Discovery of bacterial NAD+-dependent DNA ligase inhibitors: Improvements in clearance of adenosine series

Article abstract of DOI:10.1016/j.bmcl.2011.11.071

Optimization of clearance of adenosine inhibitors of bacterial NAD ⁺-dependent DNA ligase is discussed. To reduce Cytochrome P-450-mediated metabolic clearance, many strategies were explored; however, most modifications resulted in compounds wi

Full text of DOI:10.1016/j.bmcl.2011.11.071

Bioorganic & Medicinal Chemistry Letters 22 (2012) 85–89
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of bacterial NAD+-dependent DNA ligase inhibitors: Improvements
in clearance of adenosine series
⇑
Suzanne S. Stokes , Madhusudhan Gowravaram, Hoan Huynh, Min Lu, George B. Mullen, Brendan Chen,
Robert Albert, Thomas J. O’Shea, Michael T. Rooney, Haiqing Hu, Joseph V. Newman, Scott D. Mills
Infection Innovative Medicines Unit, AstraZeneca R&D Boston, 35 Gatehouse Drive, Waltham, MA 02451, USA
a r t i c l e i n f o
a b s t r a c t
Optimization of clearance of adenosine inhibitors of bacterial NAD+-dependent DNA ligase is discussed.
To reduce Cytochrome P-450-mediated metabolic clearance, many strategies were explored; however,
most modifications resulted in compounds with reduced antibacterial activity and/or unchanged total
clearance. The alkyl side chains of the 2-cycloalkoxyadenosines were fluorinated, and compounds with
moderate antibacterial activity and favorable pharmacokinetic properties in rat and dog were identified.
Ó 2011 Elsevier Ltd. All rights reserved.
Article history:
Received 12 October 2011
Revised 16 November 2011
Accepted 18 November 2011
Available online 23 November 2011
Keywords:
Bacterial DNA ligase
Antibacterial
LigA
Adenosines
Oxidative metabolism
Over the past three decades, there has been an increase in infec-
tions caused by antibiotic-resistant pathogens and a concomitant
decrease in the number of new antibacterial compounds in the
drug development pipeline.1 This confluence of circumstances
has resulted in the urgent need for new antibacterial drugs.
The discovery and preliminary SAR of an adenosine series of
NAD+-dependent DNA ligase (LigA) inhibitors with antibacterial
activity (Fig. 1) was previously described.2 NAD+-dependent DNA
ligases are essential enzymes for DNA replication, repair, and
recombination in all bacterial species.3 Because of their essential
nature and distinct structure compared to human homologs, LigA
represents a potentially valuable target for identifying novel and
selective antibacterial agents.4 The adenosine series of LigA inhib-
itors was identified using high-throughput screening (HTS) and
subsequently optimized for antibacterial activity against the Gram
positive pathogens Streptococcus pneumoniae (S. pneumoniae) and
Staphylococcus aureus (S. aureus) by modification of the 2-position
of the adenine ring and the 30- and 50-substituents on the ribose.2
Adenosine analogs were identified with antibacterial activity
and some drug-like physical properties such as high solubility
and low plasma protein binding. These compounds had small or
medium-sized alkyl rings attached via an ether to the 2-position
of the adenine, and the sugar was 50-deoxyribose or 50-deoxy-
50-fluororibose. However, these 2-alkoxy-50-deoxyadenosines were
rapidly cleared in rat pharmacokinetic studies (Table 1), with
1: X = F; R = cyclopentyl
2: X = H; R = cyclopentyl
NH2
3: X = F; R =
*
N
N
N
R
N
O
4: X = H; R =
2
*
O
*
OH
5: X = H; R =
X
3'
5'
OH
6: X = H; R =
*
Figure 1. Examples of antibacterial adenosine LigA inhibitors, 1–6. Substituents at
2-position of adenine ring and 50 of the ribose are noted.
Table 1
In vivo rat clearance and antibacterial activity of adenosine analogs 1–6
Compd
Clearance, rat ml/min/kg (%HBFa)
MIC Spnb
(lg/ml)
1
2
3
4
5
6
83 (>100)
43 (62)
73 (>100)
89 (>100)
96 (>100)
50 (72)
4
2
2
1
1
2
Clearance was determined by measuring the concentration of 1–6 in rat plasma
over time after iv dosing of compounds at 3 mg/kg.
a
Percent of hepatic blood flow (100% = 70 ml/min/kg).
