Discovery and optimisation of potent, selective, ethanolamine inhibitors of bacterial phenylalanyl tRNA synthetase

Article abstract of DOI:10.1016/j.bmcl.2005.03.003

High throughput screening of Staphylococcus aureus phenylalanyl tRNA synthetase (FRS) identified ethanolamine 1 as a sub-micromolar hit. Optimisation studies led to the enantiospecific lead 64, a single-figure nanomolar inhibitor. The inhibitor series shows selectivity with respect to the mammalian enzyme and the potential for broad spectrum bacterial FRS inhibition.

Full text of DOI:10.1016/j.bmcl.2005.03.003

Bioorganic & Medicinal Chemistry Letters 15 (2005) 2305–2309
Discovery and optimisation of potent, selective, ethanolamine
inhibitors of bacterial phenylalanyl tRNA synthetase
Richard L. Jarvest,a,* Symon G. Erskine,b Andrew K. Forrest,a
Andrew P. Fosberry,a Martin J. Hibbs,a Joanna J. Jones,a Peter J. OÕHanlon,a
Robert J. Shepparda and Angela Worbya
aGlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK
bGlaxoSmithKline, 1250 South Collegeville Road, Collegeville, PA 19426, USA
Received 13 December 2004; revised 25 February 2005; accepted 2 March 2005
Available online 30 March 2005
Abstract—High throughput screening of Staphylococcus aureus phenylalanyl tRNA synthetase (FRS) identified ethanolamine 1 as a
sub-micromolar hit. Optimisation studies led to the enantiospecific lead 64, a single-figure nanomolar inhibitor. The inhibitor series
shows selectivity with respect to the mammalian enzyme and the potential for broad spectrum bacterial FRS inhibition.
ꢀ 2005 Elsevier Ltd. All rights reserved.
The continued spread of bacterial antibiotic resistance
and its attendant threat to antimicrobial and general
medicine, has been well-documented in medical and sci-
entific publications. The rise of methicillin resistant S.
aureus (MRSA) has received particular attention but
antibiotic resistance in the major pathogens responsible
for respiratory tract infections is also of concern. In the
search for new molecular targets for novel antibacterial
agents that would not be compromised by established
resistance mechanisms, we have focused on bacterial
aminoacyl tRNA synthetases. These enzymes couple
proteinogenic amino acids to their cognate tRNA in
an essential step in protein biosynthesis. The enzyme
family is validated clinically by the topical antibiotic
mupirocin (marketed as Bactrobanꢁ) which acts by
inhibition of isoleucyl tRNA synthetase.
ery of new inhibitors of bacterial FRS by high through-
put screening.4,5
In a pioneering series of studies, Santi et al. character-
ised FRS from Escherichia coli6 and went on to iden-
tify phenethylamine derived inhibitors with potency
down to 140 nM against the E. coli enzyme.7 Structurally,
FRS is a large complex heterodimeric a2b2 enzyme8
whose subunits are encoded by a linked transcriptional
operon in S. aureus.9 Because of the complexity of the
enzyme, the high throughput screening of the Smith-
Kline Beecham compound bank was carried out with
a crude preparation of FRS isolated from wild-type
S. aureus Oxford. Subsequently, the enzyme from S.
aureus WCUH29 was cloned, overexpressed in E. coli,
purified and characterised.9 This recombinant enzyme
was used for testing compounds in IC50 assays using
the full aminoacylation reaction with an SPA
readout.10,11
We targeted high throughput screening of the aminoacyl
tRNA synthetases from S. aureus. For example, we have
described the identification of a file compound hit
against S. aureus methionyl tRNA synthetase and opti-
misation of the series to give highly potent inhibitors
with excellent antibacterial activity.1–3 Here we describe
the discovery and optimisation of a hit against S. aureus
FRS. Others have also very recently reported the discov-
The ethanolamine 1 was identified as potent hit in the
high throughput screen. When tested in the standard
aminoacylation assay, 1 was found to inhibit S. aureus
FRS with an IC50 value of 160 nM. Further kinetic char-
acterisation of the enzyme inhibition by 1 showed that
its mechanism was best described as competitive with re-
spect to phenylalanine but non-competitive with respect
to ATP. Compound 1 thus formed an excellent starting
point for a hit-to-lead optimisation programme.
