improvement in
potency was rationalized by a co-crystal structure obtained at
2.35 Å in a laboratory X-ray source of 1o with Arg II, which re-
vealed a water-mediated interaction between the tertiary amine
of the inhibitor side chain and the carboxylic acid side chain of
Asp200, (Fig. 2), rather than with Asp202, as in the complex with
the hydroxypropyl compound 1h (Fig. 1). The close proximity of
the water molecule to the amine nitrogen (2.47 Å) and the near
symmetrical interactions of the water with both carboxylic acid
oxygens (3.19 and 3.72 Å) might suggest that these functional
groups are present as ammonium and carboxylate ions, respec-
tively. Such an interaction is not possible with the hydroxyl group
in 1h which may, at least in part, explain the difference in potency.
It is apparent from the structure–activity data (Table 1) that the
compounds presented here show little evidence of in vitro selectiv-
ity between Arg I and Arg II which ultimately may be desirable for
a therapeutic. Given the close homology between Arg I and Arg II in
the active site region, this is not surprising and may represent a
significant challenge in the future.
1o
1p
370
320
980
620
0.38
0.52
N
N
a
b
c
All compounds were tested as their racemates.
Cyclizes to the lactone under the assay condition.
Not tested.
Acknowledgments
This work was funded by Mars, Incorporated. We thank the
IGBMC Structural genomics platform staff (in particular Pierre
Poussin-Courmontagne and Dr. Alastair McEwen). Infrastructures
used in this research were supported by the Centre National de
la Recherche Scientifique (CNRS), the Institut National de la Santé
et de la Recherche Médicale, the Hôpital Universitaire de Stras-
bourg (H.U.S) and the Université de Strasbourg.
Supplementary data
In conclusion, we have shown that, relative to the literature
benchmark compound ABH, improvements in arginase inhibitory
potency can be achieved by targeting new interactions with the
amino acid residues of the enzyme at the mouth of the active site
pocket, particularly Ser155 and Asp200 in the case of Arg II. Further
optimization of these novel compounds, including synthesis of
optically active analogs, further improvement of in vitro activity
as well as pharmacokinetic data and in vivo POC for myocardial
ischemia, will be reported elsewhere.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2013.
02.024.
References and notes
1. Morris, S. M., Jr. Br. J. Pharmacol. 2009, 157, 922.
2. (a) Ryoo, S.; Gupta, G.; Benjo, A.; Lim, H. K.; Camara, A.; Sikka, G.; Lim, H. K.;
Sohi, J.; Santhanam, L.; Soucy, K.; Tuday, E.; Baraban, E.; Ilies, M.; Gerstenblith,

2030
A. Golebiowski et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2027–2030
G.; Nyhan, D.; Shoukas, A.; Christianson, D. W.; Alp, N. J.; Champion, H. C.;
(background) was subtracted from all the values. The % inhibition of Arg I
activity was calculated in the presence of each compound and the Prism
application was used to generate an IC50 curve and to determine the IC50 values
for each compound (n = 3).
Huso, D.; Berkowitz, D. E. Circ. Res. 2008, 102, 923; (b) Morris, C. R.; Gladwin, M.
T.; Kato, G. J. Curr. Mol. Med. 2008, 8, 620; (c) Michell, D. L.; Andrews, K. L.; Chin-
Dusting, J. P. Front. Biosci. (Schol. Ed). 2011, 3, 946; (d) Cama, E.; Colleluori, D.
M.; Emig, F. A.; Shin, H.; Kim, S. W.; Kim, N. N.; Traish, A.; Ash, D. E.;
Christianson, D. W. Biochemistry 2003, 42, 8445; (e) Munder, M. Expert Rev. Clin.
Pharmacol. 2010, 3, 17; (f) Witte, M. B.; Barbul, A. Wound Repair Regen. 2003, 6,
419; (g) Xu, L.; Hilliard, B.; Carmody, R. J.; Tsabary, G.; Shin, H.; Christianson, D.
W.; Chen, Y. H. Immunology 2003, 1, 141.
10. Jung, D.; Biggs, H.; Erikson, J.; Ledyard, P. U. Clin. Chem. 1975, 21, 1136.
11. Colleluori, D. M.; Ash, D. E. Biochemistry 2001, 40, 9356.
12. Roholt, O. A.; Greenberg, D. M. Arch. Biochem. Biophys. 1956, 62, 454.
13. Schoffers, E.; Golebiowski, A.; Johnson, C. R. Tetrahedron 1996, 52, 3769.
14. Crystals of the human Arg I/inhibitor complexes were obtained by soaking
crystals of the native enzyme with the inhibitor. Crystals of the native enzyme
3. (a) Dowling, D. P.; Ilies, M.; Olszewski, K. L.; Portugal, S.; Mota, M. M.; Llinás,
M.; Christianson, D. W. Biochemistry 2010, 49, 5600; (b) Abebe, T.; Hailu, A.;
Woldeyes, M.; Mekonen, W.; Bilcha, K.; Cloke, T.; Fry, L.; Seich Al Basatena, N.