Minimum inhibitory concentration against S. pneumoniae D39, as described in
b
⇑
Corresponding author.
E-mail address: suzanne.stokes@astrazeneca.com (S.S. Stokes).
Ref. 2c.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.11.071

86
S. S. Stokes et al. / Bioorg. Med. Chem. Lett. 22 (2012) 85–89
clearance approaching hepatic blood flow indicating the com-
pounds were rapidly cleared from systemic circulations of the ani-
mals. Evaluation of several analogs in bile duct cannulated rats
indicated that biliary clearance was negligible, and in vitro studies
indicated that compounds of the adenosine series were stable in
whole blood. Consistent with observations from microsome- and
hepatocyte-stability experiments, metabolite identification studies
demonstrated that oxidative metabolism of the cycloalkyl substi-
tuent was the major metabolic hotspot (Fig. 2). Interestingly, no
deglycosylation and no cleavage of the ether–adenine linkage were
detected. Also, only minor amounts of products from glucuronida-
tion and oxidation of the adenine core were observed. During an
experiment to determine efficacy of 3 in a mouse model of lung
infection with S. pneumoniae, in vivo exposure was dramatically in-
creased when the animals were pre-treated with aminobenzotriaz-
ole (ABT), a mechanism-based inactivator of Cytochrome P-450
(CYP) (Fig. 3).5 The difference in reduction of Colony Forming Units
(CFU) in the lungs of animals with and without the ABT pre-treat-
ment is presumably due to the difference in clearance between the
two groups. This is another indication that clearance was due to
CYP-mediated metabolism. Subsequently, a chemistry effort was
undertaken to reduce CYP-mediated oxidative metabolism of the
cycloalkyl side chain.
1.E+09
1.E+08
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
Input
Vehicle, qid, q3h + ABT
135 mg/kg tid, q3h + ABT
90 mg/kg, bid, q3h + ABT
180 mg/kg qid, q3h + ABT 180 mg/kg, qid, q3h
Figure 3. In vivo activity of 3 on S. pneumoniae in the lungs of immunocompetent
mice (for experimental details, see Ref. 2c). Results are plotted as means; error bars
represent the standard errors of the means. Dosing regimens are given in mg/kg per
day (bid = 2x per day, tid = 3x per day, qid = 4x per day; q3 h = 4x per day every 3 h).
ABT administered as a 100 mg/kg oral dose 2 h prior to treatment.
General strategies to reduce CYP-mediated metabolism include
removing or sterically blocking the vulnerable site, reducing
hydrogen-abstraction potential of the site, or reducing affinity of
the compound for the CYP active site by decreasing lipophilicity.6
All of these approaches were pursued in this series. Table 2 shows
examples of 2-alkoxyadenosine analogs designed to reduce clear-
ance due to oxidative metabolism of the 2-position cycloalkyl sub-
stituent. Replacement of a carbon from the cyclopentyl groups of 1
and 2 by an oxygen resulted in 7 which was less lipophilic (LogD
ꢀ0.06 for 7 compared to LogD 1.5 for 1).7 For 9, a methyl substitu-
ent was incorporated on the cyclopropylmethoxy side chain of 8 to
sterically block access to the CYP active site. However, in vivo total
clearance was not improved for 7 and 9 compared to 2 and 8,
respectively. In some cases, total clearance was reduced by fluori-
nation of the cycloalkyl ring. Incorporation of 4-trifluoromethyl
cyclohexanol at the 2-position resulted in 10 [relative stereochem-
istry is trans, cis isomers not investigated] which exhibited a three-
fold reduction in clearance compared to the non-fluorinated
methylcyclohexyl-containing 4. Difluorination of the cyclohexyl
ring resulted in 11 which also exhibited a lower clearance. How-
ever, there were also analogs for which fluorination of the side
chain did not result in a reduction of total clearance compared to
the non-halogenated versions. Examples include 12, containing a
trifluoromethyl substituent designed to block oxidation of the
methylene carbon of cyclopropane methanol, and the tetrafluoro-
cyclobutane methanol-containing 13. Total clearance was equiva-
lent for those compounds compared to the non-halogenated
analogs 8 and 3. To reduce metabolic liability from the methylene
carbon, it was removed in 14 and 15. For these compounds,
fluorination of the cycloalkyl group resulted in reduced clearance.