*
Corresponding author. Tel.: +44 01438 762074; fax: +44 01438
763620; e-mail: richard.l.jarvest@gsk.com
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.03.003

2306
R. L. Jarvest et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2305–2309
The role of the meta-trifluoromethyl group in the left
OH
H
N
hand side ring was investigated with a series of ana-
logues containing an ortho alkoxy substituent in the
right hand side ring, as shown in Table 2. It is clear from
the positional isomers and di-substitution, that only
meta substitution is favoured and that further substitu-
tion is highly detrimental.
1
CF3
The most efficient preparation of arrays of analogues of
1 was by reductive alkylation of ethanolamines with aryl
aldehydes, as shown in step (iii) of Scheme 1. The limit-
ing factor for this chemistry was the availability of the
ethanolamine starting materials 4. An array methodol-
ogy for the synthesis of these compounds was developed
as shown in steps (i) and (ii) of Scheme 1. In a one-pot
process, in situ formation of the silylated cyanohydrins
3 was followed by LAHreduction and parallel normal
phase chromatography to afford the ethanolamines 4
in moderate yields and good purity.
Further analogues to identify the preferred meta substi-
tuent were then prepared (Table 3). Compact non-polar
substituents, such as halogen, methyl and trifluoro-
methyl were preferred at this position. Larger (21) or
polar (25) substituents resulted in a significant reduction
in potency.
Amongst the compounds synthesised in the meta-substi-
tuted array was the bromothiophene analogue 34. This
had an FRS IC50 value of 50 nM, a 3-fold improvement
over the corresponding (trifluoromethyl)phenyl ana-
logue. The consistent improvement in properties of the
bromothiophene resulted in this becoming the preferred
left hand side moiety.
Me3SiO
Ar
O
i.
Ar
H
Thus to enhance potency further, the SAR of the 2-alk-
oxy position was developed in the bromothiophene ser-
ies. Several groups of analogues were prepared via array
chemistry at the benzaldehyde level. Direct alkylations
of salicylaldehyde (26) resulted in final compounds such
N
ii.
2
3
OH
4
OH
iii.
H
N
NH2
Ar
Ar
R
5
Table 2. FRS inhibition of (trifluoromethyl)phenyl ethanolamines
Scheme 1. Reagents and conditions: (i) TMSCN (1.1 equiv)/ZnI2
(0.1 equiv)/Et2O; (ii) LiAlH4/Et2O/D; (iii) R-PhCHO/NaCNBH3 or
resin CNBH3/AcOH/MeOH.
OH
H
N
X
O
R
Initial studies of the SAR of the right hand side phenyl
ring showed that substitution was only tolerated at the
ortho position. The enzyme assay results for ortho
substituted analogues are reported in Table 1. The car-
bon-linked and nitrogen-linked substituents of com-
pounds 6–8 resulted in a significant reduction in
inhibition. However, several oxygen-linker substituents,
such as those of compounds 12 and 13, afforded a signif-
icant enhancement in potency.
No.
X
R
IC50 (nM) S. aureus FRS
14
11
15
16
17
2-CF3
3-CF3
H>10,000
H150
4-CF3
2,5-Di-CF3
3,5-Di-CF3
H>10,000
H>10,000
Ph
>10,000
Table 3. FRS inhibition of 3-(substituted)phenyl ethanolamines
OH
H
Table 1. FRS inhibition of 3-(trifluoromethyl)phenyl ethanolamines
N
OH
H
O
R
N
X
R
No.
X
R
IC50 (nM) S. aureus FRS
CF3
11
18
19
20
21
CF3
Br
H150
H110
No.