K.; Corware, N. K.; Modolell, M.; Munder, M.; Tacchini-Cottier, F.; Müller, I.;
Kropf, P. PLoS Negl. Trop. Dis. 2012, 6, e1684.
4. Baggio, R.; Elbam, D.; Kanyo, Z. F.; Carroll, P. J.; Cavalli, R. C.; Ash, D. E.;
Christianson, D. W. J. Am. Chem. Soc. 1997, 119, 8107.
5. Christianson, D. W. Acc. Chem. Res. 2005, 38, 191.
6. Ilies, M.; Di Costanzo, L.; Dowling, D. P.; Thorn, K. J.; Christianson, D. W. J. Med.
Chem. 2011, 54, 5432.
7. Van Zandt, M.; Golebiowski, A.; Ji, M. K.; Whitehouse, D.; Todd, R.; Beckett, P.
WO 2011 133653.
8. O’Donnell, M. J. Aldrichim. Acta 2001, 34, 3.
9. Arg I converts L-arginine to L-ornithine and urea, and relative Arg I activity is
Arg I were obtained by vapor diffusion method at 4 °C. Drops containing 4.0
of protein at 5 mg/mL in 10 mM TrisÁHCl pH 7.5, 10 mM NaCl, 1 mM b-
mercaptoethanol, 0.2 mM MnSO4 and 4 L of precipitant solution [100 mM
lL
l
malonic acid, imidazol, boric system pH 5, 25% polyethylene glycol (PEG) 1500]
were equilibrated against a reservoir with precipitant solution. Then the native
crystals of Arg I were soaked with 15 mM of the inhibitor during 1 week. They
were cryo-protected by the precipitant solution containing 30% ethylene glycol
prior to flash cooling in liquid nitrogen.
Crystals of human Arg II/inhibitor complexes were obtained by co-
crystallization with 2 mM of inhibitor. Crystals of complexes were obtained
by vapor diffusion method at 4 °C. Protein was concentrated at 5 mg/mL in
10 mM TrisÁHCl pH 8.0, 10 mM NaCl, 1 mM b-mercaptoethanol, 0.2 mM
MnSO4. Silver Bullets screen (Hampton Research) allowed to obtained
crystals. They were cryo-protected by the reservoir solution containing 30%
ethylene glycol prior to flash cooling in liquid nitrogen.
determined by measuring urea levels in a colorimetric assay according to the
published method.10 We used purified recombinant full length human Arg I
protein to develop
dissolved in DMSO. Both Arg I and compounds were further diluted in the assay
buffer (0.1 M NaPO4, 130 mM NaCl, 1 mg/mL ovalbumin, pH 7.4) prior to
addition to well. L-Arginine (Sigma) was dissolved in a solution containing
glycine and MnSO4. Final concentration of MnSO4 and glycine was 0.3, and
13.8 mM, respectively, in each well. Each compound was tested at the final
a
96 well plate Arg
I
assay. Tested compounds were
15. Structural refinement and structure analysis: the atomic coordinates of rat
arginase protein complexed with a boronic inhibitor (Protein Data Bank (PDB)
code 1D3V) were used to solve the structure of all Arg I/II: inhibitor complexes.
Crystallographic refinement was performed initially with the CCP4 suite and
later with the program PHENIX. Amino acid side-chains were fitted into
2Fo À Fc and Fo À Fc electron density maps. The final Fo À Fc map indicated
clear electron density for the different inhibitors. Water molecules were fitted
into difference maps and in the final cycles riding H-atoms were introduced.
The program Coot was used for fitting the models to the electron density. The
Arg I complex was solved from a twin crystal and was refined according to
‘Àh,Àk, l’ as twin law. The atomic coordinates have been deposited in the PDB
(PDB codes: 4IE1, 4IE2 and 4IE3) and will be released upon publication. In
order to study the inhibitor binding sites, Coot was used for analysis, while
figures were built with the PyMOL Molecular Graphics System (Schrödinger
LLC). Data collection and refinement statistics are shown in the Supplementary
data.
concentration of 0.005, 0.01, 0.04, 0.12, 0.37, 1.11, 3.33, and 10
Compounds, Arg I (67 ng/mL–1.34 nM), and -arginine (8.3 mM) were added
lM.
L
to each well and the plate was incubated at 37 °C for 1 h. For experimental
determination of the Ki, inhibition was determined using six concentrations of
compound and eight concentrations of substrate. Duplicate wells were done for
each condition. The amount of urea in each well was measured by addition of a
chromogenic substrate which binds to urea. The absorbance was measured at
wavelength 530 nm using a spectroMAX plate reader. The absorbance in each
well is directly proportional to the amount of urea produced by arginase-1. The
urea production in the absence of any compound was considered the
maximum arginase-1 activity. The absorbance in the absence of L-arginine