However, many of these modifications also resulted in decreased
antibacterial activity (7, 10, 11, 13–15 compared to the non-
halogenated parents 1–4, 8).
The major contributors to clearance of the adenosines were
hepatic metabolism and renal clearance. To study the relationship
between clearance and lipophilicity, renal clearance was measured
and compared to total clearance. Initial studies suggested that
renal clearance accounted for a greater proportion of the total
clearance for compounds with LogD <1.5, such as 7 compared to
2. For example, for 14, renal clearance accounted for 70% of the
total clearance, and LogD = 1.3. However, SAR for antibacterial
activity indicated that compounds which balance antibacterial
activity and drug-like physical properties have LogD 1.5–2.5;2a
therefore, these more polar analogs were not pursued. In some of
the fluorinated compounds, it is possible that metabolic switching
occurred such that oxidation of the cycloalkyl ring no longer ac-
counted for the major metabolites. Another mechanism, such as
glucuronidation or oxidation of the adenine core may have become
more prevalent. Metabolites from oxidation of the core and glucu-
ronide conjugates were also more prevalent in the qualitative
metabolite identification studies in compounds with halogenated
minor metabolites detected:
oxidation of core
NH2
major metabolites detected:
oxidation of side chain(M +16, M +48)
N
N
N
N
2 O
O
OH
F
3'
5'
OH
minor metabolites detected:
hydroxy glucouronide conjugates
Figure 2. Qualitative metabolite identification study of 3. Samples were collected and pooled from the rat pharmacokinetic assay and analyzed by LC/MS/MS. Data shown are
representative of metabolite identification studies run with other 2-alkoxy-50-deoxyadenosines from this series.

S. S. Stokes et al. / Bioorg. Med. Chem. Lett. 22 (2012) 85–89
87
Table 2
Antibacterial activity, LogD, total clearance, and renal clearance of 7–15, designed to reduce CYP-mediated oxidation of the 2-position side chain
Compd Structure
MIC S. pneumoniae
g/ml)
LogD
total clearance Cl (ml/
min/kg)
Renal clearance Clr (ml/min/kg) (%
total)
(
l
(ACDlogD7.4a)
NH2
N
O
N
N
N
O
7
16
ꢀ0.06
(ꢀ0.2)
(0.3)
2.2
67
49
13 (19%)
13 (27%)
11 (10%)
2 (7%)
O
OH
OH
NH2
N
N
N
N
O
O
8
9
8
O
O
OH
F
F
OH
NH2
N
N
N
N
4
115
OH
OH
NH2
F
F
F
N
N
N
N
O
O
10
11
12
13
2
27
O
O
O
O
OH
OH
NH2
F
N
N
N
F
N
8
1.4
33
15 (45%)
OH
OH
NH2
F
F
N
N
F
N
N
O
8
1.8
73
5 (7%)
OH
OH
NH2
N
N
F
F
N
N
O
4
1.4
85
17 (20%)
F
F
OH
OH
NH2
F
N
N
F
N
N
O
14
16
1.3
23
16 (70%)
F
F
O
OH
OH
NH2
N
N
N
N
O
15
16
1.7
58
8 (14%)
F
O
F
OH
OH
Calculated LogD using commercially available software package ACD labs from Advanced Chemistry Development.
a

88
S. S. Stokes et al. / Bioorg. Med. Chem. Lett. 22 (2012) 85–89
2-position side chains [data not shown]. However, further quanti-
tative experiments would be necessary to determine the actual
mechanism of clearance in these compounds.
remaining alcohol to yield 18. Fluorination of the resulting ketone
gave a mixture of 19 and 20. In order to separate the desired 19
from the co-eluting side product 20 resulting from elimination of
HF, dihydroxylation of the olefin was performed. This resulted in
a diol which was significantly more polar than the difluoro inter-
mediate 19, so the two components could be easily separated by
flash chromatography. Deprotection resulted in the desired alcohol
21 which could then be used in the subsequent displacement reac-
tion of the previously described 2-chloroadenosine intermediate
17.
In conclusion, the adenosine class of LigA inhibitors derived
from an HTS screen was optimized for potency, physical properties,
and clearance. The adenosine class represents the first series re-
ported to show efficacy in an animal model of infection due to inhi-
bition of bacterial DNA ligase, thereby validating LigA as a useful
target for antibacterial drug discovery. The major issue in this ser-
ies was the high clearance due to CYP-mediated metabolism of the
2-alkoxy substituent or high renal clearance for polar analogs.