R
IC50 (nM) S. aureus FRS
Me
OCF3
OCH2Ph
H110
H190
H>10,000
6
Me
790
7
CO2H2500
Morpholino
OH100
8
ꢀ10,000
13
22
23
24
25
CF3
Cl
Ph
Ph
Ph
Ph
570
50
31
32
83
9
10
11
12
13
OCF2H260
OMe
Oallyl
I
OMe
CH2OHPh
160
58
OCH2Ph
50

R. L. Jarvest et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2305–2309
2307
H
i.
H
O
O
O
O
OH
26
OEt
27
O
iii.
ii.
H
H
iv.
O
O
O
R
O
O
( )n
N
( )n
OH
R'
31 n = 1
32 n = 2
33 n = 3
28 n = 1
29 n = 2
30 n = 3
Scheme 2. Reagents and conditions: (i) Br(CH2)3CO2Et/K2CO3/DMF/D (96%); (ii) propiolactone/NaH/DMF/D (35%); (iii) KOH/DMF (94%); (iv)
RR0NH/EDC/HOAt/THF/DMF (50–70%).
as 44–48. More remote diversity was introduced by
amide arrays as shown in Scheme 2. The acid intermedi-
ates were prepared by bromoester alkylation and base
hydrolysis (30), or by propiolactone alkylation (29),
whilst 28 was commercially available. The acids were
converted to amides by standard water-soluble carbodi-
imide coupling methodology.
highly enantiospecific with a ratio of over 250-fold be-
tween the two isomers.
OH
61 racemic
NH2
S
62 (R)-enantiomer
63 (S)-enantiomer
Br
In contrast to the relatively tight SAR constraints else-
where in the molecule, a wide variety of ortho-alkoxy
substituents was tolerated as evidenced by many ana-
logues with IC50 values <50 nM in Table 4. A diverse
range of functional groups is accommodated, including
hydrogen bond donors and acceptors and heterocycles,
as well as ionisable groups such as amine and carboxylic
acid. Reduced activity was seen only for the phenoxy
compound 35 and several of the analogues with a nitro-
gen atom separated by two carbons from the ortho oxy-
gen (39, 40, 51, 52). The diverse functionality possible at
this position allows tuning of the overall physicochemi-
cal properties of the inhibitors. The tolerance for such a
range of polar functional groups suggests that in the en-
zyme–inhibitor complex, the alkoxy group is likely to be
at least partially solvent-exposed on the protein surface.
In view of the high affinity of the compounds for S. aur-
eus FRS, we wished to test the spectrum of inhibition
against FRS enzymes from organisms that are impor-
tant pathogens in respiratory tract infections. The FRS
enzymes from Streptococcus pneumoniae and Haemoph-
ilus influenzae, the key Gram-positive and Gram-nega-
tive RTI pathogens, were overexpressed and purified
essentially as described for the S. aureus enzyme.9 Se-
lected analogues were tested in assays against these en-
zymes and the results are shown in Table 5.13 The
initial hit 1 did not inhibit the H. influenzae enzyme at
the highest concentration tested. In contrast, the opti-
mised compounds 48 and 64 showed a relatively bal-
anced spectrum of inhibition across the three species.
All three enzymes had a strong preference for the (R)
stereochemistry. These results are encouraging for the
potential of an RTI spectrum of FRS inhibition.
Representative compounds from the series were tested
for selectivity against mammalian FRS, isolated from
rat liver. Very little inhibition of the mammalian enzyme
was observed. For example, compound 47 had an IC50
of 90,000 nM, resulting in a selectivity ratio of roughly
5000-fold in favour of the bacterial enzyme.
Despite potent inhibition of S. aureus FRS, none of
these compounds showed whole cell antibacterial activ-
ity against the S. aureus organism, when tested at con-
centrations up to 64 lg/mL. Although it is possible
that the target potency needs improving still further,
the extremely poor antibacterial activity seems likely
to be due to issues of bacterial penetration or efflux.
In artificial membrane assays, representative analogues
from the series showed high permeability (data not
shown). Thus bacterial efflux may be the major factor
contributing to the large discrepancy between enzyme
potency and antibacterial activity for the ethanolamine
inhibitors.
The racemic amine 61 was resolved by chiral HPLC on a
Chiralpak AD column, affording the enantiomers 62
(faster eluting) and 63 (slower eluting) with 99.8% ee
and 98.0% ee, respectively. The absolute configuration
of the enantiomers was assigned by ab initio optical
rotation.12 The enantiomeric amines were taken through
in the usual way to the (R)- and (S)-enantiomers of 38,
compounds 64 and 65, respectively. When tested against
S. aureus FRS, 64 had an IC50 value of 3.7 nM whereas
for 65 the IC50 was >1000 nM. Thus the inhibition is

2308
R. L. Jarvest et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2305–2309
Table 4. FRS inhibition of 4-bromo-2-thienyl ethanolamines
39/40 and 64/65, respectively, Greg Jonas for the chiral
preparative chromatography to resolve 62/63; Doug
Minick for assignment of absolute configuration; Ste-
phen Rittenhouse and his team for antibacterial testing;
and Nicola Wallis and Christine Richardson for useful
discussions.
OH
H
N
S
O
R
Br
No.
R
IC50 (nM) S. aureus FRS
34
35
36
37
38
39
40
41
42
Me
Ph
50
120
35
References and notes
CH2OMe
CH2CN
1. Jarvest, R. L.; Berge, J. M.; Berry, V.; Boyd, H. F.;
Brown, M. J.; Elder, J. S.; Forrest, A. K.; Fosberry, A. P.;
Gentry, D. R.; Hibbs, M. J.; Jaworski, D. D.; OÕHanlon,
P. J.; Pope, A. J.; Rittenhouse, S.; Sheppard, R. J.; Slater-
Radosti, C.; Worby, A. J. Med. Chem. 2002, 45, 1959.
2. Jarvest, R. L.; Berge, J. M.; Brown, M. J.; Brown, P.;
Elder, J. S.; Forrest, A. K.; Houge-Frydrych, C. S. V.;
OÕHanlon, P. J.; McNair, D. J.; Rittenhouse, S.; Shepp-
ard, R. J. Bioorg. Med. Chem. Lett. 2003, 13, 665.
3. Jarvest, R. L.; Berge, J. M.; Brown, P.; Houge-Frydrych,
C. S. V.; OÕHanlon, P. J.; McNair, D. J.; Pope, A. J.;
Rittenhouse, S. Bioorg. Med. Chem. Lett. 2003, 13,
1265.
49
(CH2)2OH8
(CH2)2NH2
220
250
18
(CH2)2phthalimido
(CH2)2morpholino
Allyl
18
43
44
45
46
47
48
CH2Ph
10
26
43
18
16
26
CH2(4-MeSO2Ph)
CH2(2-pyridyl)
CH2(3-pyridyl)
CH2(4-pyridyl)
CH2(2-benzimidazolyl)
4. (a) Yu, X. Y.; Finn, J.; Hill, J. M.; Wang, Z. G.; Keith, D.;
Silverman, J.; Oliver, N. Bioorg. Med. Chem. Lett. 2004,
14, 1339; (b) Yu, X. Y.; Finn, J.; Hill, J. M.; Wang, Z. G.;
Keith, D.; Silverman, J.; Oliver, N. Bioorg. Med. Chem.
Lett. 2004, 14, 1343.
49
50
51
52
CH2CONH2
CH2CONHEt
42
32
CH2CONHCH2CONHMe
CH2CONMe(CH2)2OH110
120
5. Beyer, D.; Kroll, H.-P.; Endermann, R.; Schiffer, G.;
Siegel, S.; Bauser, M.; Pohlmann, J.; Brands, M.; Zieg-
elbauer, K.; Haebich, D.; Eymann, C.; Bro¨tz-Oesterhelt,
H. Antimicrob. Agents Chemother. 2004, 48, 525.
6. (a) Santi, D. V.; Danenberg, P. V.; Satterly, P. Biochem-
istry 1971, 10, 4804; (b) Santi, D. V.; Danenberg, P. V.
Biochemistry 1971, 10, 4813; (c) Santi, D. V.; Danenberg,
P. V.; Montgomery, K. A. Biochemistry 1971, 10,
4821.