Although many strategies were pursued to address the metabo-
lism, fluorination of the cycloalkyl moiety was the most successful
way to decrease total clearance without a significant loss of anti-
bacterial activity; this was more successful for compounds without
a methylene between the ether and cycloalkyl ring. Through this
approach, 10 was designed and synthesized which exhibited good
in vivo stability in both rat and dog with high oral bioavailability.
To obtain adenosine analogs with potential clinical utility, these
Compound 10 ( Fig. 4) demonstrated sufficient antibacterial
activity and metabolic stability such that it was chosen for further
evaluation in rat and dog. Pharmacokinetic analyses were per-
formed determining dog clearance to be lower than rat clearance
(20% vs 39% hepatic blood flow), and oral bioavailability of 10
was high: 88% and 86% in dog and rat, respectively. For the aden-
osine series, in general, trends for mouse, rat, and dog clearance
rates were the same. Testing of 10 in an animal model of infection,
without the addition of ABT, should lead to a reduction in colony
forming units thus demonstrating its efficacy. However, other
directions for this project were pursued before this experiment
could be performed.
Synthesis of the antibacterial adenosine LigA inhibitors is de-
scribed in Schemes 1 and 2. Compounds 1–15 were synthesized
as described previously from commercially available 2-chloroade-
nosine (Scheme 1).2,8 Protection, tosylation, nucleophilic displace-
ment with either hydride or fluoride, displacement using an
alkoxide, and final deprotection yielded the desired compounds
1–15. Most alcohols that were used as precursors to the 2-alkoxy
substituents were either available commercially or made by reduc-
tion of the corresponding ester or carboxylic acid.
Synthesis of 4,4-difluorocyclohexanol (Scheme 2) was made by
mono-protection of 1,4-cyclohexanediol and oxidation of the
NH2
MIC (S. pneumoniae) = 2 µg/ml
Log D = 2.32
CF3
N
N
N
solubility = 800 µM
human plasma protein binding = 20% free
rat i.v. Cl = 27 ml/min/kg (39% HBF)
oral bioavailability = 86%
N
O
O
OH
dog i.v. Cl = 7.7 ml/min/kg (20% HBF)
oral bioavailability = 88%
OH
Figure 4. Profile of compound 10 including antibacterial activity, physical properties, and pharmacokinetic properties.
NH2
NH2
N
N
N
N
N
N
a, b
c or d
N
N
Cl
Cl
O
O
OH
O
TsO
HO
OH
O
16
NH2
NH2
N
N
N
N
e, f
N
N
N
N
Cl
OR
O
O
O
OH
X
X
X = H or F
O
OH
17
1 - 15
Scheme 1. Synthesis of 2-alkoxy-50-deoxyadenosines 1–15. Reagents and conditions: (a) 2,2-Dimethoxypropane, pTsOH, acetone, 45 °C (85%); (b) TsCl, pyridine, ꢀ10 °C
(85%); (c) LiEt3BH, THF (74%); (d) TBAF, CH2Cl2 (55%); (e) ROH, NaOH, THF, 75 °C (10–85%); (f) AcOH or HCO2H, H2O, 85 °C (80–90%).

S. S. Stokes et al. / Bioorg. Med. Chem. Lett. 22 (2012) 85–89
89
Metabolism
& Pharmacokinetics Department at AstraZeneca,
Boston for physical property measurements, pharmacokinetic
parameter determinations, and metabolite identification studies.
c
O
a, b
HO
O
OH
O
O
References and note
18
1. (a) Gilbert, D. N.; Guidos, R. J.; Boucher, H. W.; Talbot, G. H.; Spellberg, B.;
Edwards, J. E., Jr.; Scheld, W. M.; Bradley, J. S.; Bartlett, J. G. Clin. Infect. Dis. 2010,
50, 1081; (b) Projan, S. J.; Bradford, P. A. Curr. Opin. Microbiol. 2007, 10, 441; (c)
Spellberg, B.; Guidos, R.; Gilbert, D.; Bradley, J.; Boucher, H.; Scheld, W. M.;
Bartlett, J. G.; Edwards, J., Jr. Clin. Infect. Dis. 2008, 46, 155; (d) Talbot, G. H.;
Bradley, J.; Edwards, J. E., Jr.; Gilbert, D.; Scheld, M.; Bartlett, J. G. Clin. Infect. Dis.