7. (a) Anderson, R. T., Jr.; Santi, D. V. J. Med. Chem. 1976,
19, 1270; (b) Santi, D. V.; Cunnion, S. O.; Anderson, R.
T., Jr.; Webster, R. W., Jr. J. Med. Chem. 1979, 22, 1260.
8. Mosyak, L.; Reshetnikova, L.; Goldgur, Y.; Delarue, M.;
Safro, M. G. Nat. Struct. Biol. 1995, 2, 537.
53
54
55
(CH2)2CO2H58
(CH2)2CONHCH2CONHMe
(CH2)2CONMe(CH2)2OH25
36
30
56
57
58
(CH2)3CO2H43
(CH2)3CONHCH2CONHMe
(CH2)3CONMe(CH2)2OH26
59
60
(CH2)4NH2
(CH2)4phthalimido
22
69
Table 5. Inhibition of various bacterial FRS enzymes by selected
analogues
9. Savopoulos, J. W.; Hibbs, M.; Jones, E. J.; Mensah, L.;
Richardson, C.; Fosberry, A.; Downes, R.; Fox, S. G.;
Brown, J. R.; Jenkins, O. Protein Expr. Purif. 2001, 21,
470.
Stereochem.
FRS IC50 (nM)
Streptococcus Streptococcus Haemophilus
aureus
pneumoniae
influenzae
´
10. Macarron, R.; Mensah, L.; Cid, C.; Carranza, C.; Benson,
1
RS
160
10
ND
280
250
NI @ 300
1500
150
N.; Pope, A.; Diez, E. Anal. Biochem. 2000, 284, 183.
11. Assay conditions used 1 · Km[Phe], 10 · Km[ATP] and a
large excess of tRNA.
43 RS
48 RS
26
3.7
>1000
64
65
R
S
190
>1000
56
>1000
12. Specific rotations were calculated for models of 2-amino-
1-(4-chlorothien-2-yl)ethanol at the sodium D line using
HF/SCF wavefunctions with the 6-31G* basis set (Cl was
used in place of Br as fourth row elements are not
supported in the current version of the Dalton program).14
Predicted specific rotations were averaged using Boltz-
mann statistics. Experimental values were measured in
methanol (c = 0.004 mg/mL, 20 ꢂC) at the sodium D line
and are compared to the predicted values:
ND—not done; NI—no inhibition.
In summary, a sub-micromolar HTS hit against S. aur-
eus FRS was identified and optimised to give a highly
potent sub-10 nM enantiospecific inhibitor. Whilst the
series was selective with respect to mammalian FRS
and showed significant broad spectrum inhibition, use-
ful antibacterial activity against S. aureus was not
attained.
[a]D (calcd)
[a]D (exp)
[a]D
(calcd ꢁ exp)
Acknowledgments
R
S
R
S
62
63
+9
+9
ꢁ9
ꢁ9
+13
ꢁ14
ꢁ4
ꢁ22
We thank Margarita Rodriguez-Lens and David
McNair for assistance with preparation of compounds
+22
+5

R. L. Jarvest et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2305–2309
2309
Both the sign and magnitude of rotation are in good
agreement for 62 as (R) and 63 as (S). The sign of the rotation
predicted in this way is correctly predicted in about 95% of
cases for small molecules, although the exact magnitude may
be more variable.15,16 It is considered that the current
configurational assignment is highly reliable. Moreover, this
assignment is consistent with the sign of rotation of all
literature aryl and heteroaryl ethanolamine enantiomers.
13. For each bacterial enzyme assay, the concentration of
phenylalanine was equal to Km for that enzyme, to allow
comparison across the species.
14. ÔDaltonÕ,
a molecular electronic structure program,
Release 1.2, 2001, written by T. Helgaker et al.
15. Polavarapu, P. L. Mol. Phys. 1997, 91, 551.
16. Stephens, P. J.; Devlin, F. J.; Cheeseman, J. R.; Frisch, M.
J. J. Phys. Chem. A 2001, 105, 5356.