2006, 42, 657.
2. (a) Stokes, S. S.; Huynh, H.; Gowravaram, M.; Albert, R.; Cavero-Tomas, M.; Chen,
B.; Harang, J.; Loch, J. T.; Lu, M.; Mullen, G. B.; Zhao, S.; Liu, C.-F.; Mills, S. D.
Bioorg. Med. Chem. Lett. 2011, 21, 4556; (b) Stokes, S. S.; Huynh, H.; Gowravaram,
M.; Lu, M.; Mullen, G.; Mills, S.D. Presented at the 49th Interscience Conference
on Antimicrobial Agents and Chemotherapy, San Francisco, CA, Sep 2009,
abstract F1-1980.; (c) Mills, S. D.; Eakin, A. E.; Buurman, E. T.; Newman, J. V.;
Gao, N.; Huyhn, H.; Johnson, K. D.; Karantzeni, I.; Lahiri, S.; Shapiro, A.; Walkup,
G. K.; Yang, W.; Stokes, S. S. Antimicrob. Agents Chemother. 2011, 55, 1088; (d)
Gowravaram, M.; Stokes, S.S.; Huynh, H.; Lu, M.; Albert, R.; Chen, B.; Mullen, G.;
Cavero-Tomas, M.; Loch, J.; Zhao, S.; O’Shea, T.; Newman, J. V.; Mills, S. D.
Presented at the 242nd ACS National Meeting & Exposition, Denver, CO, Aug
2011, abstract MEDI-93.
+
d, e, f
O
F
O
O
F
F
19
20
NH2
N
F
N
N
F
g, h
F
F
HO
N
O
O
OH
OH
21
11
Scheme 2. Synthesis of 4,4-difluorocyclohexanol analog 11. Reagents and condi-
tions: (a) Benzoyl chloride, NEt3, CH2Cl2 (99%); (b) PCC, mol sieves, CH2Cl2 (83%); (c)
DAST, CH2Cl2, 0 °C [not isolated]; (d) OsO4 (aq), acetone/H2O; (e) SiO2 column
chromatography (78% over 2 steps); (f) KOH (aq), THF (99%); (g) NaOH, 17, THF,
75 °C; (h) HCO2H, H2O, 85 °C.
3. (a) Lehman, I. R. Science 1974, 186, 790; (b) Shuman, S. J. Biol. Chem. 2009, 284,
17365; (c) Tomkinson, A. E.; Vijayakumar, S.; Pascal, J. M.; Ellenberger, T. Chem.
Rev. 2006, 106, 687.
4. (a) Dwivedi, N.; Dube, D.; Pandey, J.; Singh, B.; Kukshal, V.; Ramachandran, R.;
Tripathi, R. P. Med. Res. Rev. 2008, 28, 545; (b) Swift, R. V.; Amaro, R. E. Expert
Opin. Drug Discov. 2009, 4, 1281; (c) Meier, T. I.; Yan, D.; Peery, R. B.; McAllister,
K. A.; Zook, C.; Peng, S.-B.; Zhao, G. J. FEBS 2008, 1; (d) Brötz-Oesterhelt, H.;
Knezevic, I.; Bartel, S.; Lampe, T.; Warnecke-Eberz, U.; Ziegelbauer, K.; Habich,
D.; Libischiniski, H. J. Biol. Chem. 2003, 278, 39435; (e) Ciarrocchi, G.; MacPhee,
D. G.; Deady, L. W.; Tilley, L. Antimicrob. Agents Chemother. 1999, 43, 2766.
5. Ortiz deMontellano, P. R.; Mathews, J. M. Biochem. J. 1981, 195, 761.
6. Van de Waterbeemd, H.; Smith, D. A.; Beaumont, K.; Walker, D. K. J. Med. Chem.
2001, 44, 1.
compounds require further improvements in potency while main-
taining in vivo clearance which results in sufficient exposure to
treat infections.
Acknowledgments
7. The 50fluoro substituent does not alter the physical properties; 1 and 2 have
nearly identical clogD values. Therefore comparison with either compound is
valid.
The authors thank the following groups: the Biosciences Depart-
ment at AstraZeneca, Boston for determination of MIC values and
efficacy data from the animal models of infection, and the Drug
8. For experimental details and 1HNMR data, see Cavero-Tomas, M.; Gowravaram,
M.; Huynh, H., Ni, H.; Stokes, S. Patent WO 2006040558.